GekkoScience has a new pod miner, just in time for Christmas

[sidehack]

Introducing the newest quiet consumer miner from GekkoScience, the newest addition to the Terminus pod-miner line, the R909.

It may not look like much right away, but it's a significant improvement over the R606 in almost every way.

This miner is built with the same physical dimensions as the R606 so anything built around them would still fit this unit. Airflow ducting is the same, same 80mm fan. However, you'll notice ports and jacks are located at the exhaust end instead of accessible from the side.

This makes the unit more readily stackable, as you no longer have to keep three sides clear - two for airflow and one for cabling.

It also allowed us to expand the circuit board by 15%, making room for a better main regulator and improved thermal management. Heatsink area over the chips is increased by 40% over the R606, and we've added more board mounting legs for improved stability especially during shipping.

The miner runs on the S17's BM1397 ASIC, as used in our Compac-F USB stick miner. Kano has been hard at work improving the driver, and has added several features that will be very attractive to tuning enthusiasts, though the stock settings will already be fairly excellent.

The miner is guaranteed to run at 1.5TH from under 100 DC watts at stock voltage and do it nearly silently. Better performance isn't guaranteed, but is probably attainable. Take for example the test unit I had running 1.8TH at 70 watts, or the above-photographed machine which pulled down almost 2.1TH from 88 watts (on stock voltage, no voltage tuning) for over a week:

The Terminus R909 needs an external controller running cgminer, with USB connection to the miner (a 6-foot USB cable will be included with purchase). The driver has been tested extensively on linux PC and Raspberry Pi. Like previous Terminus models, the miner runs on 12V power from either a standard 2.1/5.5mm barrel jack (rated for 8A) or a PCIe 6-pin jack.

The "stock" speed will be set to 400MHz, expected to see 1.6TH, but higher frequencies are possible. Kano will explain the details, but he's been working on adding tuning features which include setting individual chip frequencies. Core voltage is adjustable in the range of 1.4V-1.6V per chip, same as on the Compac F, and though the R606 had a fancy push-button interface, we've rolled back to a basic knob you turn with a screwdriver on this one. That may be improved on future versions but I won't make promises just in case.

This miner also features Blinky Lights, but they're not as intrusive as previous models of pod or stick. The rate of blinking is proportional to hashrate. The base light color is a delightful pink-magenta.

The BM1397 chips have a significant temperature coefficient, which means that they pull noticeably more power as they get warmer. Nobody at GekkoScience will be offended if a tuner or enthusiast upgrades the fan, as more airflow could mean reducing power consumption by 5-10% just from maintaining lower temperatures. These units take a standard 80mmx25mm 3/4-wire 12V case fan.

And like all GekkoScience products, the Terminus R909 is Made In The USA.

Approximate dimensions: 3.5 x 4.2 x 6 inches.

SALES

So far, two resellers have confirmed to stock this machine, and should have stock in time to start delivering orders by Christmas. Production is already well underway on the first batch. This post will be updated as more sellers are added.

bitcoinmerch.com shipping from California
419mining.com shipping from Ohio
bitshopper.de in Europe/Germany (forum user MacEntyre)

The Terminus R909 will include a 6ft USB cable. It'll be up to resellers to provide any extras, power supplies or anything.

[1715675645]

1715675645

[1715675645]

1715675645

[1715675645]

1715675645

[1715675645]

1715675645

[1715675645]

1715675645

[kano]

BETA release of cgminer for the new R909

The 'beta' comment is that there's quite a few changes, so we'll see how it goes,

I've moved this, since it allows me to edit it more easily, and also gives me full HTML markup to make it easier to read.
Here: https://kano.is/gekko.php

I'll comment here when there are any major updates.

2022-12-11 R606 performance issues have been resolved.
2022-12-13 Windows binary for v4.12.1 is available now at https://kano.is/cgminer.php
2022-12-17 Git updated to add per chip tuning information. Windows binary updated. https://kano.is/gekko.php updated.
2022-12-21 Update to improve performance and tuning. Windows binary updated. https://kano.is/gekko.php updated.
2023-01-14 Update only necessary for the old R808 miner.

[pkspks]

Hoping for a reseller in APAC. International shipping cost to Australia is always a killer.

[cygan]

a very good news to finally present the r606 successor. i had been waiting for this release for a long time and will definitely order at least 1 exemplar.
@sidehack can you perhaps tell us the range in which the new r909 will be priced?

[kano]

The resellers decide their pricing.

[Powell]

Looks like I am going to have to grab a few!  Good work Sidehack and crew!

[willi9974]

Thanks for the new device, looks nice and will also Order one

[n0nce]

BM1397-based pod miner, amazing! The efficiency of these chips is so much better than what's in Futurebit's Apollo; depending on pricing, I can see the R909 making the Apollo 'Standard Unit' obsolete.
I'm definitely going to pick one up. (for mining and tinkering.. )

[Sledge0001]

Love home mining gear like this.
Well done!

Now just need Santa to deliver!

[Skot]

Looks amazing, nice work! Is this based on six BM1397 chips?

[cygan]

BM1397-based pod miner, amazing! The efficiency of these chips is so much better than what's in Futurebit's Apollo; depending on pricing, I can see the R909 making the Apollo 'Standard Unit' obsolete.
I'm definitely going to pick one up. (for mining and tinkering.. )

i see exactly the same as you
and the power consumption of the new r909 is in my opinion unbeatable at this performance then

[o_solo_miner]

BM1397-based pod miner, amazing! The efficiency of these chips is so much better than what's in Futurebit's Apollo; depending on pricing, I can see the R909 making the Apollo 'Standard Unit' obsolete.
I'm definitely going to pick one up. (for mining and tinkering.. )

That depends on what you are planning to do, mining on your own bitcoind or not.
IF this CGMiner version is not reduced in its original function, it will be a good replacement, if not, not worth to buy.

[kano]

BM1397-based pod miner, amazing! The efficiency of these chips is so much better than what's in Futurebit's Apollo; depending on pricing, I can see the R909 making the Apollo 'Standard Unit' obsolete.
I'm definitely going to pick one up. (for mining and tinkering.. )

That depends on what you are planning to do, mining on your own bitcoind or not.
IF this CGMiner version is not reduced in its original function, it will be a good replacement, if not, not worth to buy.

Please avoid spreading FUD for your own personal agenda, and at least word it truthfully.

To correct his FUD, what o_solo_miner is saying is that if I don't fix the bug in ck's code, that o_solo_miner insists I must fix for him, where it wont currently solo mine directly to a local bitcoind, then for some reason that is somehow relevant.
Sounds like a lousy attempt at coercion to me.

I do know how to fix it since I've already done it myself to ckpool ... but fixing it in cgminer hasn't been a priority since I'm also unsure about suggesting people do something that could easily lead to losing a block and then having them blame me for it.
While this may or may not seem likely to some, even o_solo_miner's attempt at coercion above makes me wary of such things.

I will also point out that someone put up a pull request to my git to fix this but that pull request can generate work that the miner will never get the reward if they find a block, in certain circumstances.

As I have made clear many times, I don't trust people's code that isn't tested properly.
Especially when it's involved with bitcoin blocks.

--

Edit: I will also point out that I have attempted to get core to simplify block testing for developers.
I've raised an issue quite a while ago and discussed it as can be seen:
https://github.com/bitcoin/bitcoin/issues/22119
I also explain there how I do full block testing at lower difficulty and the non-testing related issues with doing that.

[heslo]

This looks really nice! Might have to pick one up if there's a reseller for my region. As always, love your work sidehack!

[o_solo_miner]

BM1397-based pod miner, amazing! The efficiency of these chips is so much better than what's in Futurebit's Apollo; depending on pricing, I can see the R909 making the Apollo 'Standard Unit' obsolete.
I'm definitely going to pick one up. (for mining and tinkering.. )

That depends on what you are planning to do, mining on your own bitcoind or not.
IF this CGMiner version is not reduced in its original function, it will be a good replacement, if not, not worth to buy.

Please avoid spreading FUD for your own personal agenda, and at least word it truthfully.

First, there is no FUD, and there is no personal agenda.

To correct his FUD, what o_solo_miner is saying is that if I don't fix the bug in ck's code, that o_solo_miner insists I must fix for him, where it wont currently solo mine directly to a local bitcoind, then for some reason that is somehow relevant.
Sounds like a lousy attempt at coercion to me.

You better word it truthfully:
There is no bug in ck's code, there was a change from core after ck stop the support of cgminer and that stoped cgminers ability to mine
on the bitcoind.

I pointed that out here:
https://bitcointalk.org/index.php?topic=5253096.msg54839526#msg54839526

While this may or may not seem likely to some, even o_solo_miner's attempt at coercion above makes me wary of such things.

There is no coercion by me!
You pointed it out the last time, and I get it, you won't do it, for obvious reason.

For me (and not only for me) it is a vital point of having the freedom of choice to solo mine on a bitcoind or a pool.
If all trustfull pools/solopools went away, I still got the opertunity to solo mine directly to core.

End of story

[sidehack]

Sounds like a Core problem. Also sounds like a problem that's already been discussed elsewhere and doesn't really need to be here too.

[o_solo_miner]

Sounds like a Core problem. Also sounds like a problem that's already been discussed elsewhere and doesn't really need to be here too.

Unfortunaly not, it is a cgminer problem, and because there is no alternative to it for running this miner it belongs here.
Sorry, you did a great job on developing this miner and I am sure it will sell like "hot cakes" even without real solo mining.

[Skot]

Sounds like a Core problem. Also sounds like a problem that's already been discussed elsewhere and doesn't really need to be here too.

Unfortunaly not, it is a cgminer problem, and because there is no alternative to it for running this miner it belongs here.
Sorry, you did a great job on developing this miner and I am sure it will sell like "hot cakes" even without real solo mining.

What about just running a stratum proxy along with bitcoin core and connecting to that with cgminer and your sweet new R909?

[kano]

Sounds like a Core problem. Also sounds like a problem that's already been discussed elsewhere and doesn't really need to be here too.

Unfortunaly not, it is a cgminer problem, and because there is no alternative to it for running this miner it belongs here.
Sorry, you did a great job on developing this miner and I am sure it will sell like "hot cakes" even without real solo mining.

What about just running a stratum proxy along with bitcoin core and connecting to that with cgminer and your sweet new R909?

Yes there are various ways to do it, just a certain person wants to make a deal out of me not directly fixing the problem that's in cgminer

The flags were deprecated in bitcoin core 0.12 in 2015 (it's now empty)
https://github.com/bitcoin/bitcoin/commit/d449772cf69c01932fc5d72c46054815d6300b3c
And remove during 0.19 in 2019 (since it is empty)
https://github.com/bitcoin/bitcoin/commit/e9a27cf338dc618b8ecab8984abc54d588de8a05

He seems to want to make it an issue for sidehack for some reason - no idea why.

Again, as before, I'm doing a driver that's opensource and free for everyone to use - with a couple of miners sent to me so I can write it.

Yes as I also stated clearly, cgminer still cannot mine directly to bitcoind, it fails to work, which is called a bug.
(since those changes above in core)

[n0nce]

What about just running a stratum proxy along with bitcoin core and connecting to that with cgminer and your sweet new R909?

Yes there are various ways to do it, just a certain person wants to make a deal out of me not directly fixing the problem that's in cgminer

Wouldn't even call it a problem (even though it can classify as a bug); separation of concerns is actually usually a pretty good idea.
Have one process for Bitcoin Core: bitcoind
One process for running the miner: cgminer
One process for connecting the two: Local Stratum server

[Powell]

For me (and not only for me) it is a vital point of having the freedom of choice to solo mine on a bitcoind or a pool.
If all trustfull pools/solopools went away, I still got the opertunity to solo mine directly to core.

End of story

So to me it seems like you are the equivalent to a Dooms Day Prepper for the Bitcoin world?  I don't mean that to come off as an attack, but this will be my 9th year on this forum and there are still trustworthy pools/solo pools you can mine on.  What I do notice though is you follow any project Kano is involved with because YOU want something changed.  I am sure like you said there are others but it's YOU that follows into Sidehack's product releases talking about how it's not truly something.  Which as of right now and for the forseeable future it is as I do not see pools disappearing for awhile, and by the time they do this hardware will be the equiv to running an Antminer U1 on a modern day pool.

Another analogy when it comes to the end of the pools, would be again comparing a person watching for the end of the world and prepping for it.  It would be the same as calling every gun manufacturer and complaining that even though they only make the firearm that nobody will be able to make bullets for it if the world were to come to an end.  You are literally worried about something that isn't a problem and have been all over any Gekkoscience release because Kano is involved.  

So how about you set up your own stratum server and run instead of worrying about mining directly to Bitcoin core?  OR, how about you pay to have your own developer write a new miner specifically for people like you and the others you mention and then you don't have to follow people like Kano and Sidehack constantly trying to tell people that it's not a solo miner because in the future you can't just mine directly to core?  

[cygan]

if you want to run the new r909 miner together with other compac f sticks, what should the cmd line look like to start all the miners via cgminer?

[kano]

I've posted the details now back in the 2nd post
https://bitcointalk.org/index.php?topic=5423227.msg61350496#msg61350496

[Dlikrot]

When is it back in stock?

[sidehack]

It's a brand-new product and we busted our butts to get the assembly line up to speed in time for the first batches to be out for Christmas, which is still 19 days off. Reseller stock requests are being filled right now, so they'll be "in stock" with resellers in one to two weeks.

[Dlikrot]

Thank you! Very excited about this miner.

[cygan]

so i will run my 2 gekko usb hubs (with 3 compac f sticks each) together with the 2 new r909 pod miners using a rpi4. if everything runs well and is configured well i would get ~3.6th/s together

[BiticaMan]

It's a brand-new product and we busted our butts to get the assembly line up to speed in time for the first batches to be out for Christmas, which is still 19 days off. Reseller stock requests are being filled right now, so they'll be "in stock" with resellers in one to two weeks.

sidehack how many s17 chips does this miner pod have?

From what I understand I could use 3 or 4 on a PI connected directly to the USB port?
Could you please confirm.

Thanks.

[sidehack]

I've only tested 1 at a time on my Pi, but we're running five per USB bus (10 per machine) on the old Dell 755 testers. If a current model Pi has comparable horsepower to half an office PC from like 2010 you'd be fine. Given a Pi will run something like 12-14 Compac F, 3-4 R909 shouldn't be difficult.

By the way, 1.5TH is the baseline spec. The last batch off testers were running 1.6-1.7TH and still had headroom. Kano and I have both tested units at 2TH and up. Of course the top-end of your units will be dependent on the hardware behaving nicely but I bet you could get 3.6TH just from the R909s.

[NebulaMiner]

Hi, I am working on a BM13xx High Level Analyzer for Saleae Logic 2. You can find it on Saleae Logic 2 market place as an extension (and also in GH : https://github.com/GPTechinno/bm13xx-hla). Maybe you can have a use of it

Currently, basic communicatin (FIL/VIL mode, R/W register, manage chip address, job/nonce) is well handled. I am digging into the register functions and fields now.

I am sure your level of knowledge on the BM13xx serial protocol is higher han mine, but such a practical tool can be handy for your dev also.

Let me know what you think about it.

[yxt]

It's a brand-new product and we busted our butts to get the assembly line up to speed in time for the first batches to be out for Christmas, which is still 19 days off. Reseller stock requests are being filled right now, so they'll be "in stock" with resellers in one to two weeks.

can it purchased directly (where?) or only via reseller?

[sidehack]

Resellers only, same as we've done for the last several years. Mine is a small shop so we focus on R&D and manufacture and let resellers handle advertising, distribution and customer support.

[BiticaMan]

I've only tested 1 at a time on my Pi, but we're running five per USB bus (10 per machine) on the old Dell 755 testers. If a current model Pi has comparable horsepower to half an office PC from like 2010 you'd be fine. Given a Pi will run something like 12-14 Compac F, 3-4 R909 shouldn't be difficult.

By the way, 1.5TH is the baseline spec. The last batch off testers were running 1.6-1.7TH and still had headroom. Kano and I have both tested units at 2TH and up. Of course the top-end of your units will be dependent on the hardware behaving nicely but I bet you could get 3.6TH just from the R909s.

sidehack I have 350gh/s in my each compacF, if the R909 reaches 3Th/s that would give between 8 to 9 compacF, is this comparison correct?

I would like to know how many compacF or S17 processors have 1 R909 for me to calculate here and see what it pays for me to assemble my next miner, can you help me with this information?

[os2sam]

I would like to know how many compacF or S17 processors have 1 R909

My  guess would be 6

[kano]

Update:

as now mentioned in the 2nd post, the latest code has problems with the older miners.

Specifically the R606 wont scale up performance properly.
It performs badly and keeps dropping the frequency.

The CompacF, and R909 of course, work well with all the new features.

Switching back to the previous version will fix this for the old miners, for now, until I do regression testing on my old miners later.
The git commands to switch to the previous 4.12.0 version is:

Code:

cd
cd cgminer
git checkout f357f61c1d
CFLAGS="-O2 -march=native -fcommon" ./autogen.sh --enable-gekko --enable-icarus
make

The second last line is to make sure it say it's v4.12.0

[NotFuzzyWarm]

It should be pointed out that at least when using Raspian on a Pi3, it is a good idea to run multiple instances of cgminer when your sticks/pods are along side older ones that have vastly lower hash rate. In my case a Compac-F in one instance and 1 Newpac + 1 2Pac in a 2nd instance.

If need be just run the latest/greatest instance for F's & R909 and older one for others until Kano gets it worked out

[419mining]

419Mining opened the preorder sales, limited to 20 units. https://www.419mining.com/shop/miners/gekkoscience-r909/

[Dlikrot]

Ordered! No PlayStation 5 for me this christmas. Lottery mining for my family, here we go!

[Powell]

Got mine ordered as well!

[Dlikrot]

Should a Raspberry Pi 3 suffice for running it? Or even a 2?

[sidehack]

The Pi4 doesn't even blink, so a Pi3 should handle just fine.

[kano]

Anyone waiting for the code update for the R606 issues, it's been resolved.
See the 2nd post (or https://kano.is/gekko.php )

[jwinterm]

Just to make sure I am mathing correctly:

With 1.5 Th/s at current network difficulty you would get ~$0.15 in BTC per day? So if reseller charges $600 and BTC price doesn't change then you would recover purchase cost in 4000 days or about 11 years (not including electricity costs)?

I guess you mine for a year and then pray BTC price goes up 10x in USD terms?

Anyway, product looks awesome, I am tempted even with the above mathing. Nice work kano and sidehack

[Skot]

Just to make sure I am mathing correctly:

With 1.5 Th/s at current network difficulty you would get ~$0.15 in BTC per day? So if reseller charges $600 and BTC price doesn't change then you would recover purchase cost in 4000 days or about 11 years (not including electricity costs)?

I guess you mine for a year and then pray BTC price goes up 10x in USD terms?

Anyway, product looks awesome, I am tempted even with the above mathing. Nice work kano and sidehack

The time to recover the purchase cost of a Bitcoin node is infinity, but tons of people do that also because it's fun, educational and helps in a small way to secure the world's best money. Solo mine on your R909 and you'll even have a chance at 6.25+ BTC!

[sidehack]

Yeah nobody here is going to lie to you about the numbers. It's a toy, not a business case. Lot of people will solo mine with them because, for a bitcoin lottery ticket, you can't really beat a nearly silent 2TH miner using under $10 of power per month.

[Dlikrot]

I even hope it produces a bit heat. I will use it to heat a small isolated shed that has new products etc. that needs to be dry and not get moisture from outside.

I would have put a 80w oven there (though, with thermostat, but still) if it weren't for this purchase.

The only usecases would be for educational purpose or lottery mining (as a heater?). It COULD end up being genius lol.

[NotFuzzyWarm]

I even hope it produces a bit heat.

Folks... Ja it would make for a nice foot warmer but remember it is only drawing 60-maybe 100(?) watts and that means very little heat is produced

 The smallest space heater Ive ever found pulls ~150w on low and 300w on high and is sold as a <drum roll please> under the desk foot warmer...
Hmm, found a nice little radiant heat panel, 170W

[Dlikrot]

My father may have meant a oven pulls about 80w on a average (a small heater that is). Anyhow. Will come back with review after I get it ^^ Hope for some heat at least... haha

[kano]

As per the 2nd post,
There is a Windows binary for v4.12.1 available now at https://kano.is/cgminer.php
That should work fine with the older Gekko miners also (let me know if you have problems vs v4.12.0)
and will of course work with the new R909 when you get one.

[n0nce]

Does anyone (sidehack or European customers) know if Bitshopper already sold out or if it's just 'not available yet'?
The website says 'Out of stock', but it's not clear whether stock ran out or whether it has not been stocked yet.

[Dlikrot]

419mining delivers to europe too. still preorder open (last I checked).

[n0nce]

419mining delivers to europe too. still preorder open (last I checked).

Good call! Would be a good bit more expensive, though.
$599 = 570€, plus international shipping, VAT and import tax will end up near 700 bucks.
Meanwhile Bitshopper sells for 599€ incl. VAT and cheap shipping.

[o_solo_miner]

Bitshopper just listed the miner, I don't think you can order them before christmas.
On the CompacF there was a "early Bird" Discount, but not with this miner.

[cygan]

Bitshopper just listed the miner, I don't think you can order them before christmas.
✂

the r909 has been in stock at bitshopper.de since today and can now be ordered.
currently there are 5 miner left...
actually i would order 1-2 now but i am still waiting for an answer from the bitshopper owner to organize a kind of group buy

[o_solo_miner]

Bitshopper just listed the miner, I don't think you can order them before christmas.
✂

the r909 has been in stock at bitshopper.de since today and can now be ordered.
currently there are 5 miner left...
actually i would order 1-2 now but i am still waiting for an answer from the bitshopper owner to organize a kind of group buy

WOW that was quick!
now just 3 left

[cygan]

✂
WOW that was quick!
now just 3 left

ordered the last 2 r909 miners today
when i put them into working i can give a little update then

[kano]

Git updated to add per chip tuning information (see sig)
Windows binary updated.
https://kano.is/gekko.php updated.

[cygan]

Git updated to add per chip tuning information (see sig)
Windows binary updated.
https://kano.is/gekko.php updated.

do i understand correctly that only users who run the r909 through windows need to update cgminer?
my 4.12.0 version runs on linux, so i would not have to do anything - did i understand that correctly

[kano]

Git updated to add per chip tuning information (see sig)
Windows binary updated.
https://kano.is/gekko.php updated.

do i understand correctly that only users who run the r909 through windows need to update cgminer?
my 4.12.0 version runs on linux, so i would not have to do anything - did i understand that correctly

4.12.0 doesn't handle the new R909
My changes in git for it, I also updated the version to 4.12.1
All the new tuning is only in 4.12.1

[cygan]

Git updated to add per chip tuning information (see sig)
Windows binary updated.
https://kano.is/gekko.php updated.

do i understand correctly that only users who run the r909 through windows need to update cgminer?
my 4.12.0 version runs on linux, so i would not have to do anything - did i understand that correctly

4.12.0 doesn't handle the new R909
My changes in git for it, I also updated the version to 4.12.1
All the new tuning is only in 4.12.1

gotcha!
and what is the easiest command to update to the 4.12.1 version under linux?

[n0nce]

and what is the easiest command to update to the 4.12.1 version under linux?

Probably just download the latest release (or git pull all changes) and copy over your existing miner config file.
In case you installed it to your path, re-installing will simply overwrite the old binary / symlink wherever it sits on your path.

[willi9974]

Git updated to add per chip tuning information (see sig)
Windows binary updated.
https://kano.is/gekko.php updated.

do i understand correctly that only users who run the r909 through windows need to update cgminer?
my 4.12.0 version runs on linux, so i would not have to do anything - did i understand that correctly

4.12.0 doesn't handle the new R909
My changes in git for it, I also updated the version to 4.12.1
All the new tuning is only in 4.12.1

gotcha!
and what is the easiest command to update to the 4.12.1 version under linux?

Hi cygan,

Than you have your miners at home, let us know your config and the result that you will get.
Interested in the mining speed and configuration what you have.

Best regards,
Willi

[kano]

...
and what is the easiest command to update to the 4.12.1 version under linux?

https://kano.is/gekko.php#lin

If you typed "make install" last time you built it, then add that also at the end to overwrite the last one.

Edit: I've added a comment about that at the link

[n0nce]

Hey @sidehack; why did you choose to go with a USB-Mini Type-B cable?


I'm not even asking that everything becomes USB-C; but my issue with Type-B (Mini and Micro) is that they're (always?) often (like in this case) SMD-soldered. This means they can be ripped off the circuit board and cause almost irreparable damage.
Meanwhile both USB-A Type-B (shown below) and some USB Type-C sockets have through-hole pins that hold it firmly in place.

[sidehack]

And that's why we reinforce each one with epoxy.

[RKG456]

I just ordered one from Bitcoinmerch. 

I really need to get going on getting this setup. I now have 4 Compac-Fs, 2 NewPacs, a Terminus 808, and now this RS909. I have been so busy, I buy the items when I see them in stock but keep getting side tracked. I have never even powered them up. Hopefully none are duds.
I also have multiple USB hubs two from Bitcoinmerch and 2 Sipolar-USB Hub- 10 Ports USB 2.0 powered hubs as well as multiple RPI 4s in 4GB and 8GB as well as Radza Zeros, Libra Computer, a Neo PI, RockPi4b and multiple RPI Zeros and Zero2s.

Anyways enough rambling, on to my actual questions / concerns -
Are there issues with running multiple hosts? I ask as I am not sure if the USB 2.0 ports on an single RPI4 can handle all of the miners above across two powered USB2.0  hubs so I am thinking I may I run two RPI4s as hosts and run the Compac-Fs and the 909 on one and the older miners (two Newpacs and the 808) on the other.  Will that work with two hosts mining to the same pool from the same Public IP address, or do I need to introduce a some kind of proxy to regulate work for the two miners?  (I am not sure if a proxy is needed, but I read not to long ago about a proxy being an issue with mining Monero and thought I would ask about it. )  - Given the USB hubs have their own power supply does that make the USB power concern a moot point?

