GekkoScience BM1384 Project Development Discussion

[Bicknellski]

With regards to Hammer stickminers:

I got the datasheet for the Hammer ASIC.  It does not contain power info at 0.55V but indicates that the bottom end power efficiency is 0.55W/GH although that is also the rated power at 0.6V and 430MHz (this is at the typical corner).  This would be 3.5W which is too high, but even taking 0.55W/GH as your best power consumption estimate you could probably run off USB down at about 4GH or so including VRM losses.

Would anyone look at a Hammer stickminer (~2.5W, ~4GH) with interest or is that just too low of a hashrate?  You could clock it at double that easily (with sufficient USB power) assuming we build the power supply capable of handling that.  It would be a fair amount more work for us to do an additional board including serial protocol for the Hammer but it would keep us away from bitmain's (potential) price hikes.

Let me know if this is something you would consider worth picking up, the attraction for us is that we would likely be able to get some old Hammers a little cheaper than bitmain might sell us chips.  The downside is that the result would probably not compare particularly well to a BM1384 stickminer, which would not be much extra work for us to produce.

--
novak

don't make them.  the bm1384  would be much better.

having had sticks of all types having sold sticks of all types on ebay.  your specs mentioned on  the bm1384  were a lot better.

If the bm1384 dial under .4 watts/gh that is a big seller.

  the hammer at 2.5/4 = .625 w per gh  it is substandard efficiency .

If I run 4 sticks at .4 watts or less (BM1384)  Balance them  with a solo pool and a pair of pps pools Positive roi is possible due to the low watts per gh.

If I do it with 4 sticks at .625 watts (Hammer)  Positive roi is not going to happen.

USB's won't ROI. That is just dream city the USB sales price is not going be low enough. That threshold was already reached about 1 year ago when price of BTC and the network difficulty was far more reasonable than now to miners.

The point of the USB is novelty, educational with possibly some sort of secondary use built in.

Ideally you could solar power a few and it be a green miner the question would be price of the chip and cost of design. Both could be possible if you had two footprints for each chip.

Also a great proof of concept and allows people to see what sort of team you got here. Wish them luck.

[1714801531]

1714801531

[1714801531]

1714801531

[1714801531]

1714801531

[1714801531]

1714801531

[1714801531]

1714801531

[philipma1957]

With regards to Hammer stickminers:

I got the datasheet for the Hammer ASIC.  It does not contain power info at 0.55V but indicates that the bottom end power efficiency is 0.55W/GH although that is also the rated power at 0.6V and 430MHz (this is at the typical corner).  This would be 3.5W which is too high, but even taking 0.55W/GH as your best power consumption estimate you could probably run off USB down at about 4GH or so including VRM losses.

Would anyone look at a Hammer stickminer (~2.5W, ~4GH) with interest or is that just too low of a hashrate?  You could clock it at double that easily (with sufficient USB power) assuming we build the power supply capable of handling that.  It would be a fair amount more work for us to do an additional board including serial protocol for the Hammer but it would keep us away from bitmain's (potential) price hikes.

Let me know if this is something you would consider worth picking up, the attraction for us is that we would likely be able to get some old Hammers a little cheaper than bitmain might sell us chips.  The downside is that the result would probably not compare particularly well to a BM1384 stickminer, which would not be much extra work for us to produce.

--
novak

don't make them.  the bm1384  would be much better.

having had sticks of all types having sold sticks of all types on ebay.  your specs mentioned on  the bm1384  were a lot better.

If the bm1384 dial under .4 watts/gh that is a big seller.

  the hammer at 2.5/4 = .625 w per gh  it is substandard efficiency .

If I run 4 sticks at .4 watts or less (BM1384)  Balance them  with a solo pool and a pair of pps pools Positive roi is possible due to the low watts per gh.

If I do it with 4 sticks at .625 watts (Hammer)  Positive roi is not going to happen.

USB's won't ROI. That is just dream city the USB sales price is not going be low enough. That threshold was already reached about 1 year ago when price of BTC and the network difficulty was far more reasonable than now to miners.