Regarding power - I am thinking of getting a UPS with surge protection  or a 10 plug AC power strip  I have that also has surge protection to plug the RPI4s, POD miners and USB hubs with the usb miners into.  Any concerns there? 
I am also thinking that if I go the multiple RPI4 host route, I can then split the electricity draw between multiple wall outlets so not to overload one wall outlet.

Thoughts and guidance much appreciated.

[kano]

As per the 2nd post,
There is a new Windows binary for v4.12.1 available now at https://kano.is/cgminer.php

2022-12-21 Update to improve performance and tuning. Windows binary updated. https://kano.is/gekko.php updated.

[kano]

...
Thoughts and guidance much appreciated.

The basics is that an R909 is equivalent to 6xCompacF - but requires no power from USB to run the chips.
Basically cgminer is now running 6xBM1397 chips in an R909, rather than one chip in a CompacF.
The R909 gets it's mining power from the power connectors on the R909 itself, so no hub or any hub will do.

Simplest solution is to just run another RPi for the R909.
It should also be OK running 1xR909 and 4xCompacF on one RPI4.
Then put the older miners on a separate RPi.

As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

[cygan]

✂
As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

thank you for the note...
i was actually planning to run 2x r909 and 2x hubs on my rpi4 with 6 compacf sticks in total...
could this work well, what do you think?

[kano]

✂
As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

thank you for the note...
i was actually planning to run 2x r909 and 2x hubs on my rpi4 with 6 compacf sticks in total...
could this work well, what do you think?

Let me know how it goes

[cygan]

✂
As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

thank you for the note...
i was actually planning to run 2x r909 and 2x hubs on my rpi4 with 6 compacf sticks in total...
could this work well, what do you think?

Let me know how it goes

i will definitely do it.
but what could happen in the worst case if the rpi4 can't handle all the hardware?

[kano]

✂
As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

thank you for the note...
i was actually planning to run 2x r909 and 2x hubs on my rpi4 with 6 compacf sticks in total...
could this work well, what do you think?

Let me know how it goes

i will definitely do it.
but what could happen in the worst case if the rpi4 can't handle all the hardware?

Worst case, you'll get less hash rate and all that fun API data I've added will tell you why.

[AFox]

Got my R909 unit yesterday.

It was equipped with an Artic F8 fan, that has a good airflow : 52.7 m³/h at 22.5 dB
Replaced by a Noctua NF-A8 PWM : 55.5 m³/h at 17.7 dB

Plugged it to a Raspberry Pi 3 Model B, Quad Core CPU 1.2 GHz, 1 GB, with Raspberry Pi OS Lite.
It's using around 38~40% of the CPU. Edit: The CPU percentage goes to 400% on my model. See Kano's explanation here (link).



I've turned the voltage knob clockwise, to add voltage.
It's running at around 2.061 Th/s for 99 watts at the plug (including power supply, Raspberry Pi and R909).
The heatsink of the mining chips are around 43°c.
I'm not used to be running CGMiner, what numbers should I look at for the "rejected shares" and "hardware errors" ?

[willi9974]

Got my R909 unit yesterday.

...

Thanks for the Information.
is it possible to show us the power supply and could you write a short installation stepp info? What raspi image do you use, and what for software package you have installed and what have you changed in the configuration... ans so one...

Best regards,
Willi

[sidehack]

Nice, 99W all in. 2.055TH is about 97% of ideal for that speed, which is pretty solid considering 525MHz is an overclock. I'm pretty proud of the R909.

[AFox]

is it possible to show us the power supply and could you write a short installation stepp info? What raspi image do you use, and what for software package you have installed and what have you changed in the configuration... ans so one...

For the power supply, I'm using a +10 year old Corsair TX650 (650w, 80Plus Bronze). It won't be the most efficient power supply, because of it's age.
Also power supplies are the most efficient when you use 50% of their total rated output power. But hard to find (impossible?) a 200w 80Plus Platinum computer power supply. And I didn't want to buy a new one, to only save ~1$ per month of electricity.
It is kept on, with a "jump start" adapter, like these : https://i.imgur.com/ERai2HX.jpg
And I use the 6-pin PCIe connector on the R909. In theory that plug is rated for 75watts (link). But I don't worry about that. Just check it's well plugged in.

For the configuration and installation of the Raspberry Pi, I've got an old model, so I went with the command line OS, no fancy interface or desktop.
To "flash" the SD card, I used Raspberry Pi Imager (link), selected Raspberry Pi OS Lite, and in the advance options activated "Enable SSH" giving a username and password. I'm using a SSH connection to connect to the Raspberry Pi.

Once everything is copied, put the SD card in the Raspberry, and turn it on.
Connect the Raspberry to the network. You'll need it's IP address to connect to it. You can find it on the router, or with a program like Advance IP Scanner (link).
Once you found the IP address, you can connect to it with the program Putty (link).
It will ask for the username and password you entered earlier.

For the installation of CGMiner, I used Kano's website : https://kano.is/gekko.php
sudo apt-get update
sudo apt-get upgrade -y
sudo apt-get install -y build-essential autoconf automake libtool pkg-config libcurl4-openssl-dev libudev-dev libusb-1.0-0-dev libncurses5-dev zlib1g-dev git
cd
git clone https://github.com/kanoi/cgminer.git
cd cgminer
CFLAGS="-O2 -march=native -fcommon" ./autogen.sh --enable-gekko --enable-icarus
make
sudo apt-get install -y openjdk-8-jre-headless

To run CGminer, and start mining, I go in the CGminer folder (cd CGMiner), and enter this command :
sudo ./cgminer -o mining pool -u btc address -p x --gekko-start-freq 400 --gekko-r909-freq 525
I saw somewhere that you should be able to enter that command without "sudo". But if I don't enter "sudo", it can't access the USB peripherals. I'm too lazy to find the fix.

I use "Screen" to be able to run multiple programs in the terminal (CGMiner and HTop) : https://raspi.tv/2012/using-screen-with-raspberry-pi-to-avoid-leaving-ssh-sessions-open
I use "HTop" to see the CPU and RAM usage : https://linuxhint.com/raspberry-pi-cpu-usage-monitoring/

Nice, 99W all in. 2.055TH is about 97% of ideal for that speed, which is pretty solid considering 525MHz is an overclock. I'm pretty proud of the R909.

I'm happy with it also. I ordered six Compac F, before the announcement of the R909. I ended up not using the Compac F, as the R909 is much easier to power, is compact, and needs only one fan to cool down (compared to the Compac F that needs one fan each, plus the USB Hub that has a fan also).




For my question what numbers should I look at for the "rejected shares" and "hardware errors", if I looked up correctly, these are the calculation to get the "rejected shares" and "hardware errors" percentage :

Number taken from my earlier screen capture. You need to take the values that are at the top
rejected shares percentage  =  R * 100 / (R+A+HW)  =  6944 * 100 / (6944+14955024+7616)  =  0.046%
hardware errors percentage  =  HW * 100 / (R+A+HW)  =  7616 * 100 / (6944+14955024+7616)  =  0.051%

If these calculation are correct, I think I have some headroom to lower the voltage, and watt usage. 2Th/s was my goal, I don't want to push it more.
On graphic cards, I'm happy when these values are bellow 1%.

[kano]

The Rejects depend on your connection to the pool.
Also note that on linux the CPU% has a max based on the number of threads/cores.
So on an RPi if it's 2 cores the max is 200% and 4 cores the max is 400%
On an intel CPU they all have 2 threads per core, so a 4 core intel CPU would have a max of 800%

Edit: also be aware that hw errors on the gekko driver reports a correct value, that literally no other driver or miner does.
The hw error count is the difficulty of the errors, not the count of errors.
Since the R909 mines internally at 64 difficulty, the number of errors returned is your number/64 - however that's not meaningful, since each error represents 64 difficulty and thus comparing apple with apples for hw vs accepted/rejected at the top of the screen requires making hw based on the internal work difficulty. I had to change the hw code in cgminer to allow this correction (back when I added the ferrari)

Also the lower accepted rejected count on your display doesn't mean much since share difficulty can change, so that's no longer comparing equally weighted numbers, just look at the top A/R/HW.

[willi9974]

Many thanks to your guide and your effort to post the information !!!

THANKS !!!

[RKG456]

Got my R909 unit yesterday.

It was equipped with an Artic F8 fan, that has a good airflow : 52.7 m³/h at 22.5 dB
Replaced by a Noctua NF-A8 PWM : 55.5 m³/h at 17.7 dB

Plugged it to a Raspberry Pi 3 Model B, Quad Core CPU 1.2 GHz, 1 GB, with Raspberry Pi OS Lite.
It's using around 38~40% of the CPU. Edit: The CPU percentage goes to 400% on my model. See Kano's explanation here (link).

I've turned the voltage knob clockwise, to add voltage.
It's running at around 2.061 Th/s for 99 watts at the plug (including power supply, Raspberry Pi and R909).

How much of a clockwise turn did you do? 1/8, 1/4, 1/2?

Also you mentioned you are not using the compac Fs.   Are planning on using them eventually or selling them?    You could get another 1.8TH with those.

[RKG456]

I just took a peek.. Bitcoinmerch website shows to have 4 left in stock for those that may still be looking for the 909. That is where I bought mine. They are bundling with the power brick.

[AFox]

Also note that on linux the CPU% has a max based on the number of threads/cores

Didn't know Linux worked that way. My original post was updated.

Many thanks to your guide and your effort to post the information !!! THANKS !!!

You're welcome. I hope it can help a lot of other people also. It can be applied for those that use Compac F, replace "--gekko-r909-freq 525" by "--gekko-compacf-freq 525".
Also, I use "525" frequency in my setup, but I recommend starting with 400, and gradually going up. To see if there's a point where it overheats (I have an infrared thermometer) and/or gets a lot of hardware errors.

How much of a clockwise turn did you do? 1/8, 1/4, 1/2?
Also you mentioned you are not using the compac Fs.   Are planning on using them eventually or selling them?    You could get another 1.8TH with those.

I barely turned it. It must have been 1/8. I took pictures before turning it. I'll post "before" and "after" pictures, but I don't have access to it now, to take an "after" picture.
I'll be using the Compac F's
Edit: Here are the before and after pictures https://imgur.com/a/F1V84qf

[os2sam]

Got my R909 unit yesterday.

Great.  It really was "just in time for Christmas"!

Merry Christmas all.
Sam

[cygan]

i finally connected my 2 r909 pod miners to my existing setup today
on the picture you can see the two new r909 and on top of it a gekoo usb hub with another 3 compacf sticks. i have 3 more compacf miners, but since my rpi4 has no usb-port left, they can't be used now and in addition one of the usb hubs almost died (one usb pair still works)...
the whole fun consumes now ~250watt and the r909 fans are very quiet in my opinion ... but as AFox has already done, i will also replace it with the noctua nf-a8 pwm



and this is how the whole thing looks like at the moment:

Code:

cgminer version 4.12.1 - Started: [2022-12-22 17:14:20.873]
--------------------------------------------------------------------------------
 (5s):4.787T (1m):4.870T (5m):4.807T (15m):4.773T (avg):4.706Th/s
 A:6051658  R:0  HW:2768  WU:65748.6/m
 Connected to solo.ckpool.org diff 3.08K with stratum as user cygan
 Block: 7034e12b...  Diff:35.4T  Started: [18:46:51.507]  Best share: 3.88M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10051796: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 373.6G / 319.3Gh/s WU: 4460.3/m
 1: GSF 10051619: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 351.5G / 351.7Gh/s WU: 4913.7/m
 2: GSF 10053622: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 357.5G / 342.6Gh/s WU: 4786.4/m
 3: GSF 10070003: BM1397:06+ 525.00MHz T:515 P:523 (4:2)   | 88.7% WU:^83% | 1.785T / 1.760Th/s WU:24588.4/m
 4: GSF 10070009: BM1397:06+ 525.00MHz T:525 P:513 (4:2)   | 91.7% WU: 91% | 1.968T / 1.932Th/s WU:26999.8/m
--------------------------------------------------------------------------------

and here is the usage of the rpi4:



do you see any potential for improvement in performance?

[AFox]

i will also replace it with the noctua nf-a8 pwm
do you see any potential for improvement in performance?

You won't regret the fan change. Even less noise, and less vibration as the Noctua fans have anti-vibration pads on each corner.

To get more performance, you could turn the voltage knob clockwise, giving it more voltage. The hashrate should increase.
Here's pictures of the knob on mine (before/stock, and after) : https://imgur.com/a/F1V84qf

[100knot2dae]

Is it possible to disable the magenta LEDs (resp. the LEDs in general) via jumper or DIP switch on the R909? Or via driver/software?

[sidehack]

Nope

[gt_addict]

Is it possible to disable the magenta LEDs (resp. the LEDs in general) via jumper or DIP switch on the R909? Or via driver/software?

Just get a Sharpie or similar and put a little dab in the leds. It will "dim" them slightly. Personally I love the blinky lights 

[cygan]

✂
As reported in my pool discord #gekko channel, I've run 2xR909 and 2xCompacF on a single RPi4 and it uses about 50% of the 400% of CPU.

thank you for the note...
i was actually planning to run 2x r909 and 2x hubs on my rpi4 with 6 compacf sticks in total...
could this work well, what do you think?

Let me know how it goes

on my rpi4 are now 2 r909 and 2 gekko hubs with a total of 4 compacf sticks connected
the performance is also impressive

Code:

cgminer version 4.12.1 - Started: [2022-12-24 11:22:54.187]
--------------------------------------------------------------------------------
 (5s):5.025T (1m):5.340T (5m):5.293T (15m):5.254T (avg):5.145Th/s
 A:8849622  R:2850  HW:2208  WU:71879.2/m
 Connected to solo.ckpool.org diff 2.85K with stratum as user cygan
 Block: 1dbb8e6f...  Diff:35.4T  Started: [13:13:09.645]  Best share: 7.62M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10051796: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 298.7G / 316.7Gh/s WU: 4424.9/m
 1: GSF 10051619: BM1397:01+ 525.00MHz T:515 P:525 (24:12) |  100% WU:100% | 316.5G / 347.5Gh/s WU: 4854.7/m
 2: GSF 10053622: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 290.3G / 344.1Gh/s WU: 4807.7/m
 3: GSF 10053618: BM1397:01+ 525.00MHz T:515 P:525 (24:12) |  100% WU:100% | 368.8G / 343.2Gh/s WU: 4794.8/m
 4: GSF 10070003: BM1397:06+ 525.00MHz T:515 P:521 (4:2)   | 88.7% WU:^84% | 2.109T / 1.783Th/s WU:24908.5/m
 5: GSF 10070009: BM1397:06+ 525.00MHz T:515 P:525 (4:2)   | 98.5% WU:^95% | 1.613T / 2.010Th/s WU:28088.4/m
--------------------------------------------------------------------------------

[RKG456]

Got my R909 delivered today and got all my miners going except the R808 I got off of ebay appears to have pooped out. It was working then got stuck on Found 0 Chips.. But that is not the point of this post.
I have to keep an eye on the Compac-F (item 5) as it appears to be sluggish. I was reporting 35%  until I captured this output. Right when I chose copy it jumped to 100% but as you can see the hash is way lower than the other compac-f (3, 4 and 6).  With this setup though I have seen the overall 5s hash shoot as high as 3.799T.

Now to start saving $$ for another R909.

Code:

 cgminer version 4.12.1 - Started: [2022-12-30 15:12:45.297]
--------------------------------------------------------------------------------
 (5s):3.215T (1m):3.442T (5m):2.976T (15m):1.703T (avg):3.212Th/s
 A:464622  R:39064  HW:355  WU:44943.4/m
 Connected to solo.ckpool.org diff 2.42K with stratum as user USER.NPandT
 Block: 68b3a9c7...  Diff:35.4T  Started: [15:13:51.054]  Best share: 262K
--------------------------------------------------------------------------------
[U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070012: BM1397:06+ 550.00MHz T:550 P:550 (3:2)     |  100% WU:^96% | 2.294T / 2.134Th/s WU:29835.9/m
 1: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 99.1% WU:^95% | 81.85G / 80.95Gh/s WU: 1130.9/m
 2: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 98.8% WU:^94% | 83.98G / 80.72Gh/s WU: 1127.8/m
 3: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:445 (28:14)   |  100% WU:^98% | 272.1G / 231.0Gh/s WU: 3227.6/m
 4: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 405.9G / 284.4Gh/s WU: 3973.2/m
 5: GSF 10050104: BM1397:01+ 440.00MHz T:440 P:347 (28:15)   |  100% WU: 58% | 119.0G / 134.0Gh/s WU: 1872.6/m
 6: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 339.5G / 270.1Gh/s WU: 3775.4/m
--------------------------------------------------------------------------------
 [2022-12-30 15:23:31.244] Accepted 10ae7164 Diff 3.93K/2422 GSF 4
 [2022-12-30 15:23:34.321] 5: GSF 3 - setting [all] frequency to 420.00MHz (420.00) (168/2/1/5)
 [2022-12-30 15:23:34.322] 5: GSF 3 - new frequency 415.00MHz -> 420.00MHz
 [2022-12-30 15:23:38.089] 5: GSF 3 - setting [all] frequency to 425.00MHz (425.00) (170/2/1/5)
 [2022-12-30 15:23:38.091] 5: GSF 3 - new frequency 420.00MHz -> 425.00MHz
 [2022-12-30 15:23:42.391] 5: GSF 3 - setting [all] frequency to 430.00MHz (430.00) (172/2/1/5)
 [2022-12-30 15:23:42.393] 5: GSF 3 - new frequency 425.00MHz -> 430.00MHz
 [2022-12-30 15:23:43.882] Accepted 1a8fdabb Diff 2.47K/2422 GSF 0
 [2022-12-30 15:23:47.545] 5: GSF 3 - setting [all] frequency to 435.00MHz (435.00) (174/2/1/5)
 [2022-12-30 15:23:47.546] 5: GSF 3 - new frequency 430.00MHz -> 435.00MHz
 [2022-12-30 15:23:48.225] Accepted 17b81fcb Diff 2.76K/2422 GSF 2
 [2022-12-30 15:23:48.794] Accepted 06bbbbce Diff 9.73K/2422 GSF 0
 [2022-12-30 15:23:51.762] Accepted 0535c27c Diff 12.6K/2422 GSF 2
 [2022-12-30 15:23:52.555] 5: GSF 3 - setting [all] frequency to 440.00MHz (440.00) (176/2/1/5)
 [2022-12-30 15:23:52.557] 5: GSF 3 - new frequency 435.00MHz -> 440.00MHz

[sidehack]

If it's running slow, best bet is probably pull it, adjust the voltage up a little, and plug it back in.

[RKG456]

If it's running slow, best bet is probably pull it, adjust the voltage up a little, and plug it back in.

How do you adjust the voltage on a compac-F?  I do not see a dial.



Also I just got a notice of Ticket Value confirmed after 5000 nonces..   what does that mean?  - it looked to be on the R909.

[sidehack]

Tiny knob in the corner, same as all other Gekko sticks. Slight clockwise turn.

Ticket value confirmed message is a leftover from some early testing. Just means we understood the format of the ticket mask register assignment correctly when we reverse-engineered the protocol.

[RKG456]

Tiny knob in the corner, same as all other Gekko sticks. Slight clockwise turn.

Ticket value confirmed message is a leftover from some early testing. Just means we understood the format of the ticket mask register assignment correctly when we reverse-engineered the protocol.

Thanks Sidehack.  The stick seems better now but just a little slow compared to the others.  I will tweak the tuning.


Here is a video of the blinking lights.  You would not think this would generate that much heat, but so far it is enough to raise the overall temp in a 500 sq ft room by 4 degrees F.  That corner went from 72 - 73 degrees to 77.7 degrees.  However the HP Power supply and the  USB hub power bricks are also putting off some heat.

https://youtu.be/ErYD_5Pi5Zc

[MoparMiningLLC]

Anyone willing to buy one or two with fiat from 419mining and sell to me for BTC? Or willing to trade 2 of these for 4 r606’s?

[kano]

Got my R909 delivered today and got all my miners going except the R808 I got off of ebay appears to have pooped out. It was working then got stuck on Found 0 Chips..
...

Check it's not a software version issue - i.e. if it was working before, use the version of software from before just for the R808 and see if it's ok.
I've no idea if the R808 works with the current code.
I've run R909, R606, CompacF, NewPac, 2Pac and Compac on the new version (and you should use the new 4.12.1 version for those six)

[n0nce]

Got my R909 unit yesterday.

It was equipped with an Artic F8 fan, that has a good airflow : 52.7 m³/h at 22.5 dB
Replaced by a Noctua NF-A8 PWM : 55.5 m³/h at 17.7 dB

Are you sure that you got the PWM fan? If so, it is now running at 100% fixed fan speed, since the R909 only outputs a 3-pin DC fan signal and Noctua PWM fans default to 100% fan speed if they get a (any) 3-pin DC fan signal.

@sidehack does the R909 have any fan speed control at all or is it just outputting a fixed 12V? In that case, it wouldn't matter (at least in case of Noctua fans!) which type we choose.

I need to replace my fan, too, since I find it too loud (at least the type of noise it's making is annoying me, somehow).

[AFox]

Are you sure that you got the PWM fan? If so, it is now running at 100% fixed fan speed, since the R909 only outputs a 3-pin DC fan signal and Noctua PWM fans default to 100% fan speed if they get a (any) 3-pin DC fan signal.

Yes, I confirm I bought the PWM model. The PWM part isn't important for the R909, but it is the 80mm Noctua model that has the most airflow.

[n0nce]

Are you sure that you got the PWM fan? If so, it is now running at 100% fixed fan speed, since the R909 only outputs a 3-pin DC fan signal and Noctua PWM fans default to 100% fan speed if they get a (any) 3-pin DC fan signal.

Yes, I confirm I bought the PWM model. The PWM part isn't important for the R909, but it is the 80mm Noctua model that has the most airflow.

I see, that makes a lot of sense! Only one question: the pin header outputs constant 12V? So, no control at all even on DC (3-pin) fans? Because in that case, I'd get the PWM, too. But if it meant losing fan control, I'd prefer an actual DC fan whose RPM can be controlled by the miner. No reason to 'over-cool' the hashboard, if there's more noise reduction to be had, right?

[AFox]

I see, that makes a lot of sense! Only one question: the pin header outputs constant 12V? So, no control at all even on DC (3-pin) fans? Because in that case, I'd get the PWM, too. But if it meant losing fan control, I'd prefer an actual DC fan whose RPM can be controlled by the miner. No reason to 'over-cool' the hashboard, if there's more noise reduction to be had, right?

It seems the miner only outputs 12v, with no control on the voltage.
You could use the Noctua NA-FC1 (link) to control a PWM fan.

I didn't play with the fan speed, because with the Noctua NF-A8 PWM, and at 525MHz, the heatsink is at 55°c; you can barely keep your finger on it. You need the fan at full blast to cool it properly.
You'll probably need to lower the frequency, to be able to lower the fan speed.

[sidehack]

Yep. Just like with the R606, I didn't put fan speed control on here because the fan's quiet enough even at full speed to be pretty unintrusive, so why bother with less than full speed if there's probably going to be a heat/efficiency penalty?

[RKG456]

Is there a general thread for Gekkoscience miners?  Not just for the  R909?

I will mention this here since there is an R909 in the mix, but It may be better suited for a different thread. Let me know.  I tried mining the nicehash pool this morning via cgminer, and though it works, the difficulty seems to be a little much. it is very slow and the R909 seems to hang at around 89% but still hashes at around 2.5TH/s.  The Compac-Fs see to be the ones that struggle the most as a few of them will reset the frequency down to 200mhz and ramp back up.   I am running 1 R9-9, 4 Compac-Fs and 2 New Pacs with an over all hash rate of 3.1TH to as high as 4.2TH/s.  The pool settles on a difficulty of 40k and and overall it does get excepted shares, just really slow to handle them.  That being said, I was curious if anyone else had tried to mine nicehash with usb miners.  Is it worth it or is there an increased risk of damage to the miners?  Anyone have suggested difficulty for nicehash pool?

Here is what I used: NiceHash External Mining:
sudo ./cgminer -o stratum+tcp://sha256.auto.nicehash.com:9200 -u WALLET-ADDRESS.Workername -p x --gekko-newpac-freq 350 --gekko-compacf-freq 400 --gekko-r909-freq 550

I was thinking of trying it again with the --suggest-diff 20000  flag and then with --suggest-diff 10000 flag to see how those work.  I have been mining ckpool with --suggest-diff 10000 flag  to see how that went.  it seems stable just getting about half the shares than if I let it auto set the difficulty.  When I do let it auto set on ckpool it usually sets it somewhere between 2400 and 3400. 

Here is what I used to mine ckpool: 
sudo ./cgminer -o stratum+tcp://solo.ckpool.org:3333 -u WALLET-ADDRESS.Workername  -p x --gekko-newpac-freq 375 --gekko-compacf-freq 450 --gekko-r909-freq 575 --suggest-diff 10000

Earlier today ckpool reported these stats for my worker. These are the frequency same settings I normally run only difference is the suggested difficulty.

Code:

{
 "hashrate1m": "6.14T",
 "hashrate5m": "4.81T",
 "hashrate1hr": "3.68T",
 "hashrate1d": "3.17T",
 "hashrate7d": "1.14T",
 "lastshare": 1672677213,
 "workers": 1,
 "shares": 196345749,
 "bestshare": 283475270.1970135,
 "bestever": 283475270,
 "worker": [
  {

Code:

 cgminer version 4.12.1 - Started: [2023-01-02 11:04:46.777]
--------------------------------------------------------------------------------
 (5s):3.477T (1m):3.632T (5m):3.624T (15m):3.608T (avg):3.589Th/s
 A:9450000  R:10000  HW:4260  WU:50140.9/m
 Connected to solo.ckpool.org diff 10K with stratum as user bc1qwgc4p4gf6qlpv0m.NPandT
 Block: 3df2368c...  Diff:35.4T  Started: [14:08:11.563]  Best share: 21M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070012: BM1397:06+ 575.00MHz T:575 P:575 (4:2)     | 98.7% WU: 97% | 1.991T / 2.256Th/s WU:31524.2/m
 1: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 99.8% WU: 98% | 82.56G / 83.99Gh/s WU: 1173.4/m
 2: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 98.0% WU: 98% | 88.03G / 83.83Gh/s WU: 1171.2/m
 3: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:450 (29:14)   |  100% WU:100% | 301.0G / 270.4Gh/s WU: 3777.5/m
 4: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 290.8G / 297.1Gh/s WU: 4150.3/m
 5: GSF 10050104: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 271.5G / 299.6Gh/s WU: 4185.5/m
 6: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (29:14)   |  100% WU:100% | 415.4G / 297.7Gh/s WU: 4158.8/m
--------------------------------------------------------------------------------
 [2023-01-02 14:06:44.553] Accepted 048725ba Diff 14.5K/10000 GSF 0
 [2023-01-02 14:06:56.428] Accepted 025d6339 Diff 27.7K/10000 GSF 0
 [2023-01-02 14:07:10.289] Accepted 04fd253a Diff 13.1K/10000 GSF 4
 [2023-01-02 14:07:24.395] Accepted 0859ff5b Diff 2.01M/10000 GSF 0
 [2023-01-02 14:07:25.985] Accepted 0476518e Diff 14.7K/10000 GSF 0
 [2023-01-02 14:07:30.342] Accepted 0558b45a Diff 12.3K/10000 GSF 0
 [2023-01-02 14:07:48.383] Accepted 02d12650 Diff 23.3K/10000 GSF 4
 [2023-01-02 14:08:09.490] Accepted 0168869e Diff 46.5K/10000 GSF 3
 [2023-01-02 14:08:11.563] Stratum from pool 0 detected new block at height 770072
 [2023-01-02 14:08:25.631] Accepted 030ba34c Diff 21.5K/10000 GSF 3
 [2023-01-02 14:08:52.728] Accepted 01897353 Diff 42.6K/10000 GSF 2
 [2023-01-02 14:09:07.898] Accepted 0392977d Diff 18.3K/10000 GSH 1
 [2023-01-02 14:09:11.982] Accepted 045f91be Diff 15K/10000 GSF 0
 [2023-01-02 14:09:13.723] Accepted 026ca67b Diff 27K/10000 GSF 0
 [2023-01-02 14:09:18.526] Accepted 40db734b Diff 259K/10000 GSF 1

[os2sam]

Is there a general thread for Gekkoscience miners?  Not just for the  R909?

I am running 1 R9-9, 4 Compac-Fs and 2 New Pacs

All the Gekko threads are pretty close to the  top of the list so you should be able to see them.

Newpac:
https://bitcointalk.org/index.php?topic=5053833.msg47068859#msg47068859

Compac F:
https://bitcointalk.org/index.php?topic=5355470.msg57735149#msg57735149

[NotFuzzyWarm]

Is there a general thread for Gekkoscience miners?  Not just for the  R909?

I am running 1 R9-9, 4 Compac-Fs and 2 New Pacs

All the Gekko threads are pretty close to the  top of the list so you should be able to see them.
Newpac:
https://bitcointalk.org/index.php?topic=5053833.msg47068859#msg47068859

Compac F:
https://bitcointalk.org/index.php?topic=5355470.msg57735149#msg57735149

Long ago Kano recommended that the slower original Compacs, 2Pacs & NewPacs be ran in a separate instance of cgminer and the F's & obviously the R909 in their own instance. You can of course still run them off of the same hub and code - cgminer is perfectly happy running multiple instances of itself, AIR it is due to the vastly different data rates involved with the chips and how the CPU(s) & cgminer has to process the data for them.

[sidehack]

Is that a 2.26TH average speed from the R909? Nice. The fastest I've pushed one stable was 2.1TH so, you know, cool.

[RKG456]

Is that a 2.26TH average speed from the R909? Nice. The fastest I've pushed one stable was 2.1TH so, you know, cool.

I have been running the last round of testing with the R909 at 600mhz and it appears stable. Difficulty is locked at 10K and it seems to be handling it.
575mhz on the R909 seems to be the sweet spot for mine though. The Best share resets every time i restart the miner.. So far the best share I have seen is with the R909 at 550mhz and it was 293M and the difficulty was at 3211.  
EDIT: I did have to turn the blue know about 1/8th Clockwise to get 600mhz to be stable otherwise the R909 would not work well above 575mhz.

Code:

cgminer version 4.12.1 - Started: [2023-01-02 18:44:28.984]
--------------------------------------------------------------------------------
 (5s):4.040T (1m):3.779T (5m):3.681T (15m):3.670T (avg):3.671Th/s
 A:4280000  R:0  HW:2747  WU:51295.3/m
 Connected to solo.ckpool.org diff 10K with stratum as user bc1qwgc4p4gf6qlpv0mh95ju6rtqvlhmnwtyn87dpa.NPandT
 Block: 740932e4...  Diff:35.4T  Started: [19:49:00.595]  Best share: 4.3M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070012: BM1397:06+ 600.00MHz T:600 P:600 (3:2)     | 95.7% WU: 97% | 2.813T / 2.342Th/s WU:32720.5/m
 1: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 97.2% WU: 98% | 86.06G / 83.85Gh/s WU: 1171.4/m
 2: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 99.4% WU: 98% | 86.39G / 83.46Gh/s WU: 1166.0/m
 3: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:449 (28:14)   |  100% WU:100% | 268.3G / 265.7Gh/s WU: 3711.9/m
 4: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 328.1G / 300.6Gh/s WU: 4199.6/m
 5: GSF 10050104: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 276.8G / 298.5Gh/s WU: 4169.8/m
 6: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 292.9G / 297.5Gh/s WU: 4156.0/m
--------------------------------------------------------------------------------
 [2023-01-02 20:16:18.754] Accepted 23d05810 Diff 468K/10000 GSF 2
 [2023-01-02 20:16:29.731] Accepted 0498d7c1 Diff 14.3K/10000 GSF 0
 [2023-01-02 20:16:33.821] Accepted 059fe3a0 Diff 11.7K/10000 GSF 0
 [2023-01-02 20:16:58.581] Accepted 0159b5a0 Diff 48.5K/10000 GSH 0
 [2023-01-02 20:16:59.384] Accepted 02e3922a Diff 22.7K/10000 GSF 0
 [2023-01-02 20:17:03.497] Accepted 04fcfc43 Diff 13.1K/10000 GSH 1
 [2023-01-02 20:17:07.362] Accepted 017e3b91 Diff 43.9K/10000 GSF 1
 [2023-01-02 20:17:14.330] Accepted 0323fce4 Diff 20.9K/10000 GSF 1
 [2023-01-02 20:17:21.062] Accepted 02faca58 Diff 22K/10000 GSF 0
 [2023-01-02 20:17:34.834] Accepted 02a5b268 Diff 24.8K/10000 GSF 0
 [2023-01-02 20:17:47.444] Accepted 02063572 Diff 32.4K/10000 GSF 0
 [2023-01-02 20:17:49.857] Accepted 01f438d1 Diff 33.5K/10000 GSF 0
 [2023-01-02 20:18:01.234] Accepted 052019bb Diff 12.8K/10000 GSF 0
 [2023-01-02 20:18:08.316] Accepted 01c339b1 Diff 37.2K/10000 GSF 0
 [2023-01-02 20:18:17.076] Accepted 05934f47 Diff 11.8K/10000 GSF 0

[RKG456]

So today I have been letting it run and letting cgminer and the pool manage the difficulty, and it is set low at 2587.  Yet somewhere along the line today, this little setup pulled a Best share of 1.15G.

 

Code:

cgminer version 4.12.1 - Started: [2023-01-03 08:12:27.546]
--------------------------------------------------------------------------------
 (5s):3.382T (1m):3.570T (5m):3.589T (15m):3.575T (avg):3.567Th/s
 A:25203274  R:49005  HW:12163  WU:49832.5/m
 Connected to solo.ckpool.org diff 2.59K with stratum as user MyWallert.NPandT
 Block: 60df4130...  Diff:34.1T  Started: [16:34:05.450]  Best share: 1.15G
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070012: BM1397:06+ 575.00MHz T:575 P:575 (3:2)     | 99.1% WU: 96% | 2.258T / 2.229Th/s WU:31141.5/m
 1: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 97.3% WU: 98% | 89.83G / 84.13Gh/s WU: 1175.4/m
 2: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 98.4% WU: 98% | 86.88G / 84.08Gh/s WU: 1174.7/m
 3: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:445 (28:14)   |  100% WU:100% | 299.3G / 270.0Gh/s WU: 3772.5/m
 4: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 392.4G / 298.9Gh/s WU: 4175.3/m
 5: GSF 10050104: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 359.9G / 299.3Gh/s WU: 4181.4/m
 6: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 301.3G / 301.5Gh/s WU: 4211.8/m

[Atrax]

My 909 was working without any flaw.
Now, after some days, I get this messages:

[2023-01-06 13:25:48.954] Started cgminer 4.12.1-wrk                    
 [2023-01-06 13:25:48.955] Loaded configuration file /home/gilkane/.cgminer/cgminer.conf                    
 [2023-01-06 13:25:48.956] Probing for an alive pool                    
 [2023-01-06 13:25:49.173] Pool 0 difficulty changed to 8192                    
 [2023-01-06 13:25:49.174] Pool 2 difficulty changed to 8192                    
 [2023-01-06 13:25:49.345] Pool 1 difficulty changed to 262144                    
 [2023-01-06 13:25:50.255] 0: GSF 0 - Toggling ASIC nRST to reset                    
 [2023-01-06 13:25:51.494] 0: GSF 0 - R909 Bitcoin Miner (10070007)                    
 [2023-01-06 13:25:51.497] Network diff set to 34.1T                    
 [2023-01-06 13:25:53.394] 0: GSF 0 - set ticket to 0xfc/64 work 128/128.0                    
 [2023-01-06 13:25:53.597] 0: GSF 0 - setting bauddiv : 00 61 (ftdi/2)                    
 [2023-01-06 13:25:53.673] 0: GSF 0 - setting [all] frequency to 400.00MHz (400.00) (160/2/1/5)                    
 [2023-01-06 13:25:53.694] GSF 0 failure, disabling!    

Frequency was originally set on 450 in my configuration. Changed it to 400, but still the same result.

Rebooting the Raspi didn't change anything.

edit:
Now, since a while, I only got this message:

[2023-01-06 13:33:51.305] GekkoScience detect (1:11) failed to initialise (incorrect device?), resetting

Any advice where I should begin with finding the error?

[Atrax]

Get the same behavior when I plug in the 909 on my desktop computer. So I guess there's something I should be able to fix on the 909?

[Atrax]

Okay, seems it is a power problem. Had to find a more stable PSU for we have here massive problems with the current in our house. Hope it will be better once I put the device out of the testing environment in my house and  on a solar device.

[RKG456]

For those still looking for an R909. I see them listed on Amazon for $620 with free prime shipping.  Sold by Eyeboot and ships from Amazon.

https://www.amazon.com/gp/product/B0BQZ35B6G/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1 

[MoparMiningLLC]

For those still looking for an R909. I see them listed on Amazon for $620 with free prime shipping.  Sold by Eyeboot and ships from Amazon.

https://www.amazon.com/gp/product/B0BQZ35B6G/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1 

419mining has them as well for 599 but they dont take btc - neither does amazon, so I am unable to order.

[sidehack]

R909 seem to be a bit particular about power, yeah. The fairly cheap brick on my linux laptop is noisy and I had to replace it before it'd detect the R909 over USB. Sometimes it'd load up and work for a few seconds then drop out again, but most of the time it just doesn't work. I ended up in a pinch just using a second lab PSU for the laptop, just like the one powering my test R909, and it's been running beautifully at 600MHz for the last day on that setup.

Didnt have the problem on a Pi controller, because it was running off the same clean power source as the pod. Low quality power for either the pod or controller could cause ground loop noise issues that throw off the USB signal voltages and make for a poor connection.

[yettihead]

Bitcoinmerch has them and I believe they take bitcoin through Coinbase Commerce.

https://bitcoinmerch.com/products/bitcoin-merch%C2%AE-geoscience-terminus-sha256-bitcoin-r909-pod-miner-led-version

For those still looking for an R909. I see them listed on Amazon for $620 with free prime shipping.  Sold by Eyeboot and ships from Amazon.

https://www.amazon.com/gp/product/B0BQZ35B6G/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1 

419mining has them as well for 599 but they dont take btc - neither does amazon, so I am unable to order.

[n0nce]

The BM1397 chips have a significant temperature coefficient, which means that they pull noticeably more power as they get warmer. Nobody at GekkoScience will be offended if a tuner or enthusiast upgrades the fan, as more airflow could mean reducing power consumption by 5-10% just from maintaining lower temperatures. These units take a standard 80mmx25mm 3/4-wire 12V case fan.

Speaking of this: is there any way to read out the chips' temperatures?
Or would we need to manually place a thermocouple and monitor it that way? I'm asking, because I replaced the stock fan and now I'm not sure whether it's getting too hot.

[kano]

So I tried to do something that I'd not suggest anyone else to do
Of course try it if you really know what you are doing and have a 5kRPM fan, high quality power and are using linux

The aim was to see how fast I can get the R909 going - and I've got a record so far, we'll see who can beat it
My production R909 has one chip not as good as the other 5, so someone with 6 perfect chips should be able to beat it.

Anyway, I put in a 5kRPM fan, set the blue screw close to maximum, set 5 chips to ... 725MHz ... and the dodgy chip to 500MHz.

Results after hours of running (that I posted in discord earlier):

Code:

(5s):5.067T (1m):5.046T (5m):5.105T (15m):5.098T (avg):5.024Th/s
 A:36261324  R:23056  HW:8786576  WU:70192.7/m
 Connected to               diff 2.88K with stratum as user
 Block: 5fd33018...  Diff:34.1T  Started: [19:45:10.943]  Best share: 84.9M
--------------------------------------------------------------------------------
 0: GSF 10008346: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 307.9G / 292.5Gh/s WU: 4086.1/m
 1: GSF 10008000: BM1397:06+ 500.00MHz T:500 P:500 (4:2)   |  100% WU: 98% | 2.265T / 1.970Th/s WU:27525.2/m
 2: GSF 10055348: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 296.0G / 286.5Gh/s WU: 4001.8/m
 3: GSF 10070004: BM1397:06+ 687.50MHz T:500 P:665 (3:2)   | 90.4% WU: 89% | 2.433T / 2.475Th/s WU:34579.3/m

Massive >20% hardware errors, but "3: GSF 10070004: BM1397:06+" powering along at 2.475Th/s

(and of course the other number there, total of >5TH for 2xR909 and 2xCompacF)

Mine on!

Edit: I should add, when you change frequencies or it has finished ramping up the frequencies, reset the averages.
I've added the details to http://kano.is/gekko.php#perf

[sidehack]

Mine with 6 good chips is sitting at 2.4T flat at 600MHz, blue knob about 1 o'clock, pulling 104 watts DC with a stock fan and pretty low HW. My old laptop craps the latency timer at any higher MHz but if I can get the Pi back running I'm gonna try to beat that record.

[cygan]

i will also try to get to the performance/record of kanos r909. but for now i have no desire to tinker with my small lottery farm.
first wait for my new gekko usb hub and then i will first adjust my two r909 and among other things replace the fans against the noctua ones

[Atrax]

R909 seem to be a bit particular about power, yeah. The fairly cheap brick on my linux laptop is noisy and I had to replace it before it'd detect the R909 over USB. Sometimes it'd load up and work for a few seconds then drop out again, but most of the time it just doesn't work. I ended up in a pinch just using a second lab PSU for the laptop, just like the one powering my test R909, and it's been running beautifully at 600MHz for the last day on that setup.

Didnt have the problem on a Pi controller, because it was running off the same clean power source as the pod. Low quality power for either the pod or controller could cause ground loop noise issues that throw off the USB signal voltages and make for a poor connection.

Yep, once the power problem has been solved, it is hashing beautifully on 600 Mhz at room temperature:

(5s):2.887T (1m):2.433T (5m):2.418T (15m):2.422T (avg):2.401Th/s   

Now measuring for a couple of days how much power it sucks before installing it in a solar system. There it will work in constantly 15° Celsius, so looking forward on how the performance will be then.

[n0nce]

among other things replace the fans against the noctua ones

For what it's worth, do be aware that Noctua does not equal absolute silence. So far, their most quiet fans in my experience are the A12x25, which even at roughly half speed (1000RPM) provide ample cooling and static pressure for my application while being almost silent. Of course, they don't fit this miner.

Other Noctua fans however, need more RPM to cool properly and are definitely audible. Even larger ones like the NF-A14 PWM.

I installed the NF-A8 PWM in the R909 earlier today and it sounds slightly higher pitched (at full 2200RPM speed) than the stock Arctic F8 at 2000RPM, although the sound is a bit quieter and more pleasant; the F8 almost sounded 'rattly' in a way; some kind of frequency was probably spiking higher than the others.
Right now, I am running the Noctua with the included LNA (low noise adapter) at 1750RPM and that makes it almost silent from a few meters away. However, it results in a bit less airflow than even the stock fan.

For now it works without issues at 525MHz and a bit over 2TH/s, but I definitely need to check my temps somehow.

[n0nce]

I've been gradually increasing frequency, without playing with the voltage knob, and so far I've reached 600MHz. Getting 2.3TH/s out of it. Pretty amazing!

Anyway, I put in a 5kRPM fan, set the blue screw close to maximum, set 5 chips to ... 725MHz ... and the dodgy chip to 500MHz.

Results after hours of running (that I posted in discord earlier):

Code:

(5s):5.067T (1m):5.046T (5m):5.105T (15m):5.098T (avg):5.024Th/s
 A:36261324  R:23056  HW:8786576  WU:70192.7/m
 Connected to               diff 2.88K with stratum as user
 Block: 5fd33018...  Diff:34.1T  Started: [19:45:10.943]  Best share: 84.9M
--------------------------------------------------------------------------------
 0: GSF 10008346: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 307.9G / 292.5Gh/s WU: 4086.1/m
 1: GSF 10008000: BM1397:06+ 500.00MHz T:500 P:500 (4:2)   |  100% WU: 98% | 2.265T / 1.970Th/s WU:27525.2/m
 2: GSF 10055348: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 296.0G / 286.5Gh/s WU: 4001.8/m
 3: GSF 10070004: BM1397:06+ 687.50MHz T:500 P:665 (3:2)   | 90.4% WU: 89% | 2.433T / 2.475Th/s WU:34579.3/m

In this overview, what frequency is shown when a device has multiple chips? On your two R909's, I can spot one at 500MHz and one at 687.5MHz. Is it the average?
I guess so.. (725*5+500)/6 = 687.5

I'm going to need to dabble with the Java API and individual frequencies, too. Right now, just using --gekko-r909-freq 600 and restarting the process when trying another setting.

[RKG456]

Thought I would provide an update with my setup. I got the 2nd R909 and 5th Compac-F up and running on a 2nd RPi4.  I have been mining ViaBTC pool with the rig setups below and the 24hour mining results are: 10-Min Avg. HashRate fluctuates between 5.5TH/s and 7.5TH/s.   All is working great and the R909s are AWESOME!

1st Rig: 1x R909, 4 Compac-Fs and 2 NewPacs.

Code:

 cgminer version 4.12.1 - Started: [2023-01-09 22:10:42.856]
--------------------------------------------------------------------------------
 (5s):4.208T (1m):3.642T (5m):3.670T (15m):3.680T (avg):3.302Th/s
 A:39440384  R:8192  HW:47379  WU:46131.1/m
 Connected to bitcoin.viabtc.io diff 4.1K with stratum as user N0F34R.NPandT
 Block: 11fe4e31...  Diff:34.1T  Started: [11:59:55.799]  Best share: 31.6M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070012: BM1397:06+ 600.00MHz T:600 P:600 (3:2)       | 96.3% WU:^87% | 2.745T / 2.104Th/s WU:29401.4/m
 1: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)    | 98.9% WU:^88% | 87.78G / 75.03Gh/s WU: 1048.2/m
 2: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)    | 98.6% WU:^88% | 82.19G / 74.92Gh/s WU: 1046.6/m
 3: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:449 (28:14)     |  100% WU:100% | 287.0G / 242.7Gh/s WU: 3390.6/m
 4: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)     |  100% WU:100% | 337.9G / 267.3Gh/s WU: 3734.2/m
 5: GSF 10050104: BM1397:01+ 450.00MHz T:450 P:450 (28:14)     |  100% WU:100% | 317.0G / 269.7Gh/s WU: 3767.1/m
 6: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (28:14)     |  100% WU:100% | 306.3G / 267.9Gh/s WU: 3743.0/m

2nd Rig: 1x R909 and 1x Compac-F. 

Code:

 cgminer version 4.12.1 - Started: [2023-01-09 08:51:23.596]
--------------------------------------------------------------------------------
 (5s):2.266T (1m):2.598T (5m):2.554T (15m):2.528T (avg):2.234Th/s
 A:50655232  R:4096  HW:60608  WU:31211.2/m
 Connected to bitcoin.viabtc.io diff 4.1K with stratum as user N0F34R.T9andCF
 Block: 11fe4e31...  Diff:34.1T  Started: [11:59:55.782]  Best share: 118M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070032: BM1397:06+ 575.00MHz T:575 P:575 (3:2)      |  100% WU:^86% | 2.211T / 1.995Th/s WU:27878.9/m
 1: GSF 10053931: BM1397:01+ 400.00MHz T:400 P:400 (31:16)    |  100% WU:100% | 259.0G / 238.5Gh/s WU: 3332.3/m
--------------------------------------------------------------------------------

Updated Rig video link: https://youtu.be/MHAziVzA6I8

[n0nce]

All is working great and the R909s are AWESOME!

They really are! It seems to me like sidehack did good work on the thermal design. Super beefy heatsinks on both sides, with nice channel for the air, all in all running great.
I've ran 600MHz for over 48h without touching the voltage knob, and got no anomalies in error rate or anything.

Now testing 650MHz just using --gekko-r909-freq 650 on all cores, stock voltage.

Updated Rig video link: https://youtu.be/MHAziVzA6I8

Nice setup! For what it's worth, you could run multiple cgminer instances all on a single Pi, or are you getting issues when too many USB devices are connected?

[RKG456]

Nice setup! For what it's worth, you could run multiple cgminer instances all on a single Pi, or are you getting issues when too many USB devices are connected?

Essentially with all of the fans being plugged in I was running out of usb ports.   The original setup for Rig 1 was running stable as it was so I didn't want to change it. Given I have many Raspberry Pi and other Single Board computers laying about, I figured I would just spit it into two mining rigs.  That way I could relocate one if need be. Plus test mining against different pools to see how things went.  Right now it is a little cramped in that one area. With Rig 1 I have a keyboard and mouse connected. With Rig 2 I am forcing myself to work it remotely.

[n0nce]

Essentially with all of the fans being plugged in I was running out of usb ports.

I see, makes sense!

So I tried to do something that I'd not suggest anyone else to do
Of course try it if you really know what you are doing and have a 5kRPM fan, high quality power and are using linux

The aim was to see how fast I can get the R909 going - and I've got a record so far, we'll see who can beat it
My production R909 has one chip not as good as the other 5, so someone with 6 perfect chips should be able to beat it.

Anyway, I put in a 5kRPM fan, set the blue screw close to maximum, set 5 chips to ... 725MHz ... and the dodgy chip to 500MHz.

Results after hours of running (that I posted in discord earlier):

Code:

(5s):5.067T (1m):5.046T (5m):5.105T (15m):5.098T (avg):5.024Th/s
 A:36261324  R:23056  HW:8786576  WU:70192.7/m
 Connected to               diff 2.88K with stratum as user
 Block: 5fd33018...  Diff:34.1T  Started: [19:45:10.943]  Best share: 84.9M
--------------------------------------------------------------------------------
 0: GSF 10008346: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 307.9G / 292.5Gh/s WU: 4086.1/m
 1: GSF 10008000: BM1397:06+ 500.00MHz T:500 P:500 (4:2)   |  100% WU: 98% | 2.265T / 1.970Th/s WU:27525.2/m
 2: GSF 10055348: BM1397:01+ 450.00MHz T:450 P:450 (28:14) |  100% WU:100% | 296.0G / 286.5Gh/s WU: 4001.8/m
 3: GSF 10070004: BM1397:06+ 687.50MHz T:500 P:665 (3:2)   | 90.4% WU: 89% | 2.433T / 2.475Th/s WU:34579.3/m

Massive >20% hardware errors, but "3: GSF 10070004: BM1397:06+" powering along at 2.475Th/s

Hey kano, I'm running 700MHz for around an hour now, without changes to the voltage screw and the NF-A8 PWM running at 2200RPM. Seems stable and good so far. I'm finally getting over 2.5TH/s (that was my goal), so I guess I beat you, with stock voltage!

Just to make sure: worst that can happen when aiming for too high clock speeds is that it crashes or returns faulty hashes, right? Or is there any 'risk' to pushing the overclock as long as the voltage screw is not touched? And the error rate is displayed through those Java API commands, right? The only way I've checked my error rate so far was the log printed when killing cgminer.

Proof:

Code:

 (5s):2.647T (1m):2.735T (5m):2.639T (15m):2.602T (avg):2.585Th/s
 A:3259855  R:0  HW:1984  WU:36123.2/m
 Connected to stratum.kano.is diff 1.62K with stratum as user
 Block: 4b258ede...  Diff:34.1T  Started: [07:05:45.608]  Best share: 1.5M
--------------------------------------------------------------------------------
 0: GSF 10070017: BM1397:06+ 700.00MHz T:700 P:697 (3:2) | 94.2% WU: 92% | 2.615T / 2.585Th/s WU:36125.3/m

2.585Th/s

[cygan]

✂
Hey kano, I'm running 700MHz for around an hour now, without changes to the voltage screw and the NF-A8 PWM running at 2200RPM. Seems stable and good so far. I'm finally getting over 2.5TH/s (that was my goal), so I guess I beat you, with stock voltage!
✂

you definitely beat him
i will try my luck with my 2 r909 pod miner, only i would have to know how to adjust the rotations of the noctua fans ...