The point of the USB is novelty, educational with possibly some sort of secondary use built in.

Ideally you could solar power a few and it be a green miner the question would be price of the chip and cost of design. Both could be possible if you had two footprints for each chip.

Also a great proof of concept and allows people to see what sort of team you got here. Wish them luck.

 I agree the usb stick do not make positive roi most of the time but  a .35 watt usb has a chance to do positive roi a .625 watt usb is not going to do it.

Since he wants the bigger miner to use bm1384 chips I only suggest the usb stick to do the same.

It is also easier for him to do this since he has decided to work with the bm1384 chips in the larger miner.

If given a choice of creating a working set of  pcbs   for 1 chip or for 2 different chips.  He is better off using just the one chip.

Especially since the one chip has better down clocked potential.

I see the asic world with 11 to 18 month cycles  for new gen not the quick 6 month cycles we have been doing.

So a usb stick with a bigger w/gh range is better.

Just like a miner with a bigger w/gh range is better.    the current s-5 is around .51watts per gh flat more or less.

if it had dial down dial up  watts per gh  at .35 to .55 it would be all the more valuable  and it would last longer.

In fact I believe that it was intentionally left out by bitmaintech so the miner would go obsolete quicker. (also to lower cost to build the s-5)

Asic builders want miners to go obsolete so they can sell better machines every six months = old thinking.

My opinion  for what it is worth.

Miners that have wider power range on down clock and up clock are all the more desirable for a guy mining = 11 to 18 month cycles = a more stable coin and network

 also my opinion for what it is worth.

[valkir]

I do agree with you Phil (again  )

I dont think the hammer chip is worth the time and energy.

You should focus on the BM1384 and get a better watt/gh miners.

Im pretty sure the BM1384 usb miner will get out of stock pretty quick.

Can wait to see the first proto board! 

[Bicknellski]

One advantage to the hammer design. Spondoolies support and more chips. Not sure what sort of availability there is for others. I'd say Bitmain would be the only other RELIABLE source of chips.

I would wager they would be available to help support Sidehack and Novak hardware or software related with anything might need as well something that no other supplier of chips would be able to do beyond a datasheet and a few emails back and forth. Our EE was able to ask questions and get some pretty good support although ultimately we failed. I suspect that Spondoolies will do everything it can to support projects like this given what I have seen before.

[philipma1957]

One advantage to the hammer design. Spondoolies support and more chips. Not sure what sort of availability there is for others. I'd say Bitmain would be the only other RELIABLE source of chips.

I would wager they would be available to help support Sidehack and Novak hardware or software related with anything might need as well something that no other supplier of chips would be able to do beyond a datasheet and a few emails back and forth. Our EE was able to ask questions and get some pretty good support although ultimately we failed. I suspect that Spondoolies will do everything it can to support projects like this given what I have seen before.

well I never knock good support.  If needed proper support is worth a lot.

But putting out a .625 watt per gh usb stick minimum power draw.

Against putting out a .35 watt to .55 watt per gh  usb stick via a trim pot.

  If I could build one over the other I want the lower watt stick.

Well lets see what sidehack and his team do.

[Biffa]

Why Hammer? Is Rockerbox too demanding?

Hammer:
Rated HashRate 7.5 GH/s per chip
Rated Voltage   0.63 V, recommended voltage range is 0.6 V - 0.8 V
Power Consumption   0.58 W/GH/s

Rockerbox:
Rated hash rate: 200 GH/s per chip
Rated Voltage   0.7 V, recommended voltage range is 0.63 V - 0.8 V
Power Consumption   0.34 W/GHs in typical corner

[ManeBjorn]

I kept 2 of the Beast Hubs as well.
They rock.  
 

One of the biggest issues with U2's came from the size of the heat sink. It was appropriately matched for the performance but became a big obstacle when plugging into a hub. So making them with a female end would allow the user to add short 6-7 inch usb leads to the hub would be nice. The other issue was power sourcing. Simple screwdown connectors would allow the user to quickly tap power from an ATX PSU but as we all know by now, most have shifted the output too 12volts the buck convertor would need to be o.k. with that over the longhaul. This setup would allow for the use of inexpensive unpowered hubs.

So what you're saying is, you don't want a stick miner. You want a USB-tethered desk miner that runs off 12V power. You should probably wait until the TypeZero board is up and going then.

Yeah Phil, you know we like hacking hardware so if the thing isn't software-voltable it needs to be hardware-voltable.

Good I kept 2 of these stud hubs so I am ready for some nice sticks.

[Biffa]

Why Hammer? Is Rockerbox too demanding?

Hammer:
Rated HashRate 7.5 GH/s per chip
Rated Voltage   0.63 V, recommended voltage range is 0.6 V - 0.8 V
Power Consumption   0.58 W/GH/s

Rockerbox:
Rated hash rate: 200 GH/s per chip
Rated Voltage   0.7 V, recommended voltage range is 0.63 V - 0.8 V
Power Consumption   0.34 W/GHs in typical corner

Doh.. your talking about USB sticks... 200 x 0.34W is too much

[cshelswell]

Usb sticks are still fun to play with alt coins. Everyone said I would ROI with the old block erupters but mining a few alts and I did. Not sure how much that happens now. Nonetheless USBs still sound like a funny hobby / entry level device to me.

[philipma1957]

Why Hammer? Is Rockerbox too demanding?

Hammer:
Rated HashRate 7.5 GH/s per chip
Rated Voltage   0.63 V, recommended voltage range is 0.6 V - 0.8 V
Power Consumption   0.58 W/GH/s

Rockerbox:
Rated hash rate: 200 GH/s per chip
Rated Voltage   0.7 V, recommended voltage range is 0.63 V - 0.8 V
Power Consumption   0.34 W/GHs in typical corner

Doh.. your talking about USB sticks... 200 x 0.34W is too much

Did you answer your own question? LOL

[TheRealSteve]

Why Hammer? Is Rockerbox too demanding?

Doh.. your talking about USB sticks... 200 x 0.34W is too much

Did you answer your own question? LOL

Did you?
( MUST GO DEEPER! )

Biffa: on top of power and subsequent cooling requirements*, there's availability, cost and overall headache of the package issues
* For the curious: there's 3 figures posted by SPT, and some frequencies from the info sheet.  Taking the lowest of those and extrapolating from the 3 figures gives a RockerBox running at 500MHz yielding 96.5Gh/s, between 0.22 and 0.28J/Gh (depending on extrapolation).  Still way too high, and who knows if that even runs stable.

[ElGabo]

S1 and S3 I believe are directly compatible but I could be wrong. We have S1 and Prisma to experiment on. I figured folks would have Prisma chassis running around from having stripped the boards for refund.

S1 have the same chassis like S3.

If you want an S3 I can send you one.

[novak@gekkoscience]

One advantage to the hammer design. Spondoolies support and more chips. Not sure what sort of availability there is for others. I'd say Bitmain would be the only other RELIABLE source of chips.

I would wager they would be available to help support Sidehack and Novak hardware or software related with anything might need as well something that no other supplier of chips would be able to do beyond a datasheet and a few emails back and forth. Our EE was able to ask questions and get some pretty good support although ultimately we failed. I suspect that Spondoolies will do everything it can to support projects like this given what I have seen before.

well I never knock good support.  If needed proper support is worth a lot.

But putting out a .625 watt per gh usb stick minimum power draw.

Against putting out a .35 watt to .55 watt per gh  usb stick via a trim pot.

  If I could build one over the other I want the lower watt stick.

Well lets see what sidehack and his team do.

I don't want to rule out a Hammer miner just because they would be less profitable- stickminers are some of the most fun miners to play with and still rarely break even.  At the same time, we'll have most of a BM1384 stickminer design done during our prototyping phase except for PCB layout, a BM1384 miner is pretty much double a Hammer miner, plus it would be far more adjustable.  If we do wind up having a BM1384 stickminer it would likely also be done earlier, which means we'd be coming back to making something half as powerful- probably not of that much interest to anyone.