[n0nce]

you definitely beat him
i will try my luck with my 2 r909 pod miner, only i would have to know how to adjust the rotations of the noctua fans ...

Good luck!
The fan header on the R909 doesn't seem to output a fan curve so whatever you plug in, will run at its max speed. For the NF-A8 PWM, that should be 2200RPM.

You could plug in the included (with the fan) low-noise adapter that reduces it to a fixed 1750RPM but then the airflow will be less than the original fan so I'd be very careful with temps.

[kano]

Hey kano, I'm running 700MHz for around an hour now, without changes to the voltage screw and the NF-A8 PWM running at 2200RPM. Seems stable and good so far. I'm finally getting over 2.5TH/s (that was my goal), so I guess I beat you, with stock voltage!

Just to make sure: worst that can happen when aiming for too high clock speeds is that it crashes or returns faulty hashes, right? Or is there any 'risk' to pushing the overclock as long as the voltage screw is not touched? And the error rate is displayed through those Java API commands, right? The only way I've checked my error rate so far was the log printed when killing cgminer.

Proof:

Code:

 (5s):2.647T (1m):2.735T (5m):2.639T (15m):2.602T (avg):2.585Th/s
 A:3259855  R:0  HW:1984  WU:36123.2/m
 Connected to stratum.kano.is diff 1.62K with stratum as user
 Block: 4b258ede...  Diff:34.1T  Started: [07:05:45.608]  Best share: 1.5M
--------------------------------------------------------------------------------
 0: GSF 10070017: BM1397:06+ 700.00MHz T:700 P:697 (3:2) | 94.2% WU: 92% | 2.615T / 2.585Th/s WU:36125.3/m

2.585Th/s

Well that A says 1.5 hours so definitely higher hash rate than me

I wonder if 725MHz is just too high for them (as per my >20% HW) and 700 is the limit?

Also as mentioned in my case, one chips is somewhat of a dud,
so I had to keep pushing the freq higher to get closer to 2.5T,
so for you all 6 chips running OK at 700 will give a higher avg for you also "BM1397:06+ 700.00MHz"
(though my dud does about 25% of 500MHz when you set it to 500MHz)

'Error rate' is HW / (A+R+HW) = 1984 / (3259855 + 0 + 1984)

I changed cgminer to report HW on the screen correctly for the Gekkos - it used to only count HW occurrences, but alas most miners don't return nonces at diff1 - so for the Gekko's it uses the ticket per HW.
In the case of the R909 that's 64 - i.e. 1984 / 64 = 31 bad nonces returned (and each nonce returned is worth 64)

Yes you can kill it - no it's not bullet proof.
It is unlikely to kill it, but as you push it harder that means you are pushing the chips harder also.
If you overload the power it will just switch off, and then the 'fuse' will need a power cycle and letting it cool will fix that.
But the chips themselves, I've no idea under what circumstances they might be at risk.
Maybe sidehack knows?

[n0nce]

Well that A says 1.5 hours so definitely higher hash rate than me

I wonder if 725MHz is just too high for them (as per my >20% HW) and 700 is the limit?

That's definitely possible. I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now. Yet still I am getting 2.487Th/s on average. Still super close to that 2.5 mark, or 3.8% lower than my previous result, while reducing clock by 50MHz, so a 7.2% reduction in clock speed.
This indicates to me that above 650, we get into diminishing returns.

Yes you can kill it - no it's not bullet proof.
It is unlikely to kill it, but as you push it harder that means you are pushing the chips harder also.

I guess on its basic level, the transistors are going to be 'pushed' to switch faster; but in the realms of CPU overclocking I've never heard of a high frequency killing chips / diminishing their lifespan. It's usually the required higher voltage that does the most harm, to the best of my knowledge at least.

If you overload the power it will just switch off, and then the 'fuse' will need a power cycle and letting it cool will fix that.
But the chips themselves, I've no idea under what circumstances they might be at risk.
Maybe sidehack knows?

Feedback by sidehack would be amazing!

[kano]

I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now.

I initially got up to 2.497TH/s after more than an hour, hoping it would reach 2.5, but by the next day it was down to 2.475Th/s
I just let it go at 725/500 since, well, it didn't die, just the HW % kept rising.

[n0nce]

I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now.

I initially got up to 2.497TH/s after more than an hour, hoping it would reach 2.5, but by the next day it was down to 2.475Th/s

While also clocking itself down a bit? Or did you lock the frequencies?
I haven't done any such fancy stuff and it went down to 690 by itself, as mentioned before. So the sweet spot is probably between 650 and 700MHz depending on silicon lottery.
But I do think this behavior (automatic down clocking to a stable frequency) is a really great feature of cgminer. Much better than overly pushing the chip and just getting invalid blocks out of it.

I just let it go at 725/500 since, well, it didn't die, just the HW % kept rising.

For what it's worth, my HW% was always in check so far. It still downclocked. Oh well; again: diminishing returns, anyway.

[cygan]

I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now.

I initially got up to 2.497TH/s after more than an hour, hoping it would reach 2.5, but by the next day it was down to 2.475Th/s
I just let it go at 725/500 since, well, it didn't die, just the HW % kept rising.

i have now replaced the two fans with the nactua nf-a8 pwm and started the two r909 pod miners at 690mhz - before that i turned the blue screw to about 1 o'clock
now the two r909 run just over 2.5 hours and this is how the performance looks:

Code:

4: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:646 (3:2)   | 90.4% WU: 89% | 2.222T / 2.435Th/s WU:34028.6/m
5: GSF 10070003: BM1397:06+ 650.00MHz T:650 P:643 (3:2)   | 83.7% WU: 84% | 2.141T / 2.196Th/s WU:30681.2/m

somehow the 2.5th/s does not want to fall...

[n0nce]

I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now.

I initially got up to 2.497TH/s after more than an hour, hoping it would reach 2.5, but by the next day it was down to 2.475Th/s
I just let it go at 725/500 since, well, it didn't die, just the HW % kept rising.

i have now replaced the two fans with the nactua nf-a8 pwm and started the two r909 pod miners at 690mhz - before that i turned the blue screw to about 1 o'clock
now the two r909 run just over 2.5 hours and this is how the performance looks:

Code:

4: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:646 (3:2)   | 90.4% WU: 89% | 2.222T / 2.435Th/s WU:34028.6/m
5: GSF 10070003: BM1397:06+ 650.00MHz T:650 P:643 (3:2)   | 83.7% WU: 84% | 2.141T / 2.196Th/s WU:30681.2/m

somehow the 2.5th/s does not want to fall...

Weird; they're not running 690, though. Maybe decrease the voltage a little. Could be that they get too hot, return faulty computations and cgminer tells them to clock down. If they ran at stable 690.00MHz, they should right about reach 2.5TH/s.
Of course, there's some silicon lottery involved, but give that a shot. In my experience with Compac F, chips clocking down was usually a sign of bad cooling / too high voltage.

Can you show the line with A, R and HW numbers?

[cygan]

@n0nce
so you mean i should turn the blue screw again around 12 o'clock or reduce the mhz rate?🤔

[n0nce]

@n0nce
so you mean i should turn the blue screw again around 12 o'clock or reduce the mhz rate?🤔

Maybe first turn back the screw (I guess try something between 12 and 1) and try 690MHz. If that doesn't hold stable, try 680MHz, going down in 10MHz steps. In my experience, going above 650MHz was effortless at stock voltage.

[kano]

As mentioned on the gekko page,
https://kano.is/gekko.php#rang
the '[ChipNRanges]' data will give you a good idea what each chip is doing.

e.g. for my one before (data abbreviated to save space)

Code:

   [FreqLocked] => true
   [Chip0Ranges] => 882/998/1010/1035/1021/971/5917/94.54%
   [Chip0FreqReply] => 725.000000
   [Chip1Ranges] => 141/153/170/155/161/148/928/21.50%
   [Chip1FreqReply] => 500.000000
   [Chip2Ranges] => 1011/1136/1136/1082/1115/1098/6578/105.10%
   [Chip2FreqReply] => 725.000000
   [Chip3Ranges] => 834/1030/1003/1049/927/996/5839/93.29%
   [Chip3FreqReply] => 725.000000
   [Chip4Ranges] => 1022/1127/1162/1120/1164/1083/6678/106.70%
   [Chip4FreqReply] => 725.000000
   [Chip5Ranges] => 891/984/1071/1046/1028/1051/6071/97.00%
   [Chip5FreqReply] => 725.000000

Edit:
Oh I should show Dups and Nonces since that also shows if CPU/USB can't handle it:

Code:

   [Elapsed] => 31047
   [Nonces] => 1560325
   [Dups] => 7383
   [DupsReset] => 7383

   [Chip0Nonces] => 246973
   [Chip0Dups] => 8
   [Chip1Nonces] => 42874
   [Chip1Dups] => 267
   [Chip2Nonces] => 272900
   [Chip2Dups] => 57
   [Chip3Nonces] => 242216
   [Chip3Dups] => 11
   [Chip4Nonces] => 272852
   [Chip4Dups] => 6113
   [Chip5Nonces] => 243088
   [Chip5Dups] => 927

The first 3 are the most relevant numbers.

Chip4Dups might be suggesting that chip 4 can't handle 725MHz as well as the other 4 good ones, or is near it's limit - though I'm not sure

[cygan]

@n0nce
so you mean i should turn the blue screw again around 12 o'clock or reduce the mhz rate?🤔

Maybe first turn back the screw (I guess try something between 12 and 1) and try 690MHz. If that doesn't hold stable, try 680MHz, going down in 10MHz steps. In my experience, going above 650MHz was effortless at stock voltage.

now everything runs just over 24h but I still can not reach the limit of 2.

Code:

4: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:655 (3:2)   | 88.4% WU: 90% | 2.792T / 2.453Th/s WU:34281.3/m
 5: GSF 10070003: BM1397:06+ 640.00MHz T:640 P:621 (3:2)   | 89.8% WU: 90% | 1.961T / 2.331Th/s WU:32575.2/m

and here are the a, r and hw data (but there are 4 more compac f sticks connected)

Code:

A:131834880  R:54432  HW:51456  WU:85740.4/m

[n0nce]

@n0nce
so you mean i should turn the blue screw again around 12 o'clock or reduce the mhz rate?

Maybe first turn back the screw (I guess try something between 12 and 1) and try 690MHz. If that doesn't hold stable, try 680MHz, going down in 10MHz steps. In my experience, going above 650MHz was effortless at stock voltage.

now everything runs just over 24h but I still can not reach the limit of 2.

Code:

4: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:655 (3:2)   | 88.4% WU: 90% | 2.792T / 2.453Th/s WU:34281.3/m
 5: GSF 10070003: BM1397:06+ 640.00MHz T:640 P:621 (3:2)   | 89.8% WU: 90% | 1.961T / 2.331Th/s WU:32575.2/m

and here are the a, r and hw data (but there are 4 more compac f sticks connected)

Code:

A:131834880  R:54432  HW:51456  WU:85740.4/m

You've got an error rate of 0.0008025419406, in other words 0.08% which is excellent.
If I were you, I'd slightly increase clock speeds. I'm not sure if I missed something, so are you deliberately running one R909 at 640MHz and one at 680?

I wonder if 725MHz is just too high for them (as per my >20% HW) and 700 is the limit?

That's definitely possible. I'm running 650MHz now, since after a few hours it had dropped to 690MHz and I don't feel like pushing it to its max right now. Yet still I am getting 2.487Th/s on average. Still super close to that 2.5 mark, or 3.8% lower than my previous result, while reducing clock by 50MHz, so a 7.2% reduction in clock speed.
This indicates to me that above 650, we get into diminishing returns.

Update on this: It does seem like the sweet spot is right between 650MHz and 700MHz. I've turned mine back up a bit from 650MHz (just slightly over 2TH/s after a day) to 675MHz and I'm back on 2.531Th/s after an hour. Let's see what this looks like in a few hours.

[cygan]

✂
If I were you, I'd slightly increase clock speeds. I'm not sure if I missed something, so are you deliberately running one R909 at 640MHz and one at 680?
✂

no, both were started with 680 only the one went down step by step to 640
maybe i should turn the screw a little on the one r909 (640) after all?

[kano]

Got my R909 delivered today and got all my miners going except the R808 I got off of ebay appears to have pooped out. It was working then got stuck on Found 0 Chips..
...

Check it's not a software version issue - i.e. if it was working before, use the version of software from before just for the R808 and see if it's ok.
I've no idea if the R808 works with the current code.
I've run R909, R606, CompacF, NewPac, 2Pac and Compac on the new version (and you should use the new 4.12.1 version for those six)

As I mentioned in Discord, this turns out to be that the old magic '60' calc is no good for the R808.
Side effect was an average of about once per half an hour of it deciding to reset due to a slow nonce.

I've updated git with a fix for the R808 (and I've run it on all the miners now together) and the new calc is more reasonable (and more accurate)
So only if you have an old R808 you should do the update steps in https://kano.is/gekko.php#lin and it should fix that.

Edit; of course it wont fix a dead R808 - but it will fix it if it was resetting repeatedly, unnecessarily.

[n0nce]

✂
If I were you, I'd slightly increase clock speeds. I'm not sure if I missed something, so are you deliberately running one R909 at 640MHz and one at 680?
✂

no, both were started with 680 only the one went down step by step to 640
maybe i should turn the screw a little on the one r909 (640) after all?

What I noticed is that sometimes the automatic 'step down' is a bit too aggressive; i.e. if you set the target frequency too high (like 680MHz maybe), it might step down to 640MHz, even though it would hold 650MHz just fine if you used that as a target.
So before fiddling with voltages, I'd try to (manually) find a stable frequency target by increasing / decreasing it in 50MHz or 25MHz steps and letting it run for a few hours.

Right now, after 12h on 675MHz, I have a 2.310Th/s average. That's substantially more than the ~2TH at 650MHz. As you can tell, I'm having a blast fine-tuning this thing. And I'm just getting started (need to install Java & such..).

[kano]

Well, as long as you have the log going somewhere, it will say why it stepped the freq down each time.

Also, if anyone does change the step settings on the BM1397 miners, I'd suggest not to.

[100knot2dae]

Got my unit this week, really a great piece of hardware! So far I am happy with it, have it running for now at 2.1Th/s with 550MHz for about 24 hours, with the flat side of the voltage knob pointing to between 9 and 10 (clock).

Speaking of the voltage knob, is there a standardized stock voltage position that applies to all units? Reason I am asking is because mine was shipped with the knob at 11, while I saw a picture of another user with the stock position at about 9 to 10: https://imgur.com/a/F1V84qf
Which is what I have set mine to as well now, saving me about 5 Watts while still running with the same hashrate.

[sidehack]

They're factory calibrated to 4.6V which isn't always the same position on the knob due to part tolerances in the volt-setting circuit. If it'll run at speed from a lower voltage, there's no reason not to.

[RKG456]

This evening I rearranged everything and now have all my Gekkos mining off of a single Raspberry Pi 4 (4GB) host.  Looks stable so far but has only been back up for a few hours.   I let it run for a week and a half with no issues when it was spread across two hosts.  The day before yesterday my peak hash rate spiked up to 8.025TH/s for a minute or so.  Getting many spikes over 7.5TH/s

Code:

cgminer version 4.12.1 - Started: [2023-01-19 17:07:59.335]
--------------------------------------------------------------------------------
 (5s):6.593T (1m):6.293T (5m):6.322T (15m):6.277T (avg):6.097Th/s
 A:15138816  R:0  HW:15828  WU:85182.8/m
 Connected to bitcoin.viabtc.io diff 8.19K with stratum as user N0F34R.NPandT
 Block: 533c0002...  Diff:37.6T  Started: [19:50:37.341]  Best share: 99.7M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070032: BM1397:06+ 600.00MHz T:600 P:600 (3:2)     | 96.8% WU: 97% | 2.541T / 2.350Th/s WU:32840.4/m
 1: GSF 10070012: BM1397:06+ 600.00MHz T:600 P:600 (3:2)     | 96.8% WU: 98% | 2.616T / 2.381Th/s WU:33272.4/m
 2: GSH 10041663: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  |  100% WU: 98% | 89.85G / 83.92Gh/s WU: 1172.4/m
 3: GSH 10038894: BM1387:02+ 375.00MHz T:375 P:375 (100:50)  | 98.5% WU:^98% | 85.93G / 83.84Gh/s WU: 1171.4/m
 4: GSF 10052154: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 309.1G / 274.5Gh/s WU: 3834.7/m
 5: GSF 10053864: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 393.6G / 297.0Gh/s WU: 4148.7/m
 6: GSF 10050104: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 293.9G / 297.1Gh/s WU: 4150.1/m
 7: GSF 10054544: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 344.7G / 300.3Gh/s WU: 4195.1/m
 8: GSF 10053931: BM1397:01+ 450.00MHz T:450 P:450 (28:14)   |  100% WU:100% | 254.7G / 274.7Gh/s WU: 3837.1/m

[sidehack]

2.3+T  on the pods, nice.

[MoparMiningLLC]

if anyone is selling one for btc please let me know!

[100knot2dae]

They're factory calibrated to 4.6V which isn't always the same position on the knob due to part tolerances in the volt-setting circuit. If it'll run at speed from a lower voltage, there's no reason not to.

Looks like my unit is approaching an efficiency sweet spot for this frequency (550 MHz). Since the voltage decrease (12 hours ago), it's not only using 5W less but also the avg hashrate has slightly increased (+10 Gh/s  ), while the WU/m has climbed up by 110 and the HW error ratio being stable.

Very cool, it's a really impressive device what you have crafted here @sidehack!

When turning the voltage knob, is it safe to do so when the unit is powered, or even while cgminer is running? Does it matter, if so what's the recommended way?

[kano]

https://kano.is/gekko.php#vol

[n0nce]

Got my unit this week, really a great piece of hardware! So far I am happy with it, have it running for now at 2.1Th/s with 550MHz for about 24 hours, with the flat side of the voltage knob pointing to between 9 and 10 (clock).

Interesting; mine dropped a bit down to 2.1Th/s over the last 6 days. I am still on 675MHz. It started at almost 2.5Th/s. Did anyone else experience such a little bit of drop-off after the first few hours of operation?
My only theory right now is that they get slower as the heat up and the heatsink becomes fully soaked. Not sure about that, though.

[itskeule]

Interesting; mine dropped a bit down to 2.1Th/s over the last 6 days. I am still on 675MHz. It started at almost 2.5Th/s. Did anyone else experience such a little bit of drop-off after the first few hours of operation?
My only theory right now is that they get slower as the heat up and the heatsink becomes fully soaked. Not sure about that, though.

I had the same experience. With stock voltage, I tried various freq. settings from 500 MHz up to 725 MHz. At the beginning of the run, cgminer reports up to ~2.5 TH, then, after 30min or 1 hour it slowly drops the hash rate.
Currently, at 725 MHz (which seems the maximum, increasing it, the HW errors raising as well to an unacceptable ratio), the miner is dropping down to ~2.190 TH.

Also tried to increase the voltage knob from 12 o'clock to 3 o'clock in small steps, to no avail. So I turned it back to the factory setting.
The next step will be to measure temps and increase airflow if necessary.

[100knot2dae]

My only theory right now is that they get slower as the heat up and the heatsink becomes fully soaked. Not sure about that, though.

I had the same thought when I saw the hashrate increase after the slight undervolt at 550MHz. The avg hashrate is at 2.12Th/s by now, I think I will just leave it running like this for a few days to see if it goes up even further yet.

[gergar]

Hello all, i want to share my experience and some advise on cooling as i found out points to improve:

I got 4units few weeks ago, all 4 came with bolts off in the case, initially i wasnt clear where they came from i thought they were “spare bolts” 😂, but as i set them and ran for couple of hours at stock frequency 400 i saw the hashrate going down to 1.4T and they went down to zombie no chip… that made me think about temperature issues…

I open each of them, net it is not easy to open but that can be an improvement in future … i found out:
1 both heatsinks were moving missing the bolts i had at the boxes.
2 the fans are pushing the air to the inside vs pulling it out
3 there is no thermal compound on the backplate heatsinks

what i did:
1 put thermal path covering the backplate heatsink (20w/k), used same compound for the top at the BM chips.
2 when bolting it i made a cross then center then cross then center until the bolts were full secured, 6 bolt array needs a sequence to avoid the issue of having some loosen bolts due to the center bolts will cause the outer ones to lose pressure as the heatsink gets tightened.
3 inverted the fan to pull the air, why, the current setu makes hot air stays inside the case (measured with a temp camera)

Results:
550 stable - 2.2T each (48hrs no TH drops) very cold running- i kept them back there for longevity 85w per R909.
600 stable - 2.4T each (48hrs no TH drops) temperature got slightly high, but still looks fine.
650 stable - 2.6T in some / 2.5 in others (48hrs sone git back to 600Mhz) i guess silicon lottery and or needs better heatsink like cpu server with very small fins but many to increase thermal transfer and better fans <getting some 6k high cfm>

As i saw risk of damaging the asics i stop there (i had compact F damaged due to temps, 6 and replaced the asics with new AI but after monitoring i found that thermal will kill the chips so fyi)

I will do the same 750 test and report back just with one and try to measure temps with my temp camera.

hope this helps to improve the thermal protection so far with stock heatsinks these toys performs amazing just make the backplate to really work adding thermal compound on it and you will really use the backplate to remove heat.

[kano]

Sounds like you need to track down whoever you got them from and get them to explain why they were disassembled.

They are tested mining, properly assembled, before even being sent to the distributors.

[sidehack]

We've recently gotten a few reports of screws rattling loose during shipping and anything that's gone out since then has blue threadlocker on the screws. There's already a version change in planning to improve heatsink screws and ease of getting the case on and off.

It's not advisable to use thermal compound on the top chips or we would have done so. The IC package is plated and electrically connected to local ground, which means that unless there's electrical insulation between the package and heatsink, you can cause shorts and break stuff. We use a 0.5mm 20W/k silicone pad to avoid shorts.

I'll do some testing with adding TIM to the underside heatsink and see how that changes heat management.

[gergar]

Thank you yes i meant thermal pads (20w/k) both sides sorry 😞 . I used .5mm too and 1mm bottom to have symmetrical balance between heatsinks too.

yes maybe springs or locktite on them it was strange but the product is quite great maybe testing only with the outer 4 bolts…

[Sledge0001]

Alright finally just got around to this and ordered 2 of em from BitcoinMerch! 

Looking forward to tinkering with these next week!

[kano]

Added info from API-README about how to setup a web page to view your gekkos
https://kano.is/gekko.php#mph
Has a Gekko specific page and a GekkoChips page that shows the chips

Edit: Update, I added an extra command to the above steps that I left out before,
since you need to get the latest miner.php with the Gekko updates.

[100knot2dae]

Hello all, i want to share my experience and some advise on cooling as i found out points to improve:
...
3 inverted the fan to pull the air, why, the current setu makes hot air stays inside the case (measured with a temp camera)
...

Hello, what was the temperature difference you have measured with your temp camera when the (stock?) fan was blowing out? Would be interesting to know.

[n0nce]

My only theory right now is that they get slower as the heat up and the heatsink becomes fully soaked. Not sure about that, though.

I had the same thought when I saw the hashrate increase after the slight undervolt at 550MHz. The avg hashrate is at 2.12Th/s by now, I think I will just leave it running like this for a few days to see if it goes up even further yet.

That's impressively high for just 550MHz. After a few hours of running 700MHz, mine seems to have settled on around 2.3Th/s, after starting at ~2.5Th/s... Still not sure if this is temp related.

[sidehack]

It's 96% of expected speed, so not that impressive. Yours is running a little over 80% expected. I bet you could see the same 2.3TH effective from about 600.

[cygan]

My only theory right now is that they get slower as the heat up and the heatsink becomes fully soaked. Not sure about that, though.

I had the same thought when I saw the hashrate increase after the slight undervolt at 550MHz. The avg hashrate is at 2.12Th/s by now, I think I will just leave it running like this for a few days to see if it goes up even further yet.

That's impressively high for just 550MHz. After a few hours of running 700MHz, mine seems to have settled on around 2.3Th/s, after starting at ~2.5Th/s... Still not sure if this is temp related.

my 2 r909 miners have been running for ten days now and the performance at 680 and 640mhz is impressive
both pod miners have settled between 2.34 and 2.45 th/s

Code:

4: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:662 (3:2)    | 90.1% WU: 90% | 2.838T / 2.457Th/s WU:34332.8/m
5: GSF 10070003: BM1397:06+ 640.00MHz T:640 P:630 (3:2)    | 91.0% WU: 91% | 2.128T / 2.343Th/s WU:32744.4/m

[gergar]

Hello all, i want to share my experience and some advise on cooling as i found out points to improve:
...
3 inverted the fan to pull the air, why, the current setu makes hot air stays inside the case (measured with a temp camera)
...

Hello, what was the temperature difference you have measured with your temp camera when the (stock?) fan was blowing out? Would be interesting to know.

Hello,

to make easier some definitions first, 550Mhz as i tested with my gekko compactf settings temps in celcius

FP= Top Heatsink (front plate)
PCB
BP = bottom Heatsink (back plate)
EX = exhaust (stock = front / modified = back)

Stock
FP = 43 / with all bolts (50)
PCB = 54 / with all bokts (47)
BP = 27 / with all bolts (30)
EX = 48 / with all bolts (50)

Modified
FP = 53
PCB = 35
BP = 45 (temp at front = 21, two degrees lower then ambient 23)
EX = 44

what it means?, the heat was mostly staying at the PCB as the heatsinks were not fully operational, once i adjusted bolts temp lowered some how but still the lower heatsink was mot doing clean contact with the pcb (air in between. But the air as it was not strong enough to move out stayed there in the front near the cables. with the changes the backplate is draining more heat from the pcb, pcb is colder and top maintained temps so contact issue was the initial problem. Then all front part is about ambient temperature, as air gets higher speed it is 2 to 3 degrees lower just cos pressure changes, and the air going out of the fan and case is hot.

I am running it at 650 now and here the temps i just captured as i was intrigued on how much temp gets from 100Mhz more:
FP= 66 (some are hotter 69-72)
Pcb = 36-37 relative cold
BP= 50 (seems it still gets heat out)
EX = 49 - 52

I assume with current stock fan might get too hot to go higher on freq, not sure the asic temperatures down there but they should be extremely hot. 650 had maintained 24 hrs at 2.6T stable (21 degree temp intake, maybe cooler temps can allow it to go higher on freq. - mine anything higher than 650 starts to bounce back to 650-646 and i get lower hashrates.

[kano]

Actually, what matters most is the heat of the BM1397 chips.
If that heat is going away, to other components, then that's good.

One way to measure this is the W usage, since it will go up, at the same frequency, if the chip is heating up.

[gergar]

Actually, what matters most is the heat of the BM1397 chips.
If that heat is going away, to other components, then that's good.

One way to measure this is the W usage, since it will go up, at the same frequency, if the chip is heating up.

Yes and i have done that to calculate the expected BTUs etc from the asics, what i found was about a linear growth, 50w per 100Mhz of freq increase and about 10+ degree increase. i am preparing some thermistors to check temps at bottom on art of the heatsink that i assume might be closer to the asics if there is good thermal transfer.

have another question on the Api, i sent via python a post message with command=ascset parameter={0,freq,400} but it returns “Unknown Option: freq” is there any documentation. on the payload to be sent using json? i used the page that explains in summary the api but as it uses | i am not quite clear how will it read using json command/parameter format.

[n0nce]

@gergar What do you think about using 2 fans (1x intake, 1x exhaust)? I mean, the downsides of having ports on the back now is that it's not really possible, but maybe a small fan that covers part of the back side could help.
Or using 2 fans in the front. That did wonders on my Apollo.

[gergar]

@gergar What do you think about using 2 fans (1x intake, 1x exhaust)? I mean, the downsides of having ports on the back now is that it's not really possible, but maybe a small fan that covers part of the back side could help.
Or using 2 fans in the front. That did wonders on my Apollo.

it will help but in reality the way cables are today might need very small fans and to have a proper push-pull setup fans must be similar in cfm and pressure capacities (i tried some noctua 40x25mm and are large to fit) so i ordered som 6k 80mm fans from amazon one 25mm and other that is 38mm but cfm is 3x the current but noise will the the issue with these but i want to test how will it help.

this is the 38mm
https://www.amazon.com/Wathai-Cooling-80mm-Speed-Bearing/dp/B07QX3QCP5
this is the 25mm that fits with mo issues but you need in both an adaptor (link also included)
https://www.amazon.com/GDSTIME-80mm-25mm-Brushless-Cooling/dp/B00N1Y50QQ
the adaptor
https://www.amazon.com/Mainboard-Standard-Extension-Power-Cable/dp/B07Z8Z182S

once i get them netx week let me test it and i can report back so you don’t throw away money … i can try to set a push-pull using the original fans i just need to get some very large bolts/nuts to use the top bolts from the current fan to secure the front ones just i need to do some measuring to avoid bending too much the power and usb cable as i am using an atx 1600w pc psu and the pcie cable doesn’t bend too much.

hope it makes sense to you.

[sidehack]

Does push-pull really matter in a system without much flow constriction? My understanding was that helped maintain airflow in an environment with lots of obstruction, like close-packed heatsink fins, but the R909 is pretty open.

[100knot2dae]

I also replaced the stock fan with the Noctua NF-A8 today, and by that time I checked the heatsink screws as well. Almost all have been at least slightly loose and needed some re-tightening.

As a result of both, the unit is now hashing with 2.15TH/s with the undervolt at 550MHz, and the increased airflow helps to save a few more Watts. 

[n0nce]

Does push-pull really matter in a system without much flow constriction? My understanding was that helped maintain airflow in an environment with lots of obstruction, like close-packed heatsink fins, but the R909 is pretty open.

I'm not sure about that myself.
Earlier today I tried 550MHz with the NF-A8 installed and the stock fan placed in front of it, it seemed to make no difference. I remember someone trying this in a different application and they had the same results. Somehow, it was effective on the Futurebit Apollo, though. However that device was exhausting up top, with the fan pulling air through a pretty dense heatsink. So by the time it reached the top, the airspeed was quite slow. Pretty different scenario.

After those experiments, I placed the stock fan in the back, covering roughly half of the back area and set the miner to 700MHz. (I know; not very scientific. I should stick to 550 now, for further experiments).
Anyhow, I'm now seeing 2.6Th/s, something I have never experienced with this miner before.

Code:

 0: GSF 10070017: BM1397:06+ 700.00MHz T:700 P:694 (3:2) | 91.6% WU: 93% | 2.507T / 2.613Th/s WU:36505.2/m

Let's see if it can keep this up. So far, that's a 4:30h average. I'm also interested to see if the hashrate drops off over the next 24h or if this double fan setup keeps it above 2.5.

[gergar]

Does push-pull really matter in a system without much flow constriction? My understanding was that helped maintain airflow in an environment with lots of obstruction, like close-packed heatsink fins, but the R909 is pretty open.

So far with the size and fan type I don't expect any benefit but swapping to exhaust the air vs push inside did a difference (also adding the thermal path to the backplate)... with High CFM/Pressure might help, BUT heatsink design on 909 is not for that type of work, so I doubt.

I will test it so we can see if works but what I decided based on the construction of 909 was to set a stronger pull fan as I see that will be enough... at the end these are 150W tops devices and to make it push-pull cable needs to be routed differently to make the fan on top no "gaps" and create a tunnel at the 909 chassis. Also power consumption won't be great if we add fans and 6k fans pulls a lot of power at full speed. To me is a balance on the lower power possible and the highest hashrate possible, which I just haven't found it yet...

I don't see it worth to change design for push-pull, I would better look for a pipe/type dissipator/heatsink with very close fins like the ones used at server CPUs, that will maximize the airflow that passes. I am trying to find in the market any CPU like heatsink that I can try on it before I convert them to water cool as I have the compactF (just learning if the current air-cooling can get improved, but my end point is water-cooling it as summer is very hot here and air density is quite poor).

But well it is all about learning and experimentation so couple of usd worth it (at least to me).

I am using some alpha cool memory water cooling same as for the Cf to cool them down, I am working on the holders to leverage the 6 bolts and the backplate which is a great add BTW!  so doing some design/testing to assure PCB does not bend. I also ordered some SSD and custom memory watercolors to see if they fit better than the small Alphacool NDXs. 

I will test before I move them to water and share if it improves... for now 650Mhz @ 2.6T for 48hr and heat is well managed by the current design just with the few mods I did. Since the new 1397AI chips I ordered  have not arrived yet for repairs I will stay away from 650+ speeds as I need these for testing.

you both You and Kano had made an amazing product !!! thank you!!

[gergar]

Does push-pull really matter in a system without much flow constriction? My understanding was that helped maintain airflow in an environment with lots of obstruction, like close-packed heatsink fins, but the R909 is pretty open.

I'm not sure about that myself.
Earlier today I tried 550MHz with the NF-A8 installed and the stock fan placed in front of it, it seemed to make no difference. I remember someone trying this in a different application and they had the same results. Somehow, it was effective on the Futurebit Apollo, though. However that device was exhausting up top, with the fan pulling air through a pretty dense heatsink. So by the time it reached the top, the airspeed was quite slow. Pretty different scenario.

After those experiments, I placed the stock fan in the back, covering roughly half of the back area and set the miner to 700MHz. (I know; not very scientific. I should stick to 550 now, for further experiments).
Anyhow, I'm now seeing 2.6Th/s, something I have never experienced with this miner before.

Code:

 0: GSF 10070017: BM1397:06+ 700.00MHz T:700 P:694 (3:2) | 91.6% WU: 93% | 2.507T / 2.613Th/s WU:36505.2/m

Let's see if it can keep this up. So far, that's a 4:30h average. I'm also interested to see if the hashrate drops off over the next 24h or if this double fan setup keeps it above 2.5.

What I think can help is make the fan to pull / extract the air use the original as it is stronger than Noctua (Arctic label pointing the PCB will make it to extract --- I checked their specs as I have Noctuas too) and add thermal pads to the bottom heatsinks 1mm @ 13+w/k will have the highest impact in it, those few changes will give you 650@2.5-2.6T stable on a ~120W on stock voltage.

[n0nce]

I will test before I move them to water and share if it improves... for now 650Mhz @ 2.6T for 48hr and heat is well managed by the current design just with the few mods I did. Since the new 1397AI chips I ordered  have not arrived yet for repairs I will stay away from 650+ speeds as I need these for testing.

That sounds great! I will try rotating the fan, too then.

For what it's worth, my experiment with a loosely positioned 'exhaust fan' did not last long; 8h in, the hashrate dropped to 2.4Th/s again. I'm not sure why, but my chips perform well, for a few hours and then slowly drop further and further. It must be thermally related.

add thermal pads to the bottom heatsinks 1mm @ 13+w/k will have the highest impact in it, those few changes will give you 650@2.5-2.6T stable on a ~120W on stock voltage.

Just to be clear: the bottom heatsinks just cool the PCB? And by default are bolted on without any thermal interface (pads or anything else)?
I'd like to prevent touching the top heatsinks, since I believe there's a chance to rip a thermal pad if you don't screw the heatsink on perfectly straight, which would then connect grounds and lead to magic smoke.
But bottom heatsinks are a mod I may actually do. I even have 12W/mK 1mm pads here.

[tcj2001]

I have a compac-f getting 348.5 Gh/s and R909 getting 2.457 Th/s
What suggest-diff i should use for these individually.

I saw some calulation suggesting difficulty of 1 per 1 Gh/s / 1.8 and another one suggested  Gh/s * (71 * .05)^2

Any suggestion?

[os2sam]

I have a compac-f getting 348.5 Gh/s and R909 getting 2.457 Th/s
What suggest-diff i should use for these individually.

I saw some calulation suggesting difficulty of 1 per 1 Gh/s / 1.8 and another one suggested  Gh/s * (71 * .05)^2

Most pools adjust diff to 18 shares per minute regardless of what your suggest diff is.  It is just a suggestion after all.

[tcj2001]

So then I think its okay to set suggest-diff of 500 for compac-f and 5000 for r909, as pool will takes care of it and adjust it it accordingly. BTW i am using ckpool

[os2sam]

So then I think its okay to set suggest-diff of 500 for compac-f and 5000 for r909, as pool will takes care of it and adjust it it accordingly. BTW i am using ckpool

It will start at your suggest diff and then adjust to 18 shares per minute.  It doesn't really matter what you set it to, pools will adjust it automatically.

Pools also now have a minimum diff so that ASIC's won't flood them with diff1 shares.

[tcj2001]

So then I think its okay to set suggest-diff of 500 for compac-f and 5000 for r909, as pool will takes care of it and adjust it it accordingly. BTW i am using ckpool

It will start at your suggest diff and then adjust to 18 shares per minute.  It doesn't really matter what you set it to, pools will adjust it automatically.

Pools also now have a minimum diff so that ASIC's won't flood them with diff1 shares.

Pardon my ignorance, I have attached the screen shots of cgminer for both of these units
Compac-f https://imgur.com/Da9dfnc
R909 https://imgur.com/xPDxaDl

I believe the WU tells the diff1 and these are very high compared to 18.

[gergar]

I will test before I move them to water and share if it improves... for now 650Mhz @ 2.6T for 48hr and heat is well managed by the current design just with the few mods I did. Since the new 1397AI chips I ordered  have not arrived yet for repairs I will stay away from 650+ speeds as I need these for testing.

That sounds great! I will try rotating the fan, too then.

For what it's worth, my experiment with a loosely positioned 'exhaust fan' did not last long; 8h in, the hashrate dropped to 2.4Th/s again. I'm not sure why, but my chips perform well, for a few hours and then slowly drop further and further. It must be thermally related.

add thermal pads to the bottom heatsinks 1mm @ 13+w/k will have the highest impact in it, those few changes will give you 650@2.5-2.6T stable on a ~120W on stock voltage.

Just to be clear: the bottom heatsinks just cool the PCB? And by default are bolted on without any thermal interface (pads or anything else)?
I'd like to prevent touching the top heatsinks, since I believe there's a chance to rip a thermal pad if you don't screw the heatsink on perfectly straight, which would then connect grounds and lead to magic smoke.
But bottom heatsinks are a mod I may actually do. I even have 12W/mK 1mm pads here.

the pcb is taking the back heat from the asics, they come with no thermal interface on them, only air therefore the heat stays and gets into the pcb/components including asics.

what i usually do is change both up and lower but upper vs what you have are better 20w/k once you lose the bolts the bottom will fall, do 1 at time just be careful of not move them much.

overall that will help the Asics to get cooler same PCB as also power controllers getting hot will lose capacity and cause performance issues, so keeping all as cool as possible pays out in higher room to push hard the asics and stability of other components.

hope makes sense to you.

[100knot2dae]

What I think can help is make the fan to pull / extract the air use the original as it is stronger than Noctua (Arctic label pointing the PCB will make it to extract --- I checked their specs as I have Noctuas too) and add thermal pads to the bottom heatsinks 1mm @ 13+w/k will have the highest impact in it, those few changes will give you 650@2.5-2.6T stable on a ~120W on stock voltage.

Hm, why do you think the stock fan is stronger? When comparing the specs between the Noctua NF-A8 and the Arctic F8, I see the Noctua better in every discipline. Slightly higher RPM, better airflow and MUCH better static pressure. The latter of which makes it an even better candidate for blowing air into the unit (the original direction).

[os2sam]

Pardon my ignorance, I have attached the screen shots of cgminer for both of these units
Compac-f https://imgur.com/Da9dfnc


I believe the WU tells the diff1 and these are very high compared to 18.

Are you expecting the suggest-diff command to have an effect on the Worker Utility?

In your screen shot it looks like your suggest-diff of 500 is being honored.  So that equates to 9.94 shares per minute being submitted to the pool.

[tcj2001]

Pardon my ignorance, I have attached the screen shots of cgminer for both of these units
Compac-f https://imgur.com/Da9dfnc


I believe the WU tells the diff1 and these are very high compared to 18.

Are you expecting the suggest-diff command to have an effect on the Worker Utility?

In your screen shot it looks like your suggest-diff of 500 is being honored.  So that equates to 9.94 shares per minute being submitted to the pool.

What i observed is if I set difficulty too high then no of accepted shares per minute is less compared to the low difficulty setting. So I guess higher the accepted shares per minute mean more chances of calculating/generating best share.
So now I have set difficulty to 1 on both units so that the pool adjust the difficulty to whatever is feasible for the pool, that way I maximize my chances of calculation done per minute by each units.
I may be totaly wrong in my assumption.

[os2sam]

What i observed is if I set difficulty too high then no of accepted shares per minute is less compared to the low difficulty setting. So I guess higher the accepted shares per minute mean more chances of calculating/generating best share.

Setting your share difficulty has no bearing on your chances of finding a block or generating a better best share.

It's simply a filter used to keep from swamping a pool with diff 1 shares.  Using a higher share difficulty only filters out low diff shares to decrease the amount of pool resources needed to evaluate your shares and decreases the network bandwidth required to submit shares.

In your example of suggest-diff of 500, shares below a diff of 500 are not submitted and shares 500 and above are submitted.

Your share submission of 9.94 shares per minute just means your hashrate estimate by the pool will have more variance than the pools adjusting to 18 shares per minute.  If you remove the suggest-diff then ck solo will adjust for 18 share per minute which would be a difficulty of around 276.

[tcj2001]

What i observed is if I set difficulty too high then no of accepted shares per minute is less compared to the low difficulty setting. So I guess higher the accepted shares per minute mean more chances of calculating/generating best share.

Setting your share difficulty has no bearing on your chances of finding a block or generating a better best share.

It's simply a filter used to keep from swamping a pool with diff 1 shares.  Using a higher share difficulty only filters out low diff shares to decrease the amount of pool resources needed to evaluate your shares and decreases the network bandwidth required to submit shares.

In your example of suggest-diff of 500, shares below a diff of 500 are not submitted and shares 500 and above are submitted.

Your share submission of 9.94 shares per minute just means your hashrate estimate by the pool will have more variance than the pools adjusting to 18 shares per minute.  If you remove the suggest-diff then ck solo will adjust for 18 share per minute which would be a difficulty of around 276.

Thanks making sense now.
How did you calculate diff of 500 equate to 9.94 shares per minute?

[os2sam]

How did you calculate diff of 500 equate to 9.94 shares per minute?

WU:4970/Suggest-diff:500=9.94 shares per minute.

[tcj2001]

New question: Number of chip in miner vs total hash power.
In R909 there are 6 BM1397 chips, so does each chip individually participate in calculating the best share at the same time or the total hash power is used everytime to do the calculation.

[os2sam]

New question: Number of chip in miner vs total hash power.
In R909 there are 6 BM1397 chips, so does each chip individually participate in calculating the best share at the same time or the total hash power is used everytime to do the calculation.

Each Core in each chip is calculating shares.  The total hash rate increases the chances of finding a share that meets or exceeds current difficulty.

[gergar]

What I think can help is make the fan to pull / extract the air use the original as it is stronger than Noctua (Arctic label pointing the PCB will make it to extract --- I checked their specs as I have Noctuas too) and add thermal pads to the bottom heatsinks 1mm @ 13+w/k will have the highest impact in it, those few changes will give you 650@2.5-2.6T stable on a ~120W on stock voltage.

Hm, why do you think the stock fan is stronger? When comparing the specs between the Noctua NF-A8 and the Arctic F8, I see the Noctua better in every discipline. Slightly higher RPM, better airflow and MUCH better static pressure. The latter of which makes it an even better candidate for blowing air into the unit (the original direction).

My bad i checked vs the Noctua R8 (the ones i have with me), I didn’t saw you were using the A8, yes that is better to push but still it is weak, it will do better extracting thought but pushing these fans are quite weak.

having it to push and the original and/or other noctua pull will help, but i not sure how effective will be as it wont be in a close setup “tunnel” but i assume it will help to keep them cooler and accelerate the air causing it to cool a bit more.

[pkspks]

Am I the only one struggling to take these things apart? Any suggestions would be appreciated! I managed to unscrew everything except the screws for PCB standoff and I still can't slide or pop the cover off. Do I also need to unscrew the screws holding the PCB standoff? These screws seem somewhat tighter and didn't want to risk stripping it.

[kano]

Two things:

1) The CompacF mines at a minimum of 16 diff - I've coded it to not go lower.
The R909 miners at a minimum of 64 diff - I've coded to not go lower.
These values are already below what is needed.
People should expect pools to be pissed off at them if they tried to DDoS the pool with lower values.

2) All about worker difficulty: https://kano.is/index.php?k=workdiff

[sidehack]

Am I the only one struggling to take these things apart? Any suggestions would be appreciated! I managed to unscrew everything except the screws for PCB standoff and I still can't slide or pop the cover off. Do I also need to unscrew the screws holding the PCB standoff? These screws seem somewhat tighter and didn't want to risk stripping it.

You would have to flex the case outward at the bottom a little bit to get around the board. If you unscrew the PCB standoffs from the base and slide the whole innards out the front, the case would come free without issue. There's a planned board revision for the next batch that addresses heatsink screws and also ease of [dis]assembly.

[100knot2dae]

@sidehack is there really no thermal pad between the "backplate" heatsink and the PCB board (as mentioned here: https://bitcointalk.org/index.php?topic=5423227.msg61639015#msg61639015)? Was that by intention? If not, what would be the recommended choice from your side in regards to thickness and thermal conductivity of such thermal pad? And would that give some significant cooling benefit at all?

If that makes sense, I am tempted to do that. Unfortunately, I didn't check the heatsink dimensions back then when I replaced the fan. Can you please share these, so that I don't need to unscrew my unit again.

[sidehack]

Heatsink is about 45x70mm

Yes there's no interface between the bottom heatsink and the board. I'm gonna be doing some testing this week on a few potential changes to improve cooling, and underside paste is one of them.

[100knot2dae]

Heatsink is about 45x70mm

Yes there's no interface between the bottom heatsink and the board. I'm gonna be doing some testing this week on a few potential changes to improve cooling, and underside paste is one of them.

Thanks for the info, then I will probably order some thermal pad and do some tests here as well. Actually when touching the bottom heatsink, this really feels just warm, so the cooling should definitely improve with proper heat conduct. (which was in fact already pointed out/confirmed by @gergar)

[n0nce]

Heatsink is about 45x70mm

Yes there's no interface between the bottom heatsink and the board. I'm gonna be doing some testing this week on a few potential changes to improve cooling, and underside paste is one of them.

Paste is a good idea, since there is no conductivity issue on the back of the PCB. Is it possible to remove the back side heatsinks without touching the ones on the front? I'm a little hesitant touching those, since I wouldn't know where to get replacement thermal pads with 20W/mK of thermal conductivity and electrical isolation. In the name of right to repair and keeping these things running for as long as possible, I'd really appreciate if you could release more information on the ones you put into the machines.
I regard them as consumables, just like the fan, which you did a great job of documenting.

For what it's worth, I plan to design a 3D-printable shroud for the back of the unit that lets you route the cables out the top and funnels the air further back where a secondary fan will be placed. Interested to see if it makes any difference.

I love how tinker-friendly this device is, not gonna lie!

[Sledge0001]

Well 2 of the units came in tonight and I am not having much luck with one of em..

1 seems to be spot on from the factory.

The other I am not sure why but it doesn't seem to be coming to life or getting anywhere near the hashrate it should.

I've swapped USB cables, tried a different PSU, tried only 1 R909 but no luck  

Anyone have suggestions?

Code:

cgminer version 4.12.1 - Started: [2023-01-24 19:44:49.046]
--------------------------------------------------------------------------------
 (5s):1.995T (1m):1.802T (5m):653.5G (15m):244.7G (avg):2.037Th/s
 A:55629  R:0  HW:1280  WU:28262.8/m
 Connected to solo.ckpool.org diff 519 with stratum as user 
 Block: a7b97cfa...  Diff:37.6T  Started: [19:45:21.343]  Best share: 282K
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070005: BM1397:06+ 450.00MHz T:450 P:439 (4:2) |  100% WU:^97% | 1.432T / 1.753Th/s WU:24490.1/m
 1: GSF 10070040: BM1397:05+ 450.00MHz T:450 P:109 (5:3) | 21.6% WU: 18% | 367.3G / 270.1Gh/s WU: 3772.7/m

What I do notice is one shows up with 6+ and the other with 5+....

The one with 5+ is not performing in spec.

[gergar]

Heatsink is about 45x70mm

Yes there's no interface between the bottom heatsink and the board. I'm gonna be doing some testing this week on a few potential changes to improve cooling, and underside paste is one of them.

Thanks for the info, then I will probably order some thermal pad and do some tests here as well. Actually when touching the bottom heatsink, this really feels just warm, so the cooling should definitely improve with proper heat conduct. (which was in fact already pointed out/confirmed by @gergar)

thermal conductivity is impacted due to height of thermal path so, don’t exceed 1mm so still stays in a good small size and then you dont need crazy high conductivity pads 20+ and a 13+ will do a great job, 0.5 can be very thin and can get easily damaged so a .75-1mm to my experience is a good choice and still will effectively conduct heat to the heatsink.

[gergar]

Well 2 of the units came in tonight and I am not having much luck with one of em..

1 seems to be spot on from the factory.

The other I am not sure why but it doesn't seem to be coming to life or getting anywhere near the hashrate it should.

I've swapped USB cables, tried a different PSU, tried only 1 R909 but no luck  

Anyone have suggestions?

Code:

cgminer version 4.12.1 - Started: [2023-01-24 19:44:49.046]
--------------------------------------------------------------------------------
 (5s):1.995T (1m):1.802T (5m):653.5G (15m):244.7G (avg):2.037Th/s
 A:55629  R:0  HW:1280  WU:28262.8/m
 Connected to solo.ckpool.org diff 519 with stratum as user 
 Block: a7b97cfa...  Diff:37.6T  Started: [19:45:21.343]  Best share: 282K
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070005: BM1397:06+ 450.00MHz T:450 P:439 (4:2) |  100% WU:^97% | 1.432T / 1.753Th/s WU:24490.1/m
 1: GSF 10070040: BM1397:05+ 450.00MHz T:450 P:109 (5:3) | 21.6% WU: 18% | 367.3G / 270.1Gh/s WU: 3772.7/m

What I do notice is one shows up with 6+ and the other with 5+....

The one with 5+ is not performing in spec.

use Kano new miner.php and get into the gekkochip page check if all 6 are reporting, also at stats if your chips are failing to get set in frequency few causes:
Power not enough - you need a 100w as minimum to run these ones, or they are getting very hot let them cool, and try to push with no to little force the top heatsink and see if move, if it does then the heatsink bolts might got losen with shipping

[sidehack]

Definitely check the heatsinks for tight screws. Reporting 5 chips is problematic; was it doing that from the start, or after running a bit?

[Sledge0001]

Definitely check the heatsinks for tight screws. Reporting 5 chips is problematic; was it doing that from the start, or after running a bit?

Right from the go.

And its 2 r909's on a 750W psu so I know its getting enough powa!

I'll dig into this tomorrow a bit more and check the screws.

UPDATE:
This morning woke up and Zombie mode for the one that was showing on 5+ The other chugging away averaging 1.7+TH

Reboot doesn't seem to help. I'll reach out to the seller and hope they have a spare somewhere since they are sold out.

[100knot2dae]

thermal conductivity is impacted due to height of thermal path so, don’t exceed 1mm so still stays in a good small size and then you dont need crazy high conductivity pads 20+ and a 13+ will do a great job, 0.5 can be very thin and can get easily damaged so a .75-1mm to my experience is a good choice and still will effectively conduct heat to the heatsink.