It's going to depend somewhat on prices (of several things).  I'll keep you guys updated.  If spondoolies had other chips sized like the Hammer I would be busy jumping all over that but a huge chip like Rockerbox is not really what we're looking for on either of our designs despite being very efficient.

--
novak

[philipma1957]

I bookmarked this thread. Will keep an eye on it.

[valkir]

Quick question for Novak

This is the datasheet of BM1384


Will it be really possible to get 0.249 watt/gh ?

If so I will be ready to paid more to get a miner with that watt/gh rate.

I mean more chip but lower freq could give us some nice result.

Let said size of the S5 X 2 to get around 112 chip at 8.25 gh = 924 gh but only 230 watt 

I will be ready to buy right now!

[novak@gekkoscience]

Quick question for Novak

This is the datasheet of BM1384


Will it be really possible to get 0.249 watt/gh ?

If so I will be ready to paid more to get a miner with that watt/gh rate.

I mean more chip but lower freq could give us some nice result.

Let said size of the S5 X 2 to get around 112 chip at 8.25 gh = 924 gh but only 230 watt 

I will be ready to buy right now!

We will let you know after we actually get our dev board done.  As far as I know the only person who tried to run an S5 significantly undervolted was MrTeal and if I recall he had no luck getting it to start.  Although I can't say for certain without actually building a board that runs down at that efficiency, a preliminary look at the S5 indicates that this is a result of coms voltages being outside acceptable limits, not any problem with the hash engines.  I haven't seen any experimental results on efficiency, just the data sheet. 

--
novak

[philipma1957]

Quick question for Novak

This is the datasheet of BM1384


Will it be really possible to get 0.249 watt/gh ?

If so I will be ready to paid more to get a miner with that watt/gh rate.

I mean more chip but lower freq could give us some nice result.

Let said size of the S5 X 2 to get around 112 chip at 8.25 gh = 924 gh but only 230 watt 

I will be ready to buy right now!

We will let you know after we actually get our dev board done.  As far as I know the only person who tried to run an S5 significantly undervolted was MrTeal and if I recall he had no luck getting it to start.  Although I can't say for certain without actually building a board that runs down at that efficiency, a preliminary look at the S5 indicates that this is a result of coms voltages being outside acceptable limits, not any problem with the hash engines.  I haven't seen any experimental results on efficiency, just the data sheet. 

--
novak

  He got a 10 volt power source to work.  IIRC the data sheet said 9 volt psu can work but  I think 10 volts works.  That thread is buried  somewhere in hardware. 
 

[Gladimor]

I have received a message from a person impersonating SideHack. Be careful guys!

[Unacceptable]

Quick question for Novak

This is the datasheet of BM1384


Will it be really possible to get 0.249 watt/gh ?

If so I will be ready to paid more to get a miner with that watt/gh rate.

I mean more chip but lower freq could give us some nice result.

Let said size of the S5 X 2 to get around 112 chip at 8.25 gh = 924 gh but only 230 watt 

I will be ready to buy right now!

We will let you know after we actually get our dev board done.  As far as I know the only person who tried to run an S5 significantly undervolted was MrTeal and if I recall he had no luck getting it to start.  Although I can't say for certain without actually building a board that runs down at that efficiency, a preliminary look at the S5 indicates that this is a result of coms voltages being outside acceptable limits, not any problem with the hash engines.  I haven't seen any experimental results on efficiency, just the data sheet. 

--
novak

Holy crap,if you could get to that wattage,Bitmain may be VERY interested too 

Good luck guys!!!!! 

[MrTeal]

I'd have to check my notes, but I don't believe it would start at 10V. I had to start it at a higher voltage and then turn it down as it was running.
If I had to guess it's a limitation of the string design as opposed to the chip itself. The numbers Bitmain gave are likely accurate for those voltages, the S5 just won't get there. Hence why they removed the language in their thread that it could be run at 9V.