Thanks, I think I will order a set of these Iceberg pads: https://www.iceberg-thermal.com/product/consumer/thermal-pad/driftice/. 80x40 should be a near to perfect fit, just need to cut the length a little. Hope that I won't mess up with the upper heatsinks' thermal pad when re-screwing the bottom heatsinks...

Actually the unit is running very well at the moment, but I don't think I can leave it now without these pads attached, now that I know of the air gap.

It's like @n0nce has said, this neat device makes you want to tinker with it. 

@sidehack: Would be great if you could share details on the thermal pads used for the upper heatsinks.

[gergar]

thermal conductivity is impacted due to height of thermal path so, don’t exceed 1mm so still stays in a good small size and then you dont need crazy high conductivity pads 20+ and a 13+ will do a great job, 0.5 can be very thin and can get easily damaged so a .75-1mm to my experience is a good choice and still will effectively conduct heat to the heatsink.

Thanks, I think I will order a set of these Iceberg pads: https://www.iceberg-thermal.com/product/consumer/thermal-pad/driftice/. 80x40 should be a near to perfect fit, just need to cut the length a little. Hope that I won't mess up with the upper heatsinks' thermal pad when re-screwing the bottom heatsinks...

Actually the unit is running very well at the moment, but I don't think I can leave it now without these pads attached, now that I know of the air gap.

It's like @n0nce has said, this neat device makes you want to tinker with it. 

@sidehack: Would be great if you could share details on the thermal pads used for the upper heatsinks.

The upper are 0.5mm 20w/k. the ones you selected are ok for the back plate, these can be an top plate option as they are 20w/mk and they have different sizes so you can order  0.5 just in case and if you see ok to have both the same type then a 1mm (the pads can be cut with no effort with regular pair of scissors, so 1x0.5 will be enough for 1 unit top plate the two heatsinks, and also will work the same to backplate) :

https://www.amazon.com/Kritical-Thermal-Pads-Conductivity-othermicroelectronic/dp/B0B1ZF7MZ7/ref=sr_1_18?crid=2MYFZWYK4O5F2&keywords=thermal%2Bpath%2B20w%2Fmk%2B0.5&qid=1674689278&sprefix=thermal%2Bpath%2B20w%2Fmk%2B0.5%2Caps%2C152&sr=8-18&th=1

[gergar]

Two things:

1) The CompacF mines at a minimum of 16 diff - I've coded it to not go lower.
The R909 miners at a minimum of 64 diff - I've coded to not go lower.
These values are already below what is needed.
People should expect pools to be pissed off at them if they tried to DDoS the pool with lower values.

2) All about worker difficulty: https://kano.is/index.php?k=workdiff

Hi @Kano, I tired the api and made it work half way, I changed the number on ASIC ID from 0 to 1 to modify the 2nd unit and it responds "ASC 0 set OK", and it set the frequency to all chips on Asia 0 vs. 1, and also using chip lets say 0,chip,0:0 it set all to 0 on ASC 0 vs. only  chip 0. then if I change to 1,chip,0:0 it sets same all to 0 on ASC 0.

this is how I am sending the payload: {"command": "ascset", "parameter": "[{1,chip,0:0}]"} (then I encode it) is this OK? 

initially I did this but sends error, and landed making it partially work with the previous one {"command": "ascset", "parameter": "[{"ID":"1","option":"chip","value":"0:0"}]"}
or for frequency to all {"command": "ascset", "parameter": "[{"ID":"1","option":"freq","value":"0"}]"}

hope you can give me some advise.

[kano]

https://github.com/kanoi/cgminer/blob/master/API-README#L70

So {"command":"ascset","parameter":"1,chip,0:0"}

However, you don't need to use JSON, so as per https://kano.is/gekko.php

java API "ascset|1,chip,0:0" MinerIP

[gergar]

https://github.com/kanoi/cgminer/blob/master/API-README#L70

So {"command":"ascset","parameter":"1,chip,0:0"}

However, you don't need to use JSON, so as per https://kano.is/gekko.php

java API "ascset|1,chip,0:0" MinerIP

I will give it a try, if not i will try to call using shell the JAVA Api, just wanted to avoid os calls. THANK YOU!

[rdbase]

so i will run my 2 gekko usb hubs (with 3 compac f sticks each) together with the 2 new r909 pod miners using a rpi4. if everything runs well and is configured well i would get ~3.6th/s together

Did it work out to be this hashrate?
I would imagine stable at the right settings and the right fan placements for the one on the usb hubs.
After 2~ months later you should have an idea if this arrangement works.
Was eyeing a gekkoscience pod for a while and I might pull the trigger on one in this setup if the time is right.

How do they compare to buying a mars solo miner from bitcoin merch and plugging in several of the compaq F1s into a hub?
Would the price of the 909 be relativity the same for the amount of hashrate produced?

[CochnocherCrypto]

Hey everyone - so I see lots of chatter about thermal pads here. I routinely re-paste all of my asic miners every 3 months or so because they run 24/7 and my cat's dander is like a gnarly dust invasion for my miners because of all of the fans in my apartment blowing it directly at them lol. Has anyone removed the top heatsinks from the R909 and can they confirm wether it has pads or paste on the chips?

[os2sam]

Hey everyone - so I see lots of chatter about thermal pads here. I routinely re-paste all of my asic miners every 3 months or so because they run 24/7 and my cat's dander is like a gnarly dust invasion for my miners because of all of the fans in my apartment blowing it directly at them lol. Has anyone removed the top heatsinks from the R909 and can they confirm wether it has pads or paste on the chips?

Not sure if this answers your question or not, but it seems similar:

https://bitcointalk.org/index.php?topic=5423227.msg61628663#msg61628663

[gergar]

Hey everyone - so I see lots of chatter about thermal pads here. I routinely re-paste all of my asic miners every 3 months or so because they run 24/7 and my cat's dander is like a gnarly dust invasion for my miners because of all of the fans in my apartment blowing it directly at them lol. Has anyone removed the top heatsinks from the R909 and can they confirm wether it has pads or paste on the chips?

I did take all apart it has pads 20w/mk 0.5mm i replaced with new same specs.

Confirmed also here:
https://bitcointalk.org/index.php?topic=5423227.msg61628663#msg61628663

[kano]

https://github.com/kanoi/cgminer/blob/master/API-README#L70

So {"command":"ascset","parameter":"1,chip,0:0"}

However, you don't need to use JSON, so as per https://kano.is/gekko.php

java API "ascset|1,chip,0:0" MinerIP

I will give it a try, if not i will try to call using shell the JAVA Api, just wanted to avoid os calls. THANK YOU!

Whatever you use, there are 2 formats

"ascset|1,chip,0:0"
or
'{"command":"ascset","parameter":"1,chip,0:0"}'

(aside: I designed and wrote the cgminer API)

[altair_tech]

I was thinking about ways to push this beyond the 2.4-2.5Th reported by other users.
One way could be to use a bigger fan and adapter to funnel more air CFMs into the pod.
For example a 120mm or 140 mm fan can push 2-3x air through the heatsinks, while still staying fairly quiet (<25db).
I've seen some s9 space heater designs (skip to page 9) that are replacing the stock 120mm fan with 140mm fan to allow the machine to run quietly without compromising on cooling.

Would like to know from others here, if this makes sense and is worth a try.

-Aviral

[MoparMiningLLC]

Our (cryptocloaks) space heater replaces an s9’s psu fan and miner 120mm with 140mm fans - and it stays pretty quiet after that. Around 45-50 dB

If it helps, the fan we used:

Noctua NF-A14 iPPC-3000 PWM, Heavy Duty Cooling Fan, 4-Pin, 3000 RPM (140mm, Black)

 https://a.co/d/6r1YEdt

[yettihead]

I was thinking about ways to push this beyond the 2.4-2.5Th reported by other users.
One way could be to use a bigger fan and adapter to funnel more air CFMs into the pod.
For example a 120mm or 140 mm fan can push 2-3x air through the heatsinks, while still staying fairly quiet (<25db).
I've seen some s9 space heater designs (skip to page 9) that are replacing the stock 120mm fan with 140mm fan to allow the machine to run quietly without compromising on cooling.

Would like to know from others here, if this makes sense and is worth a try.

-Aviral

That is what I have done.  I created an adapter to go up to a 92mm fan as the airflow increases from 55.5 m³/h for the 80mm up to 78.9 m³/h for the 92mm.  The 120mm could easily be done, just trying to keep these somewhat small, but the airflow would increase up to 102.1 m³/h.  I do have a 140mm fan I could experiment with, but the pressure drops.  Almost everything I 3D print is for mining purposes.

https://imgur.com/a/gidpBJ8

[gergar]

https://github.com/kanoi/cgminer/blob/master/API-README#L70

So {"command":"ascset","parameter":"1,chip,0:0"}

However, you don't need to use JSON, so as per https://kano.is/gekko.php

java API "ascset|1,chip,0:0" MinerIP

I will give it a try, if not i will try to call using shell the JAVA Api, just wanted to avoid os calls. THANK YOU!

Whatever you use, there are 2 formats

"ascset|1,chip,0:0"
or
'{"command":"ascset","parameter":"1,chip,0:0"}'

(aside: I designed and wrote the cgminer API)

Thank you @Kano!!! yeah I am aware you are the designer and developer amazing job!!! it is impressive THANK YOU!!

[gergar]

I was thinking about ways to push this beyond the 2.4-2.5Th reported by other users.
One way could be to use a bigger fan and adapter to funnel more air CFMs into the pod.
For example a 120mm or 140 mm fan can push 2-3x air through the heatsinks, while still staying fairly quiet (<25db).
I've seen some s9 space heater designs (skip to page 9) that are replacing the stock 120mm fan with 140mm fan to allow the machine to run quietly without compromising on cooling.

Would like to know from others here, if this makes sense and is worth a try.

-Aviral

That is what I have done.  I created an adapter to go up to a 92mm fan as the airflow increases from 55.5 m³/h for the 80mm up to 78.9 m³/h for the 92mm.  The 120mm could easily be done, just trying to keep these somewhat small, but the airflow would increase up to 102.1 m³/h.  I do have a 140mm fan I could experiment with, but the pressure drops.  Almost everything I 3D print is for mining purposes.

https://imgur.com/a/gidpBJ8

Nice one!! i used a high CFM Axial fan, it is in some of my postings, i will share my results of my testing just needed to do one more … but those 80mm fans made a lot i can tell full speed kept the heatsinks on 37 Degree Celsius and i was running at 725Mhz 3.2-3.4 T and power was strangely flat to about 140W the issue i saw was more on the asic was not keeping up the frequency. but i can tell got interesting results I will post them.

[sidehack]

Holy crap did you say 3.2TH? 725MHz should see 2.9TH but that's still pretty nuts. The main regulator should cap at around 120 watts output.

[kano]

Alas, it does require running for a while.
I've even had 2.9TH/s reported at startup due to random luck.

As long as you reset the display as explained at the beginning of https://kano.is/gekko.php#perf
and then let it run for a few hours, then the reported hash rate is meaningful.

[cygan]

so i will run my 2 gekko usb hubs (with 3 compac f sticks each) together with the 2 new r909 pod miners using a rpi4. if everything runs well and is configured well i would get ~3.6th/s together

Did it work out to be this hashrate?
I would imagine stable at the right settings and the right fan placements for the one on the usb hubs.
After 2~ months later you should have an idea if this arrangement works.
Was eyeing a gekkoscience pod for a while and I might pull the trigger on one in this setup if the time is right.
✂

my setup is running very stable so far, as you can see below
unfortunately i have only one compac f stick running on one of the two usb hubs - due to overclocking the other ports on the hub have died (and i'm waiting for a new gekko hub delivery )
all hardware is only connected to one and the same rpi4

Code:

0: GSF 10051796: BM1397:01+ 525.00MHz T:515 P:510 (24:12)  |  100% WU:100% | 309.7G / 312.3Gh/s WU: 4363.1/m
1: GSF 10051619: BM1397:01+ 525.00MHz T:525 P:525 (24:12)  |  100% WU:100% | 321.9G / 349.9Gh/s WU: 4887.6/m
2: GSF 10053618: BM1397:01+ 525.00MHz T:525 P:525 (24:12)  |  100% WU:100% | 361.2G / 348.9Gh/s WU: 4873.6/m
3: GSF 10070009: BM1397:06+ 680.00MHz T:680 P:662 (3:2)    | 88.1% WU: 90% | 2.428T / 2.461Th/s WU:34383.3/m
4: GSF 10070003: BM1397:06+ 640.00MHz T:640 P:635 (3:2)    | 88.2% WU: 91% | 2.487T / 2.347Th/s WU:32788.6/m
5: GSF 10053617: BM1397:01+ 525.00MHz T:525 P:525 (24:12)  |  100% WU:100% | 300.3G / 343.8Gh/s WU: 4802.8/m

[gergar]

Holy crap did you say 3.2TH? 725MHz should see 2.9TH but that's still pretty nuts. The main regulator should cap at around 120 watts output.

Ohh now i see why i get flat behavior after 700mhz, the power i am measuring includes the axial fans and my RBPI to get to 725 i push my fans to 70%, i assume these are the extra 20w.

This is what i see:
- 725 not stable at all- got 3.2 for 3 minutes then crashes down to 500Mhz and as it reboots and reboots Hashrate became to 800G but power never came down. Temps on heatsinks were stable at 45 degrees celsius. I set anoush pull using original in front and exhaust acial at 50% the front i put it on tilted position resting on the cables and pushing the air towards the board (i guess at 45 degrees of angle) that caused higher heat efficiency and took the heatsinks to stable 36 degrees celcius BUT 0 effect on stability… and power got to 138 ( the 3 wats i assume got from the added fan). same 3.2 then crash the 800.

-700 push pull stable on 2.7-2.8 i did not saw these going higher than that after 24 hrs, temps stable on 35 degrees the hottest spot. consumption still high 135.

-650 the current i have only pull axial at 2700 rmps silent ops, stable 2.57- 2.6 T @ 114.5w consumption/ 42 Degree celcius at heatsinks /  net that is quite great vs other setups as looks close to 45j/t and the others were like 50.

something is wrong at 725, i have no idiea what to change ( but not sure if changing stock voltage will help) i dont want ro fry them.

so we can say 3.2 is not valid as they crashes after little time… so to me at least not messing up with voltage 2.8 is the top. i am converting to water cooling but as i saw heatsinks on 35 degrees and no improvement i don’t expect it to change, same as i want it to run 24x7 650 will give better efficiency and longevity.

i am on my iphone so cannot paste the cgminer screen but miner.php and looks ugly but here it is at 650, close to 24 hr run.

Date: 17:47:01 29-Jan-2023 UTC+00:00
Computer: cgminer 4.12.1 When: 17:47:01 29-Jan-2023 UTC+00:00 Status: Success Message: 2 ASC(s)
ASC Name ID Enabled Status Temperature MHS av
MHS 5s
MHS 1m
MHS 5m
MHS
0
GSF
0
Y
Alive
1
GSF
1
Y
Alive
Total:
2,573,644.91 3,004,809.31 2,692,198.92 2,618,194.45 2,5!
2,606,626.98 2,414,314.80 2,616,199.07 2,626,635.49 2,6
5,180,271.89 5,419,124.11 5,308,397.99 5,244,829.94 5,2

[gergar]

Alas, it does require running for a while.
I've even had 2.9TH/s reported at startup due to random luck.

As long as you reset the display as explained at the beginning of https://kano.is/gekko.php#perf
and then let it run for a few hours, then the reported hash rate is meaningful.

yeah so 3.2 was luck and not sustained over time. so mine is 2.8 T highest i got but they were running very cool!!!)

@sidekick what will be the effect on moving the variable resistor on voltage? will it give me higher Hashrate/ or put me into the power cap soon? and will the asic be damaged if all left or all right? or it is safe to play with it?

[n0nce]

what will be the effect on moving the variable resistor on voltage? will it give me higher Hashrate/ or put me into the power cap soon? and will the asic be damaged if all left or all right? or it is safe to play with it?

You can definitely increase voltage if you have the cooling (which it seems, you do). Especially if you decide to go for watercooling, you should have enough thermal headroom to give it more power through that potentiometer.
Reducing it from stock may be interesting for improving efficiency, but as far as I know these are already individually tuned by the factory and set to a good safe value. So if anything, I'd find it interesting to see how far it can go with ample cooling and higher 'power target'.

[sidehack]

You're probably butting into the power headroom already. In fact I'm surprised you didn't hit it already, but keeping the main regulator cool probably helps a little with Rdson current measurement.

More voltage should give you more ability to clock higher, but the main regulator is current limited. Shouldn't hurt anything to turn the voltage up a bit and see where it goes. If you suddenly drop to no hashing and only fans drawing power, you've tripped the OC and need to power cycle the whole unit. The voltage range capable with the pot is safe; just make sure you keep the chips cool and there should be no damage.

I've got a test unit I want to mess with, raise the current limit (will require replacing some parts to raise the safety ceiling) and repeat some of your tests to see about improved cooling efficiency and push the hashrate super high.

[n0nce]

The voltage range capable with the pot is safe; just make sure you keep the chips cool and there should be no damage.

Regarding safety and damage: I know all types of overclocking come with some amount of risk, but usually it is related to voltage and heat. What about pushing clock speed without changes to the voltage (potentially even slightly undervolting and improving cooling with better fans, ducts, thermal pads on the back...)?
Are there any hints that suggest a strong overclock deteriorates ASIC mining chips faster than stock clock speed? [while keeping voltage and heat on normal levels]

[sidehack]

As long as heat is managed, the only damage will be your hashrate tanking because the data no longer makes sense.

[n0nce]

As long as heat is managed, the only damage will be your hashrate tanking because the data no longer makes sense.

But can it produce more heat when clocked higher, even though the potentiometer is not touched?

In CPU overclocking, as long as you use fixed voltages, the heat output should not be affected by the clock speed. It is usually directly related to voltage.

[gergar]

https://github.com/kanoi/cgminer/blob/master/API-README#L70

So {"command":"ascset","parameter":"1,chip,0:0"}

However, you don't need to use JSON, so as per https://kano.is/gekko.php

java API "ascset|1,chip,0:0" MinerIP

I will give it a try, if not i will try to call using shell the JAVA Api, just wanted to avoid os calls. THANK YOU!

Whatever you use, there are 2 formats

"ascset|1,chip,0:0"
or
'{"command":"ascset","parameter":"1,chip,0:0"}'

(aside: I designed and wrote the cgminer API)

Hey @kano!!!! THANK YOU my code now works so i can test each chip and set temp telemetry on each this is amazing!!!

THANK YOU!!!

[gergar]

As long as heat is managed, the only damage will be your hashrate tanking because the data no longer makes sense.

But can it produce more heat when clocked higher, even though the potentiometer is not touched?

In CPU overclocking, as long as you use fixed voltages, the heat output should not be affected by the clock speed. It is usually directly related to voltage.

it does as more current will pass as it switches more times per second, so drains more power. these asics gets quite hot at 700-725 mhz the stock fan wont keep up, the 6600 rpm axial fan i set helps a lot to keep them but i see generating the push pull helped me to pull more heat out but i saw no hashrate improvements in my case.

i will test as @sidekick shared, i will now under and over volt it and do push pull with axials so y pass a ton of air to it to avoid having thermal issues and see where do we get

[sidehack]

Power use increases with voltage, because the tiny switch gates basically act as capacitors so with a higher voltage, there's proportionally more current transferred to flip the switch on and off.
Power use increases with clock speed, because there are more switch flips happening.

[100knot2dae]

Just wanted to report back that I have added thermal pads to the bottom heatsinks on the weekend. While everything went well, the expected cooling benefit hasn't really been met in my opinion.

Yes these heatsinks do get warmer now, so there is better heat dissipation. But not significant in a way to have a measureable impact on watts or hashrate. I had hoped these pads to be a thermal bonus for my efficiency setup (550Mhz, undervolted), but - if at all - they are probably more of use for the heavy overclocking case.

[cygan]

my small solo mining farm is now complete and i can say that a rpi4 8gb ram can handle all these miners (6 compac f sticks and 2 r909 pod miners) without any problems and crashes.
as you can also see, the rpi4 is using less than 500mb of memory.

Code:

cgminer version 4.12.1 - Started: [2023-02-23 15:39:58.371]
--------------------------------------------------------------------------------
 (5s):7.002T (1m):6.695T (5m):6.675T (15m):6.703T (avg):6.746Th/s
 A:242821020  R:119132  HW:96240  WU:94247.9/m
 Connected to solo.ckpool.org diff 4.11K with stratum as user .....
 Block: 49f875db...  Diff:43.1T  Started: [10:52:41.548]  Best share: 190M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10051796: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 320.0G / 311.3Gh/s WU: 4349.5/m
 1: GSF 10051619: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 420.0G / 349.4Gh/s WU: 4881.4/m
 2: GSF 10053617: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 294.1G / 325.8Gh/s WU: 4551.3/m
 3: GSF 10053622: BM1397:01+ 525.00MHz T:525 P:525 (24:12) |  100% WU:100% | 348.2G / 340.9Gh/s WU: 4762.2/m
 4: GSF 10051622: BM1397:01+ 525.00MHz T:515 P:521 (24:12) |  100% WU:100% | 325.9G / 307.7Gh/s WU: 4298.9/m
 5: GSF 10053618: BM1397:01+ 525.00MHz T:515 P:525 (24:12) |  100% WU:100% | 385.7G / 348.3Gh/s WU: 4866.4/m
 6: GSF 10070009: BM1397:06+ 670.00MHz T:670 P:652 (3:2)   | 87.5% WU: 90% | 2.350T / 2.430Th/s WU:33947.7/m
 7: GSF 10070003: BM1397:06+ 640.00MHz T:640 P:623 (3:2)   | 89.6% WU: 90% | 2.063T / 2.332Th/s WU:32590.6/m

[100knot2dae]

Does anybody actually power this pod miner with a barrel jack PSU, instead of a plain computer PSU?

@sidehack: You have mentioned in the first post

Like previous Terminus models, the miner runs on 12V power from either a standard 2.1/5.5mm barrel jack (rated for 8A) or a PCIe 6-pin jack

So, is the PCIe connector on this miner rated for higher current or are both connectors rated the same? Is there any (other) downside one should be aware of when using the barrel jack?

Thanks for letting me know.

[altair_tech]

Does anybody actually power this pod miner with a barrel jack PSU, instead of a plain computer PSU?

Yes, it works fine with 2.1/5.5 mm barrel jack PSU. Those are rated for upto 8 Amps, but I've gone upto 10Amps on 2 different 2.1/5.5mm barrel jack PSU's without any problem.
It is safer to run it on a PSU with 6-pin PCIe connector though, specially if you plan to overclock it over 120 Watts. Those are rated for much higher currents. For example on an antminer S9 , 9x PCIe connectors deliver 1650Watts. That's about 15 Amps per PCIe connector.

[NotFuzzyWarm]

A couple things to keep in mind:
Barrel jacks are not intended to deliver their maximum rated current 24x7x365

Barrel jacks are rated with the assumption that they are periodically disconnected/re-connected so any oxidation is scraped off and a fresh contact surface is made. At low to moderate currents this is not an issue, as you approach the connector max current rating it is definitely a concern.

As for PCIe plugs - max current is HIGHLY dependent on what kind of plating is on the pins. If silver or gold each pin *can* handle up to 5A per pin. Since there are 3x circuits and each takes 2 pins (1 feed, 1 return) you get the 15A max for the 6-pin connector. Safe max for tin plating is just 3A. Those max ratings also are highly dependent on what gauge & type of wire is used as the wire pulls heat away from the connection. Typically max current only applies to using 16ga copper wire.

[100knot2dae]

Barrel jacks are rated with the assumption that they are periodically disconnected/re-connected so any oxidation is scraped off and a fresh contact surface is made. At low to moderate currents this is not an issue, as you approach the connector max current rating it is definitely a concern.

Didn't know about that aspect, thanks for pointing this out. The PCIe connector definitely seems to be the better choice here.

[GlenAllan]

Hello, newb here... Having a pain getting my R909 to work properly on the same system as 4 Compac Fs. It was recommended i ask in here to see if anyone has an idea of what's going on instead of in my own post.

Command i'm running: /opt/homebrew/Cellar/cgminer/4.12.1/bin/cgminer -o stratum+tcp://solo.ckpool.org:3333 -u "my bitcoin address" -p x --gekko-r909-freq 500 --gekko-compacf-freq 545 --gekko-start-freq 300 --gekko-tune-up 85

I've tried running the R909 out of a free port on the Gekko Science hub, tried running both through a USB 2 hub out of the same USB port, tried running through different USB ports on the same computer, all no gos. It crashes resets and tries again. The WU: (Work uptime?) for the 909 gets stuck between 50 and 65% and never gets higher once the frequency is reached.

I should say the R909 works totally fine on a different computer by itself as a separate worker, or on the first computer without the Compac Fs running. I've tried running all devices on both an M1 Mac Mini and an iMac Pro, and on both the WU stays low for the 909, then resets.

Only other thing i can think of is i'm running the kanoi/cgminer fork that installs through homebrew. It is at version 4.12.1.

I'm new enough that this is all out of my depth. Any suggestions for what i should be doing here would be much appreciated.

Thanks!

[kano]

The sleep timer in OSX is the worst of the 3 OSes.

That's why:

API Usage
As before there's a global wait factor override to adjust the rate that work is sent to the miner.
The default should be fine which is 0.5 for Linux, 0.4 for Windows and 0.3 for Mac

One R909 is closely equivalent to 6xCompacF.
It definitely works OK on it's own, but timing can run amok when you have too many threads trying to time work to the chips on OSX.
That limit will of course be dependent upon the hardware, but by the looks of things for your case, 10 chips is too much for one.

[GlenAllan]

Okay, so are you saying there is a way to adjust the global wait factor? Or that it's not going to work on OSX at all? I'm running 18 USB devices on that machine no problem otherwise (yes i tested this without them hooked up). I did have USB issue on Intel Machines with too many hard drives connected, on a couple of machines, but i've never had any issue on Apple Silicon. I know you said OSX itself, but not sure if that would be any different.

[kano]

It's the 'timing', not exactly the number of 'devices'.

For a single R909 it has to send work around once every 4-5ms.
If it gets delays from thread switching, those delays wont normally be as low as 1ms on OSX.
You can see and compare the number in the estats output.

Check [Sleep... e.g. below

Code:

[STATS] => 1
[ID] => GSF0
[Elapsed] => 13417

[FreqComp] => 499.995331

[MaxTaskWait] => 4260
[WaitFactor0] => 0.500000
[GHGHs] => 1951.648580
[JobDataAge] => 0.004327
[JobsAvgms] => 4.27/4.27/4.27/4.27/4.27
[JobsMinMaxms] => 4.01:4.56/4.01:5.40/4.01:5.50/4.01:5.44/4.01:8.59

[Dups] => 17
[DupsReset] => 17
[Chips] => 6

[SleepN] => 16416669
[SleepAvgReq] => 1000.063289
[SleepAvgRes] => 1.074009
[SleepN1_1] => 32839824
[SleepAvgReq1_1] => 222.902610
[SleepAvgRes1_1] => 1.335763
[SleepN1_5] => 240071
[SleepAvgReq1_5] => 216.000991
[SleepAvgRes1_5] => 1.660623
[SleepInv] => 1

That's on an RPi4 doing 1.95TH/s on 1 of the 4 miners connected to it.
(2xR909, 1xCompacF, 1xNewPac)

It's calculated [FreqComp] => 499.995331 MHz is getting [GHGHs] => 1951.648580 GH/s (over the last 5 minutes)

It wants work [MaxTaskWait] => 4260 i.e. 4.2ms
Over the last 5 minutes it's averaging very well [JobsAvgms] => 4.27/4.27/4.27/4.27/4.27
Of course not every job is perfect, [JobsMinMaxms] => 4.01:4.56/4.01:5.40/4.01:5.50/4.01:5.44/4.01:8.59
JobsMinMaxms says what the min and max ms have been over the last 5 minutes

Sleep times:
Note there are normally 2 sleeps per job sent.
16416669 with an average of 1000.063289 microseconds averaged 107.4009% of the requested delay (for all that averaged less than 110% delay)
32839824 with an avg of 222.902610us avg 133.5763% of the requested delay (for all between 110 and 150%)
240071 with an avg of 216.000991us avg 166.0623% (for that were over 150%)

I'd expect your OSX numbers to be a lot worse and get progressively worse with more miners.

This all boils down to CPU and OS sleep time accuracy of usleep() on that OS
Surprisingly it is hard for all OS to sleep a thread 1ms and wake it up 1ms after the call.
The order of accuracy is Linux > Windows > OSX

Thus:

API Usage
As before there's a global wait factor override to adjust the rate that work is sent to the miner.
The default should be fine which is 0.5 for Linux, 0.4 for Windows and 0.3 for Mac

As explained on https://kano.is/gekko.php#apu
If you have CPU spare you can try lowering the wait factor.
What it means is it will try to send work faster than the R909 needs.
Which means if it is sending it slower than it needs, it will make the averages better.

[GlenAllan]

Huh, i haven't actually used the API to do anything yet, so i might give some adjustments a try.

Otherwise i do have a couple of Raspberry Pis, so i can repurpose one for this. It was being used for live video triggering for video production, but i don't think i need both for that. Interesting that a Pi running Linux can work better than a Mac or Windows machine.

Out of curiosity, what's the best distro for this on the Pi. Actually i can just go look that up.

Thanks for the info! Would be nice if this wasn't an issue, but i'll get around it. I do have CPU to spare.

[GlenAllan]

And to be sure i'm not just an idiot, is it as simple as enabling the api and then adding the wait factor commend, like this:
/opt/homebrew/Cellar/cgminer/4.12.1/bin/cgminer -o stratum+tcp://solo.ckpool.org:3333 -u "BTC Address -p x --gekko-r909-freq 500 --gekko-compacf-freq 545 --gekko-start-freq 300 --gekko-tune-up 85 --api-listen --api-allow "W:192.168.1.0/24,W:127.0.0.1" --gekko-wait-factor 0.1

Assuming that's correct (which i don't assume). How low is safe to go? 0.01, etc?

[GlenAllan]

Never mind. I found the part. Helps to actually read more. Now to figure out how to use the API properly.

java API "ascset|0,waitfactor,0.6" minerIP

Now where to put the adjustment.

[sidehack]

Yeah the Pi runs these guys pretty well but it's entirely because of the OS scheduler that it's better than more powerful machines with OSX or MSX. Our production testers are old Dell 755 machines from approximately 2008 that are stable with 10xR909 at stock speed. Meanwhile, on a newer stronger Mac...

[GlenAllan]

What a trip! Well i've got a Pi 4 with 8GB RAM running on an SSD, so i'll grab Ubuntu and get it going for mining and see what performance i'm missing from all of these then. Should be interesting to set up!

I did all the wait factor adjustments on the Mac and it just didn't really do much. I did test the commands to be sure they were working, it just didn't do anything meaningful. So Pi it is!

[sidehack]

I think the practical limit for a Pi is around 15 chips, so that'll run two R909 by itself. I don't know that anyone's gotten expected speed out of three together on one Pi.

[GlenAllan]

I see... Well right now i just want to optimize for 4 Compac Fs and 1 r909. I assume then that is actually about as much as i could do outside of adding a couple more Compac Fs.

Good to know!

[GlenAllan]

And as for the OS... I was going to go for Ubuntu MATE for being a bit lighter. Any reason to do anything different if i want a desktop UI?

[cygan]

i run my rpi 4 8gb ram with 2 r909 and a total of 6 compac f sticks (so a total of 18 chips) and that without any problems and performance loss. of the said 8gb ram my rpi doesn't even consume 500mb for this configuration. so i could imagine that you could connect at least one (third) r909 pod miner to the same rpi.
as os i use the raspios bullseye: https://www.raspberrypi.com/software/operating-systems/

[GlenAllan]

Does changing things in the api set the hardware after it's connected to something else? All of a sudden i'm only getting 3 of the 6 chips on the r909 showing up after a move to a different machine. Just wondering.

[maddhy]

Hello, newbie here and I'm also a newbie miner.

I just got a R909 and today I've been trying to connect it to Nicehash, which worked but xnsub is not supported.

The customer service told me to enable extranonce using the code "stratum+tcp://sha256.auto.nicehash.com:9200#xnsub", however, it did not work.

I wonder if there is any tweak that can make it work? Many thanks!

[paid2]

Hello, newbie here and I'm also a newbie miner.

I just got a R909 and today I've been trying to connect it to Nicehash, which worked but xnsub is not supported.

The customer service told me to enable extranonce using the code "stratum+tcp://sha256.auto.nicehash.com:9200#xnsub", however, it did not work.

I wonder if there is any tweak that can make it work? Many thanks!

Did you tried with this fork of cgminer ?

nicehash : https://github.com/nicehash/cgminer-ckolivas.git (nicheash extranonce)

[GlenAllan]

Is there some way to do a hard reset on these. Mine is now stuck only showing 3 of the 6 chips and isn't hashing above141Gh and keeps resetting.

[NotFuzzyWarm]

Did you tried with this fork of cgminer ?

nicehash : https://github.com/nicehash/cgminer-ckolivas.git (nicheash extranonce)

I highly doubt it would work as it will not have the s17 chip driver...
You *are* aware that there is a reason #xnsub is not a normal part of cgminer right? Because#xnsub silently bypasses the coinbase checks that would reject work. Without those checks the work can switch to going to anyone, anywhere, any time, any coin, any pool, anything without you ever knowing. Great for pools to take advantage of you. Because no regular mining needs to have that ability it is not a good idea to automatically include support for #xnsub

That said, it Nicehash requires it so your rental miners can change work on-the-fly w/o having to reboot.

[n0nce]

i run my rpi 4 8gb ram with 2 r909 and a total of 6 compac f sticks (so a total of 18 chips) and that without any problems and performance loss. of the said 8gb ram my rpi doesn't even consume 500mb for this configuration. so i could imagine that you could connect at least one (third) r909 pod miner to the same rpi.
as os i use the raspios bullseye: https://www.raspberrypi.com/software/operating-systems/

RAM may not be the bottleneck, but USB performance instead. I'm not sure if the Pi's even support USB 3.0; sidehack can probably elaborate better on how he reaches a theoretical max. number of 15 chips.

[sidehack]

It's less a theoretical max and more that that's about what users have reported. It's around 15 to 18, but depends on the speed you're trying to run.

That miner showing 3 chips is a hardware issue. Talk to the vendor.

[GlenAllan]

They forgot to install the blue led fan when they sent it outa and sent it to me to swap. I did that yesterday and then later that day the 3 chips went out. My guess is i'll be responsible for the "damage". Likely just out $600 and now have a dud. I'll see what they say.

What would any damage look like visibly if i wanted to remove the heat sinks and take a look? Would it be obvious? Could changing a fan even do that?

[n0nce]

They forgot to install the blue led fan when they sent it outa and sent it to me to swap. I did that yesterday and then later that day the 3 chips went out. My guess is i'll be responsible for the "damage". Likely just out $600 and now have a dud. I'll see what they say.

What would any damage look like visibly if i wanted to remove the heat sinks and take a look? Would it be obvious? Could changing a fan even do that?

Not sure what you mean by 'led fan' but if your miner came without fan and you installed one yourself, that can't damage the device.
Integrated circuits can break with or without visible damage. But if you did nothing major (like soldering / modifying the PCB), you should be able to RMA the whole device.

[sidehack]

You musta bought from Merch, whose "brand" is strapping blinky lights on everything. Not sure anything with a fan swap would stop the string at 3 chips, so it's probably an actual flaw and repair/replace should be an option. Contact the seller.

[GlenAllan]

Right, thanks all. just wanted to get feedback on the chances i did something to mess it up. Thanks much. I've requested the return. Now i wait to see what they say.

Oh, and yes, it was Bitcoin Merch

[cygan]

It's less a theoretical max and more that that's about what users have reported. It's around 15 to 18, but depends on the speed you're trying to run.
✂

this is currently my configuration running on a rpi4 8gb ram:

Code:

(5s):6.092T (1m):6.697T (5m):6.727T (15m):6.719T (avg):6.753Th/s
...
0: GSF 10051796: BM1397:01+ 525.00MHz T:515 P:525 (24:12)   |  100% WU:100% | 313.8G / 313.1Gh/s WU: 4373.6/m
1: GSF 10051619: BM1397:01+ 525.00MHz T:525 P:525 (24:12)   |  100% WU:100% | 364.3G / 349.7Gh/s WU: 4885.0/m
2: GSF 10053617: BM1397:01+ 525.00MHz T:405 P:519 (24:12)   |  100% WU:100% | 299.1G / 323.5Gh/s WU: 4519.1/m
3: GSF 10053622: BM1397:01+ 525.00MHz T:525 P:525 (24:12)   |  100% WU:100% | 339.8G / 341.1Gh/s WU: 4765.6/m
4: GSF 10051622: BM1397:01+ 525.00MHz T:505 P:524 (24:12)   |  100% WU:100% | 308.9G / 309.3Gh/s WU: 4321.4/m
5: GSF 10053618: BM1397:01+ 525.00MHz T:515 P:525 (24:12)   |  100% WU:100% | 365.0G / 349.0Gh/s WU: 4876.0/m
6: GSF 10070009: BM1397:06+ 670.00MHz T:670 P:663 (3:2)     | 91.8% WU: 90% | 2.748T / 2.430Th/s WU:33958.5/m
7: GSF 10070003: BM1397:06+ 630.00MHz T:630 P:616 (3:2)     | 89.7% WU: 92% | 2.292T / 2.337Th/s WU:32648.6/m

and i mean that is a very strong performance that all the sticks/r909 miner here get out of themselves

[bobbyg198]

Hi Everyone,
Wonder if anyone has any ideas. Got mine a few weeks ago was hashing no problem. Came home today zero hashing. Fan is on, with blue light. Red light on front is on but zero hashing, no blinking white light and no activity on solopool??

[paid2]

Hi Everyone,
Wonder if anyone has any ideas. Got mine a few weeks ago was hashing no problem. Came home today zero hashing. Fan is on, with blue light. Red light on front is on but zero hashing, no blinking white light and no activity on solopool??

Did you tried to link it to an other server/computer ? Is the issue the same ?

Anything suspicious in cgminer logs ?

[n0nce]

Hi Everyone,
Wonder if anyone has any ideas. Got mine a few weeks ago was hashing no problem. Came home today zero hashing. Fan is on, with blue light. Red light on front is on but zero hashing, no blinking white light and no activity on solopool??

'Have you tried turning it off and on again?'

Honestly, I don't know about the different LED colors. Any input by sidehack or kano on this? I couldn't find anything in this thread and it may be useful for debugging, indeed.

[kano]

Well the white blinky light is mining, the red light is power, but no idea what the blue light is.

I heard someone was putting blue led lighted fans on the ones they were selling - so if the fan has stopped working, then that would be it.

We'll have to await sidehack's comments if it's not an extra blue fan led/fan problem.

[bobbyg198]

Tried turning on and off, no use  . Been running it on Pi3b will get a laptop and download everything and see if the problem is the same there. In the meantime, where would I find the log files to see if anything happened.

Thank you

[bobbyg198]

Well the white blinky light is mining, the red light is power, but no idea what the blue light is.

I heard someone was putting blue led lighted fans on the ones they were selling - so if the fan has stopped working, then that would be it.

We'll have to await sidehack's comments if it's not an extra blue fan led/fan problem.

[The blue leds are around the fan, the fan and the lights are both working though. So its not the Pi or the Pi usb port just plugged in a newpac and set it going, hashing no problem. So has to be something on the 909]

[bobbyg198]

Sorry new here, posted in wrong place.

The blue leds are around the fan, the fan and the lights are both working though.

So its not the Pi or the Pi usb port, just plugged in a newpac and set it going, hashing no problem. So has to be something on the 909

thanks all

[n0nce]

Sorry new here, posted in wrong place.

The blue leds are around the fan, the fan and the lights are both working though.

So its not the Pi or the Pi usb port, just plugged in a newpac and set it going, hashing no problem. So has to be something on the 909

thanks all

Okay, so just your fan is running, well, bad news: the fan always runs. Maybe cgminer just crashed?
Try restarting cgminer and posting an image of what's on screen.

[GlenAllan]

Just an update...Returned the R909 to Bitcoin Merch and they will be sending me back a new unit once they are back in stock. So customer service has been very good with them.

Anyone have any ideas when the next batch of 909s are actually going out?

[bobbyg198]

Hi @ n0nce, tried cg miner several times, works for newpac, get nothing for the 909
https://imgur.com/7E3JOiw

no sign of the 909

thanks

[n0nce]

Hi @ n0nce, tried cg miner several times, works for newpac, get nothing for the 909


no sign of the 909

thanks

So the device is not recognized at all. Is your USB on the Pi working fine? You can see attached devices with commands such as lsusb and plug in the miner and / or another USB device and check whether something changes.

Also do check whether the USB socket on the miner is in good condition and whether everything is plugged in and connected well on both devices.

[bobbyg198]

Yep usb on Pi works recognises newpac no problem. The mini usb port on 909 seems fine, everything connected and tight stumped

thanks for suggestions.

[n0nce]

Yep usb on Pi works recognises newpac no problem. The mini usb port on 909 seems fine, everything connected and tight stumped

thanks for suggestions.

Does it show the NewPac with lsusb meanwhile the R909 is missing?
What about usb-devices?

[NotFuzzyWarm]

@bobbyg198
Query: You have said nothing about the PSU needed for the R909. Unlike the stickminers it is not powered from the USB hub but instead is powered either via the barrel connector or preferably the PCIe plug. Um, it does have a 12v PSU good for at least 6amps powering it right?

[bobbyg198]

Hi correct power supply.
When I do usb-devices with it plugged in get below. So Pi is fine, Pi usb is fine, correct power supply tried a few usb cables which work (so not these). Power to unit is on, power to fan is on, just nothing from chips.

T:  Bus=01 Lev=02 Prnt=02 Port=03 Cnt=02 Dev#=  4 Spd=12  MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs=  1
P:  Vendor=0403 ProdID=6015 Rev=10.00
S:  Manufacturer=FTDI
S:  Product=FT232EX
C:  #Ifs= 1 Cfg#= 1 Atr=80 MxPwr=98mA
I:  If#=0x0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)


T:  Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#=  1 Spd=480 MxCh= 1
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=1d6b ProdID=0002 Rev=05.10
S:  Manufacturer=Linux 5.10.11-v7+ dwc_otg_hcd
S:  Product=DWC OTG Controller
S:  SerialNumber=3f980000.usb
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=0mA
I:  If#=0x0 Alt= 0 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=00 Driver=hub

T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=480 MxCh= 5
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=02 MxPS=64 #Cfgs=  1
P:  Vendor=0424 ProdID=9514 Rev=02.00
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=2mA
I:  If#=0x0 Alt= 1 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=02 Driver=hub

T:  Bus=01 Lev=02 Prnt=02 Port=00 Cnt=01 Dev#=  3 Spd=480 MxCh= 0
D:  Ver= 2.00 Cls=ff(vend.) Sub=00 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=0424 ProdID=ec00 Rev=02.00
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=2mA
I:  If#=0x0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=ff Driver=smsc95xx

T:  Bus=01 Lev=02 Prnt=02 Port=03 Cnt=02 Dev#=  4 Spd=12  MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs=  1
P:  Vendor=0403 ProdID=6015 Rev=10.00
S:  Manufacturer=FTDI
S:  Product=FT232EX
C:  #Ifs= 1 Cfg#= 1 Atr=80 MxPwr=98mA
I:  If#=0x0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)

[888bold888]

Hello,

I'm a beginner trying to get started with mining using my GekkoScience R909 that I have connected to my Raspberry Pi running Umbrel.

I need step by step help and would be very grateful if anyone can help me with this.

Thank you in advance!

888bold888

[os2sam]

I need step by step help and would be very grateful if anyone can help me with this.

Link to instructions and correct software are in the 2nd post of this thread.

[frank7301]

Had commented on compac f but found this forum (sorry for all the same people) couldn't find if any one else had "found 0 chips".
So does it mean it got burnt up? I was running it at 550MHz

[GlenAllan]

Welp, ran one of my R909s for a week at 600MHz and it shorted my power supply. Plugged everything in except it and got everything else running fine again. Tried turning down the voltage and plugged it in with a barrel jack power plug and it started making whining noises, weird lights that weren't bright enough and stuttered, and then magic smoke. Offhand i think it blew a cap.

Any ideas for a cause and suggestions for a fix?

Thanks!

[os2sam]

Welp, ... and then magic smoke. ...
Any ideas for a cause and suggestions for a fix?

I don't think you'll ever be able to get the magic smoke back where it was.

[NotFuzzyWarm]

Welp, ... and then magic smoke. ...
Any ideas for a cause and suggestions for a fix?

I don't think you'll ever be able to get the magic smoke back where it was.

Ja. It takes skilled technicians to tame it and bring it under control again... 
The repair/replace process starts with you contacting the reseller you got it from.

[sidehack]

Where on the board did the magic smoke come from?

[GlenAllan]

I can't say exactly since it is enclosed and the smoke poured out from inside, but i'd guess middle front ish. it's possible if i plug it back in it'll finish smoking and i could tell better then.

[GlenAllan]

Sorry, i guess people think of where the fan is as the front. Not that side. I keep the lights side facing me. So middle back.

Also, if this is the Bitcoin Merch one, which i think it is, then they won't honor any use above 500MHz when it comes to the warranty. I assume these are software serialized, so i can't return a different one, so i might just be out 600 bucks. And to be clear, i'm not saying i like to cheat to get returns/exchanges, i just don't necessarily agree with the 500MHz limit on use.

[GlenAllan]

I would like to know though, from other who have pushed the R909, has anyone else had anything similar happen? Should i now worry about pushing the other one i have? As far as silicon lottery goes the remaining one definitely ran/runs with a much higher WU, sometimes up to 100%. The one that went poof rarely got up to 90%, sometimes as low as the high 70s. I've seen others post results and it seems the R909s tend to run below 100% unlike the CompacFs i have which almost all run at 100% or just below.

[kano]

Most R909 wont run at 100% if you push the clock speed.
This is due to the low millisecond timing of getting work to them.
If you run it overclocked on linux you'll get the best results, windows = not so good, mac = worst

As for what's going on there, I have a whole post a few pages back, you should probably, read related to that ...
https://bitcointalk.org/index.php?topic=5423227.msg61850232#msg61850232
The API info will also tell you
https://kano.is/gekko.php#apu

My picture on the support web page shows the 'front' as you call it.
https://kano.is/gekko.php#vol
With a big black capacitor there in the middle.
You should probably refer to that picture if you want help from someone - but not me - I'm a software guy

[GlenAllan]

That post a few pages back was directed at me when i was trying to run all of this on a Mac. I moved to a Pi running Ubuntu Mate and everything has gone great. That makes sense about it not running at 100%. But does the fact that the one that went smokey was running so much lower give any clues to a problem? Otherwise yeah if anyone who understands the hardware better and has done overclocking can speak to if i shouldn't have been running at 600MHz is the issue, or if it's just that one that may have had a problem to begin with, that would be cool.

Being that i think that R909 was already an exchange for a 909 going down for other reasons and Bitcoinmerch has a warranty policy regarding running it above 500Mhz, i'm now hesitant to clock the other one back up to 600 anyway. That's why i'd like to clarify if 600 is actually any issue or in that one had some flaw to begin with. I've seen a few people in here talk about running it up to 700MHz and it self regulates down if there are issues. Also why i'm hoping someone might know what would have blown and if there is actually just a way to fix it. I don't think my reseller knows anything about repairs.

[sidehack]

If Merch wimps out on the warranty, it's a Merch problem. I'm interested in seeing this one since it's the first "magic smoke" failure report I've heard in the wild, so if nothing else, you can send it back to the factory for repair.

[GlenAllan]

Hmm, well i'd definitely be interested in sending it in for repair. I don't actually see where that would be easily with a Google search. Mind pointing me in the right direction? Or do you mean through "Merch" to the factory for a paid repair?

[GlenAllan]

Or i guess i should ask, would you be the one it's going back to? Either way would like to get it looked at. Would of course prefer it to be working .

[GlenAllan]

If Merch wimps out on the warranty, it's a Merch problem. I'm interested in seeing this one since it's the first "magic smoke" failure report I've heard in the wild, so if nothing else, you can send it back to the factory for repair.

Yeah, Bitcoin Merch is not going to help me. He claims factory fixes do not exist as well. Can you give me info on where to send it and cost?

[sidehack]

Yeah let's take it to PM

[GlenAllan]

Sounds good

[sidehack]

I'mma just leave this here

[altair_tech]

😱😱😱😱😱😱😱😱😱😱😱😱😱😱😱😱😱😱😱
Beast Mode!

Is there some liquid nitrogen involved here?

[sidehack]

It maaaaaaay be a custom job a little bit. Beefy fan on stock heatsinks though; 27 watts into each chip.

[cygan]

I'mma just leave this here
✂️

holy moly!!!!
i am curious how long the performance will hold...
how could you bring the r909 to such a performance - have you made any modifications?!

[gt_addict]

  Reveal your secrets sidehack. Was the upgraded fan a leafblower ? 

[maddhy]

Hello! I recently moved homes and today at my new place when I tried to run my R909, unfortunately, it shows 0 chips found. I guess something went wrong during shipping.

Anyway, I just dissembled everything. When it's connected to the power cable but without running the CGminer, Chip '1', '2' and '5' (which are on the same side) are hot while the other 3 chips are cold. I wonder in idle, should all chips be hot or cold? and what could possibly cause this ''half hot, half cold'' scenario? Many thanks!


Picture of the dissembled board: https://www.dropbox.com/s/ctkr7t50n6vdski/PXL_20230415_164621233.jpg?dl=0

[maddhy]

An update to my yesterday's post: The 3 chips were hot in idle only if both power and data cable are connected (without plugging in the data cable all chips are cold). After leaving it overnight, all chips are found again and working fine. All chips now in the idle state are cold, so I guess probably there was a short circuit that caused the 3 chips to be hot.

[maddhy]

Sorry, It's me again.

Behind the chips there are some tiny holes formed in square shape, I wonder what are they for and if it's okay to place a thermal pad on it?

https://www.dropbox.com/s/jgdl3tlb561b9hc/PXL_20230416_092804489.jpg?dl=0

[DukeTogo13]

Question I didn't find an answer to with a quick forum search.

I have a few r909 miners I typically run on solo.ckpool with a 7d average of 7.39t.  I mess around with unmineable a bit from time to time, but it doesn't seem to be reporting my r909 stats from cgminer correctly.  After an hour or so I max out around 2.5 - 2.75t.

Before I had the r909's I used some 2pac and compac F stick miners with cgminer on unmineable, and the stats jived between the two programs.

I have an Antminer R4, and it reports to unmineable perfectly fine after 20 minutes; everything is syncd up.  Maybe I need to let the r909's go overnight and see what they look like in the morning?

Is this a common problem, or am I doing something incorrectly?  Any suggestions, or another pool recommendation?  

Thanks.

[kano]

Well, since you are solo mining, use my pool.
It also has a history graph and the best info about your mining than any other pool.
... and discord support to easily get help
I designed and wrote the cgminer API and the R909 driver, so yeah if you are mining on my pool, and I can thus see what's happening, I can help.

Though it sounds like your miner is working fine and the problem is whatever unmineable is.
I have no idea what unmineable is, but if you are having a problem with them not working, you need to ask them about it.

[DukeTogo13]

Hey Kano, Thanks for the tip. 

Unfortunately until I started researching this issue I THOUGHT I had been using your pool.  I made the discovery that I wasn't last night when I was web searching for your cgminer / r909 setup page and stumbled across your pool page.  I did switch over, and everything seems to be working as it should. 


As for unmineable, I did put a support ticket in with them earlier in the day, but they are slow to respond.

Thanks again.

[Mining21BTC]

Hey, is there a problem with R909 (cgminer) and Nicehash? Every time I mine on Nicehash with R909, the cgminer instance stops after a few hours. Sometimes it shows 0.5 ghs, sometimes 1ths, but mostly 0ths on Nicehash. Cgminer is utilizing the full Ths power, but it seems to be blocked by Nicehash. Why does cgminer stop without an error message after a few hours? Even when I do 30% solo mining on CkPool and 70% on Nicehash, cgminer stops after some time. However, if I run cgminer at 100% on CkPool, it runs for days until I stop it. Is there any way to solve this problem? My S9 miner is accepted on Nicehash.

Im using this url:

stratum+tcp://sha256asicboost.auto.nicehash.com:9200

(Same without asicboost )

[kano]

Fuckhash expect you to put a massive security breach in your miner.
The master cgminer git obviously doesn't include this hack.

The hack allows the 'pool' you are mining to, to send you work from anyone else's account, or any sha256d scamcoin, at any time.
cgminer doesn't allow this for obvious reasons.

[paid2]

Hey, is there a problem with R909 (cgminer) and Nicehash? Every time I mine on Nicehash with R909, the cgminer instance stops after a few hours. Sometimes it shows 0.5 ghs, sometimes 1ths, but mostly 0ths on Nicehash. Cgminer is utilizing the full Ths power, but it seems to be blocked by Nicehash. Why does cgminer stop without an error message after a few hours? Even when I do 30% solo mining on CkPool and 70% on Nicehash, cgminer stops after some time. However, if I run cgminer at 100% on CkPool, it runs for days until I stop it. Is there any way to solve this problem? My S9 miner is accepted on Nicehash.

Yeah Nicehash uses their own fork of cgminer, because of what kano explained in his previous message.

Personally, I wouldn't recommend using a cgminer other than this one (the one created and maintained by kano), and suggest you think twice before using shitty version from Nicehash.

[groke]

hi everyone

I have a GekkoScience R909 problem I am having is the speed is  unstable when I use it with compact f seems to go up and down and it stays at around 68 to 70% speeds range from 1.155T 1.480TH and 1.731t up to 2.274th,  when I use the r909 on it`s own there  is no issue just does not work with compact help anyone encountered this before. running the miners on i9 pc os ubuntu


thanks to anyone who can help

Grone

[altair_tech]

I saw the same issue when trying to run 8 Compac f with two R909s.
This wasn't because of CPU overload, since CPU usage was only at 30% (12th gen core i5 with windows 10 PC).
This computer can run 50+ compac f and upto 9 R909 without a problem.

The devices run fine individually but the problem only occurs when running them together in the same cgminer window.

I tried running two seperate cgminer instances on the same computer, one each for R909 and compac f and that seems to fix the unstable hashrate issues.
I used

Code:

--gekko-r909-detect

&

Code:

--gekko-compacf-detect

to enable only one type of device in each cgminer window.
I'm not sure why this works, but I'd suggest trying this out.

[yettihead]

Hello! I recently moved homes and today at my new place when I tried to run my R909, unfortunately, it shows 0 chips found. I guess something went wrong during shipping.

Anyway, I just dissembled everything. When it's connected to the power cable but without running the CGminer, Chip '1', '2' and '5' (which are on the same side) are hot while the other 3 chips are cold. I wonder in idle, should all chips be hot or cold? and what could possibly cause this ''half hot, half cold'' scenario? Many thanks!


Picture of the dissembled board: https://www.dropbox.com/s/ctkr7t50n6vdski/PXL_20230415_164621233.jpg?dl=0

Did you ever find a solution for this?  One of my R909 is displaying the 0 chips found this morning.  It has been running fine since I purchased it at the beginning of this year.

[n0nce]

The miner is guaranteed to run at 1.5TH from under 100 DC watts at stock voltage and do it nearly silently. Better performance isn't guaranteed, but is probably attainable. Take for example the test unit I had running 1.8TH at 70 watts, or the above-photographed machine which pulled down almost 2.1TH from 88 watts (on stock voltage, no voltage tuning) for over a week:

I finally got around to doing some power testing on my unit and I'm really impressed. After having owned a Futurebit Apollo that operated at about 200W, I was looking forward to the R909's sub-100W rating.

But in practice, it's even lower!
I haven't fiddled a lot with voltages, and I'm getting about 1.88TH/s with <85W at the wall (including power for the Raspberry Pi)! I'm using an old Seasonic Gold rated PC power supply that's not even at peak efficiency, since it's rated for 450W.
Without cgminer running, my meter was reading about 12W, actually. So if you're running cgminer on some sort of homeserver that you have up & running 24/7 anyway, you can save another 10W.

For one R909, a ~120W PSU would probably be perfect efficiency-wise (they are made to peak at about 80% load).

This makes it really viable to even do lottery mining in central EU where we pay on average about 30 cents per kWh. Basically, this unit will cost you under 20€ in electricity per month. Or, of course, you can go with a pool as well. cgminer runs quite stable for me on a Raspberry Pi, connected to kanopool. I will write a small guide for that soon.

The EU average price in the second half of 2022 — a weighted average using the most recent (2022, semester 2) data for electricity by household consumers — was €0.2840 per KWh.

[aniplex]

how do you get the R909 to work? Ive tried on windows10 and windows 11

Device not detected at all. so its not connecting as an unreconize device it just not connecting at all. have tried multiple ports ect as well. It powers on so guessing the power is good, maybe usb cable? never heard of them being bad, dont think ive come across 1 my whole life that was bad lol.

[os2sam]

Did you install the WinUSB driver?

[aniplex]

Did you install the WinUSB driver?

are you refering to zadig to install winusb to the device? if so zadig doesnt detect the device as connected so no ive not installed it.

[os2sam]

Well, you have to get that done, it's not optional.

[os2sam]

Did you install the WinUSB driver?

are you refering to zadig to install winusb to the device? if so zadig doesnt detect the device as connected so no ive not installed it.

Yes, I'm referring to using Zadig to install the WinUSB driver.  So if the device isn't being recognized then you must have bad hardware somewhere, miner, PC, cable, Power Supply, ...

[aniplex]

Did you install the WinUSB driver?

are you refering to zadig to install winusb to the device? if so zadig doesnt detect the device as connected so no ive not installed it.

Yes, I'm referring to using Zadig to install the WinUSB driver.  So if the device isn't being recognized then you must have bad hardware somewhere, miner, PC, cable, Power Supply, ...

thanks, ive tried 2 different pcs so doubt thats its. it does power up fans and lights so not sure about power supply. it has 3 lights it looks like on the mb only themiddle lite comes on. looks white or pinkish. not sure if that helps. so probably a dud miner ill try another cable but never heard of these going bad unless extreme useage.

[os2sam]

Well, I guess after you try a new USB cable, you need to contact the dealer you bought it from.

[kano]

Most R909 wont run at 100% if you push the clock speed.

Spamming threads to get post counts ...

Anyway, no, they will get 100% is you push the clock speed ... up to a point ...

[bigdaddymccarron]

Got my first R909 today, love it!  Nice work Sidehack!

[saeshtoshi]

Hey plebs,

I am new to this forum. Found you guys while setting up my 2x R909. I am into Bitcoin since end of 2020. Made my first newbie umbrel node in 2021. So the next step is hobby mining. Especially with the side effect of using the heat.

So now I am Solo mining with Kano's pool. First of all: thank you Kano for keeping up the good work with the cgminer. I am using cgminer around a week now and I am really happy with it. Also thanks to Sidehack for the valuable posts I have read.

Now to my R909 double team:
The controller is a RPi 4 2GB Raspiblitz Bitcoin Full Node. Through a good 3A USB 3.0 HUB the Pi-4-controller is connected to the R909s (blue USB 3.0 port at RPi). The power supplies are two brand new 150W Jack devices.

The first R909 is a mf beast. Blue voltage controller knob was set to around 11 o'clock since the beginning. The beast gave me 2.6 TH/s at 650Mhz WU:100% STABLE. So I was happy as fuck to be so lucky in the silicon lottery. Then I ordered a second one and directly ran them together... but it was pretty slow. The second R909 had the voltage knob at 1 o'clock. So I tried to play with the voltage and the frequency without having a real plan. So at first my best setting was 2x R909 at 550 Mhz making around 4.2 TH/S. The higher I went it always turned out to be the same: the second R909 lost efficiency over the course of a few hours.

Then I was fed up and wanted to analyse both of them separately. So both blue knobs to 12 o'clcok and then increased frequency by 25 Mhz, starting at 450 Mhz. giving them around 20 min. Setting display zero 1 min after target frequency was hit.

Summary:

1st R909 (blue knob 12 o'clock)
550 Mhz: 2.237 TH/s avg - WU: 100% avg
575 Mhz: 2.285 TH/s avg - WU: 98% avg
600 Mhz: 2.415 TH/s avg - WU: 100% avg
625 Mhz: 2.480 TH/s avg - WU: 98% avg
650 Mhz: 2.600 TH/s avg - WU: 100% avg (sweet spot in my opinion - takes more energy but also a better heater   )

2nd R909 (blue knob 12 o'clock)
550 Mhz: 2.107 TH/s avg - WU: 100% avg
575 Mhz: 2.250 TH/s avg - WU: 98% avg
600 Mhz: 2.350 TH/s avg - WU: 97% avg
625 Mhz: starting good at 2.400 TH/s avg - WU: 95% avg...after 15 min it started slowing down...to after 20 min: 2.335 TH/s avg WU: 90% avg
so I put the blue voltage know a little bit more clockwise to around 1 o'clock:
625 Mhz 1 o'clock: 2.450 TH/s avg WU: 97% avg (sweet spot in my opinion)
650 Mhz 1 o'clock: starting at 2.400 TH/s avg - WU: 89% avg...after 15 min it started slowing down again....
so blue knob to 2 o'clock:
650 Mhz 2 o'clock: 2.200 TH/s avg WU: 80% avg seems to be the end for the second R909

Then i ran them TOGETHER again and tried 625 Mhz at first -> which started at nearly 4.9 TH/s and it seemed to be stable for an hour... just to then slow down again...  

In the end their Team-Sweet-Spot is at 600 Mhz at around 4.5 - 4.6 TH/s average stable... until now...

Code:

 cgminer version 4.12.1 - Started: [2023-11-26 18:01:39.897]
--------------------------------------------------------------------------------
 (5s):4.467T (1m):4.668T (5m):4.617T (15m):4.247T (avg):4.544Th/s
 A:2481500  R:2836  HW:1152  WU:63478.6/m
 Connected to de.kano.is diff 2.84K with stratum as user saeshtoshi.worker1
 Block: 183ad73b...  Diff:68T  Started: [19:04:13.928]  Best share: 1.8M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070440: BM1397:06+ 600.00MHz T:600 P:600 (3:2) | 98.4% WU: 96% | 2.175T / 2.328Th/s WU:32522.3/m
 1: GSF 10070465: BM1397:06+ 600.00MHz T:600 P:600 (3:2) | 93.3% WU: 92% | 2.323T / 2.215Th/s WU:30953.2/m
--------------------------------------------------------------------------------
 [2023-11-26 19:07:15.828] Accepted 13410b25 Diff 3.4K/2836 GSF 1
 [2023-11-26 19:07:28.152] Accepted 0d8c2aab Diff 4.84K/2836 GSF 0
 [2023-11-26 19:07:34.374] Accepted 097dc97d Diff 6.91K/2836 GSF 1
 [2023-11-26 19:07:34.744] Accepted 143e19b0 Diff 3.24K/2836 GSF 0
 [2023-11-26 19:07:34.789] Accepted 0298c0e4 Diff 25.2K/2836 GSF 1
 [2023-11-26 19:07:35.535] Accepted 0693942f Diff 9.96K/2836 GSF 0
 [2023-11-26 19:07:36.359] Accepted 019a96c4 Diff 40.9K/2836 GSF 0
 [2023-11-26 19:07:37.973] Accepted 11914371 Diff 3.73K/2836 GSF 1
 [2023-11-26 19:07:44.175] Accepted 13f5f16d Diff 3.28K/2836 GSF 0
 [2023-11-26 19:07:46.069] Accepted 0a3d4f40 Diff 6.4K/2836 GSF 0
 [2023-11-26 19:07:48.452] Accepted 0c0c42dc Diff 5.44K/2836 GSF 1
 [2023-11-26 19:07:49.034] Accepted 075a675d Diff 8.91K/2836 GSF 1
 [2023-11-26 19:07:50.243] Accepted 11f06a70 Diff 3.65K/2836 GSF 1
 [2023-11-26 19:07:51.585] Accepted 065a2789 Diff 10.3K/2836 GSF 1
 [2023-11-26 19:07:54.295] Accepted 0948fd15 Diff 7.06K/2836 GSF 1
 [2023-11-26 19:07:59.152] Accepted 0f5b798c Diff 4.27K/2836 GSF 1
 [2023-11-26 19:08:00.838] Accepted 04ba0042 Diff 13.9K/2836 GSF 1
 [2023-11-26 19:08:02.279] Accepted 116e2c2d Diff 3.76K/2836 GSF 1
 [2023-11-26 19:08:02.814] Accepted 05478f78 Diff 12.4K/2836 GSF 0
 [2023-11-26 19:08:07.202] Accepted 0a07960f Diff 6.53K/2836 GSF 1
 [2023-11-26 19:08:07.621] Accepted 0248dacc Diff 28.7K/2836 GSF 1
 [2023-11-26 19:08:07.862] Accepted 05090470 Diff 13K/2836 GSF 1
 [2023-11-26 19:08:10.585] Accepted 0d0a39b0 Diff 5.03K/2836 GSF 1
 [2023-11-26 19:08:12.486] Accepted 0438f2d0 Diff 15.5K/2836 GSF 0
 [2023-11-26 19:08:13.108] Accepted 027d5f71 Diff 26.3K/2836 GSF 0
 [2023-11-26 19:08:14.564] Accepted 09a48826 Diff 6.8K/2836 GSF 1
 [2023-11-26 19:08:17.767] Accepted 08a97b42 Diff 7.57K/2836 GSF 1
 [2023-11-26 19:08:21.517] Accepted 0fb666c3 Diff 4.17K/2836 GSF 0
 [2023-11-26 19:08:24.568] Accepted 0cc7180b Diff 5.13K/2836 GSF 1
 [2023-11-26 19:08:25.145] Accepted 34d0cab0 Diff 318K/2836 GSF 0
 [2023-11-26 19:08:25.739] Accepted 0ac487f6 Diff 6.09K/2836 GSF 1
 [2023-11-26 19:08:26.358] Accepted 935999d3 Diff 114K/2836 GSF 1
 [2023-11-26 19:08:26.800] Accepted 027b839b Diff 26.4K/2836 GSF 1
 [2023-11-26 19:08:29.419] Accepted 0c637093 Diff 5.29K/2836 GSF 0
 [2023-11-26 19:08:29.786] Accepted 08686446 Diff 7.79K/2836 GSF 1
 [2023-11-26 19:08:34.817] Accepted 0a816830 Diff 6.24K/2836 GSF 1
 [2023-11-26 19:08:34.868] Accepted 10c2577d Diff 3.91K/2836 GSF 0
 [2023-11-26 19:08:36.136] Accepted 0ada510d Diff 6.04K/2836 GSF 1
 [2023-11-26 19:08:37.012] Accepted 064d8629 Diff 10.4K/2836 GSF 0
 [2023-11-26 19:08:37.638] Accepted 01e0f8e1 Diff 34.9K/2836 GSF 1
 [2023-11-26 19:08:38.539] Accepted 0b0a39fa Diff 5.94K/2836 GSF 0
 [2023-11-26 19:08:42.006] Accepted 0116d7e0 Diff 60.2K/2836 GSF 1
 [2023-11-26 19:08:45.333] Accepted 040cbd68 Diff 16.2K/2836 GSF 0
 [2023-11-26 19:08:46.091] Accepted 1624b6c6 Diff 2.96K/2836 GSF 1
 [2023-11-26 19:08:55.951] Accepted 0aeb5d2e Diff 6K/2836 GSF 1

Can someone give me advice how to run them more efficient as a duo? I can see some people running them with different frequencies in one cgminer. Haven't found out how to do this yet. So I could try to run them both at their individual sweet spot.

Thanks and Cheers to all you guys!


edit:
ah okay I found how I change the frequency when already running.....the JAVA API scripts  

Code:

 cgminer version 4.12.1 - Started: [2023-11-26 23:23:01.825]
--------------------------------------------------------------------------------
 (5s):5.149T (1m):5.123T (5m):5.018T (15m):4.559T (avg):5.001Th/s
 A:2496324  R:0  HW:1600  WU:69866.8/m
 Connected to de.kano.is diff 3K with stratum as user saeshtoshi.worker1
 Block: ce961410...  Diff:68T  Started: [23:42:56.815]  Best share: 4.1M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070440: BM1397:06+ 650.00MHz T:650 P:650 (3:2) | 98.9% WU:100% | 2.315T / 2.596Th/s WU:36271.8/m
 1: GSF 10070465: BM1397:06+ 625.00MHz T:625 P:625 (3:2) | 95.8% WU: 95% | 2.551T / 2.404Th/s WU:33591.1/m
--------------------------------------------------------------------------------

Total electrical Power used from 2x R909, RPi 4, USB-HUB = 275 Watt  

hope this plays out to be stable  

any suggestions how to tune the second device... I am not sure where the WU should be "at least"..:?  

edit2:

Finally found the ultimate sweet spot which is super stable. Both R909 at 650 Mhz. First one > blue voltage knob at 12 o'clock. Second one > at 1 o'clock.

And the thing which makes the difference: An extra FAN before the 2 R909 Fans. Means my problem was a problem of the temperature, which let the second R909 lose it's power every time after about 3 to 4 hours mining. Now I ordered 2 Noctua NF-8 Fans with 2200 RPM.

Stable at 275 Watt (including RPi and USB Hub)

Code:

 cgminer version 4.12.1 - Started: [2023-11-27 15:15:00.297]
--------------------------------------------------------------------------------
 (5s):4.865T (1m):5.213T (5m):5.108T (15m):5.093T (avg):5.112Th/s
 A:22444755  R:68859  HW:16512  WU:71420.1/m
 Connected to de.kano.is diff 3.28K with stratum as user saeshtoshi1.worker1
 Block: 28df86e9...  Diff:68T  Started: [23:04:07.887]  Best share: 16.1M
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: GSF 10070440: BM1397:06+ 650.00MHz T:650 P:650 (3:2)  | 97.9% WU: 98% | 2.489T / 2.571Th/s WU:35919.7/m
 1: GSF 10070465: BM1397:06+ 650.00MHz T:650 P:650 (3:2)  | 95.8% WU: 97% | 2.558T / 2.541Th/s WU:35500.4/m
--------------------------------------------------------------------------------
 [2023-11-27 23:22:10.824] Accepted 0810133a Diff 8.13K/3279 GSF 0
 [2023-11-27 23:22:14.424] Accepted 12bd39df Diff 3.5K/3279 GSF 0
 [2023-11-27 23:22:21.899] Accepted 11a29930 Diff 3.72K/3279 GSF 1
 [2023-11-27 23:22:24.850] Accepted 0de8b367 Diff 4.71K/3279 GSF 0
 [2023-11-27 23:22:26.784] Accepted 0cebb3a1 Diff 5.07K/3279 GSF 1
 [2023-11-27 23:22:27.203] Accepted 13123833 Diff 3.44K/3279 GSF 1
 [2023-11-27 23:22:29.407] Accepted 0af65f39 Diff 5.98K/3279 GSF 1
 [2023-11-27 23:22:29.918] Accepted 052b283e Diff 12.7K/3279 GSF 0
 [2023-11-27 23:22:30.218] Accepted 098793ad Diff 6.88K/3279 GSF 0
 [2023-11-27 23:22:31.419] Accepted 09f3f076 Diff 1.69M/3279 GSF 0
 [2023-11-27 23:22:33.494] Accepted 0d55d9c8 Diff 4.91K/3279 GSF 1
 [2023-11-27 23:22:34.971] Accepted 077f0cbb Diff 8.74K/3279 GSF 1
 [2023-11-27 23:22:35.934] Accepted 0206c30b Diff 32.3K/3279 GSF 0
 [2023-11-27 23:22:37.615] Accepted 0bffe19f Diff 5.46K/3279 GSF 0
 [2023-11-27 23:22:38.179] Accepted 04831b55 Diff 14.5K/3279 GSF 1
 [2023-11-27 23:22:38.928] Accepted 0fd38575 Diff 4.14K/3279 GSF 0
 [2023-11-27 23:22:40.797] Accepted 053f0b64 Diff 12.5K/3279 GSF 0
 [2023-11-27 23:22:45.544] Accepted 064d4063 Diff 10.4K/3279 GSF 1
 [2023-11-27 23:22:45.744] Accepted 1370ec67 Diff 3.37K/3279 GSF 1
 [2023-11-27 23:22:48.566] Accepted 0410a067 Diff 16.1K/3279 GSF 0
 [2023-11-27 23:22:50.146] Accepted 10b385bc Diff 3.92K/3279 GSF 1
 [2023-11-27 23:22:51.431] Accepted 104b1eb9 Diff 4.02K/3279 GSF 1
 [2023-11-27 23:22:51.593] Accepted 0cad1bca Diff 5.17K/3279 GSF 0

Love it!  

[n0nce]

I just posted my in-depth review here, for anyone interested:

GekkoScience Terminus R909 Review
[...]

[Valnurat]

I'm sorry. I have been away from this site for a long time.   

I do hope if you can answer if this is still a good device istead of the Compact F.

Br

[RKG456]

I have a Technical question about the R909 wattage draw at the wall. Here is the history and the question is at the bottom.

In my main mining Rig shelf I have the following all plugged into a multiport powerstip / surge protector which then runs through a wattage meter before going into the Wall.  ALL of the following pulls consistenly 509w to 515w depending on if I run the R909s at 575mhz or 600mhz.

1 small monitor, keyboard and mouse shared via KVM.
1 HP Server powersupply with 6pin breakout board.  The R909s plug into this for power via 6pin connector.
5 small external usb fans for air flow
Mining Rig#1 - I have 3 R909s running at 575mhz and 4 Compac-Fs @460mhz with attached Bitcoin merch usb fans, mining BTC on a Raspberry Pi 4 and 2 powered hubs. The R909s into one 90w Sipolar hub and the 4 Compac-Fs into another 90w Bitcoinmerch hub. (they are identical except for branding)
Mining Rig#2 - 4 newpac's @350mhz and 2 Compac-Fs @450mhz all with attached Bitcoin merch usb fans, and one Artec bendable USB Fan.


On a separate wall area I have another rig running 3 Compac-Fs @450mhz and Bitcoinmerch usb fans and one R909 @575mhz. This one the R909 is running on a 120w power brick from Bitcoinmerch that came with my first R909.  I have the R909 going into a Watt meter on its own.  This one started out pulling 115w and slowly increased so far to 117w.  

Last week I purchased a 5th r909 used from ebay. It hashes fine and is giving no problems, but I noticed that this one will start out pulling 115w from the wall but will slowly keep creeping up. I let it hash for 2 hours and it had creeped up to 132w.  I found that odd given the power strip was supposed to be 120w. I grabbed another power brick i had that came with another R909 I got off of ebay and it too was supposed to be 120w, but may only be 100w (not clear on the back), but the same thing it creeped up to 125w in about an hour before I shut it down.

Then added this 5th R909 to Rig #2 on the main shelf at the top that pulls a consistent 509w - 515w normally and it jumped to 625w and started climbing.  It got to 660w after about 5 hours and again I powered that one down just to be safe.  

Any idea what would make the wattage use creep up like that?


EDIT: It looks like I may have solved the by backing off of the blue dial by 30% counter clockwise. The R909 wattage has been stable at approx 113.8 since.  It does fluctuate up or down one or two watts sometimes but nothing like it was. 


 

[meligo]

Do have a question to Kano, If i'm running a latency check it seems that your pool is based not in Europe so it's giving me a latency of 155ms, do you have a Europe based pool also?

[paid2]

Do have a question to Kano, If i'm running a latency check it seems that your pool is based not in Europe so it's giving me a latency of 155ms, do you have a Europe based pool also?

This topic is not dedicated to Kano pool.

Anyway, the pool propose a node located in EU: https://kano.is/index.php?k=support#setup

The stratum.kano.is mining nodes are in West-USA and randomly select LosAngeles or Phoenix,
that are less than 10ms apart.

There are other mining nodes located around the world that you may be closer to:
stratum+tcp://nya.kano.is:3333 (NewYork)
stratum+tcp://nl.kano.is:3333 (Netherlands)
stratum+tcp://de.kano.is:3333 (Germany)
stratum+tcp://jp.kano.is:3333 (Japan)
stratum+tcp://se.kano.is:3333 (Seattle)
stratum+tcp://la.kano.is:3333 (LosAngeles)
stratum+tcp://px.kano.is:3333 (Phoenix)
stratum+tcp://al.kano.is:3333 (Atlanta)

[NotFuzzyWarm]

Do have a question to Kano, If i'm running a latency check it seems that your pool is based not in Europe so it's giving me a latency of 155ms, do you have a Europe based pool also?

Um, Kano has a thread covering things regarding the pool... If you look at the 1st post you will see the list of all the worldwide nodes he runs for it. There are 2 for europe - one in Germany and one in the Netherlands.

BTW: 155ms latency is *not* bad. Not great either but not bad.

[meligo]

Do have a question to Kano, If i'm running a latency check it seems that your pool is based not in Europe so it's giving me a latency of 155ms, do you have a Europe based pool also?

Um, Kano has a thread covering things regarding the pool... If you look at the 1st post you will see the list of all the worldwide nodes he runs for it. There are 2 for europe - one in Germany and one in the Netherlands.

BTW: 155ms latency is *not* bad. Not great either but not bad.

Thanks, already spun up a raspberry pi 5 with NVMe for a local Bitcoin node :-)

Got the Smart Hub and R909 today..

Some metrics, it came like this out of the box, didn't change a setting...
hashrate1m   "2.42T"
hashrate5m   "2.13T"
hashrate1hr   "2.08T"

2nd R909 sadly got a broken USB connector , broke off while plugging in the cable..