Bitcoin Forum
November 16, 2024, 03:05:45 PM *
News: Latest Bitcoin Core release: 28.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: What causes the change in TH/s ?  (Read 1591 times)
LeeC (OP)
Newbie
*
Offline Offline

Activity: 17
Merit: 0


View Profile
February 15, 2015, 01:58:52 PM
 #1

So it's not hard to confuse me about my SP20's right now but this is really don't understand.

Right now, all SP20's have their fans set to auto and i can see they are all pulling somewhat the same mining rate. Their temperatures aren't very different but over the last 5 minutes I have units reporting 1428, 1586 and 1865 . This is quite a difference and i see it reporting that rate to MRR too :

https://dl.dropboxusercontent.com/u/201234/gallery/smash1.jpg

https://dl.dropboxusercontent.com/u/201234/gallery/smash3.jpg

https://dl.dropboxusercontent.com/u/201234/gallery/smash2.jpg

The only thing i've changed of late was my pool , but this was done at MRR and not locally. I switched the 'when not mining' pool to CKPool From BTCGuide as a test.

Would this pool change have had any effect on the local hash rate of the machine (i don't think so) If not any ideas ?

philipma1957
Legendary
*
Online Online

Activity: 4312
Merit: 8855


'The right to privacy matters'


View Profile WWW
February 15, 2015, 02:07:08 PM
 #2

Btcguild   reads high for 1 or 2 hours then drops down.

Try to think in 12 hour blocks of time for your rates

▄▄███████▄▄
▄██████████████▄
▄██████████████████▄
▄████▀▀▀▀███▀▀▀▀█████▄
▄█████████████▄█▀████▄
███████████▄███████████
██████████▄█▀███████████
██████████▀████████████
▀█████▄█▀█████████████▀
▀████▄▄▄▄███▄▄▄▄████▀
▀██████████████████▀
▀███████████████▀
▀▀███████▀▀
.
 MΞTAWIN  THE FIRST WEB3 CASINO   
.
.. PLAY NOW ..
macgyver007
Member
**
Offline Offline

Activity: 105
Merit: 10


View Profile
February 15, 2015, 03:04:16 PM
Last edit: February 15, 2015, 03:22:01 PM by macgyver007
 #3

So it's not hard to confuse me about my SP20's right now but this is really don't understand.

Right now, all SP20's have their fans set to auto and i can see they are all pulling somewhat the same mining rate. Their temperatures aren't very different but over the last 5 minutes I have units reporting 1428, 1586 and 1865 . This is quite a difference and i see it reporting that rate to MRR too :


The only thing i've changed of late was my pool , but this was done at MRR and not locally. I switched the 'when not mining' pool to CKPool From BTCGuide as a test.

Would this pool change have had any effect on the local hash rate of the machine (i don't think so) If not any ideas ?


First off you can disregard the 1865 as there is no way any SP20 is hashing at that speed especially with 33C inlet temp
Easy fix here and your temps are WAY too high....all your chips are throttling for sure with the fan at 10%... what temperature is the room the miners are in?
Can you post a screenshot of the ASIC STATS please but I already know what I am going to see there  I just am curious about the numbers  Cool
You are using version 2.6.7 which is 1 of the earlier versions they added auto fan. I Strongly suggest to upgrade to 2.6.14 firmware for starters and set the upper voltage limit to .72 with auto fan and post your results please.. u should be getting ~1.5TH and 1050W with .72v cap compared to the 1200ish watts its using now for the same effect
Your temps will most likely not be in this range but the hashrate should be in check with the voltage
LeeC (OP)
Newbie
*
Offline Offline

Activity: 17
Merit: 0


View Profile
February 15, 2015, 03:34:05 PM
 #4

OK i've thrown in a bit of a red herring in the fan speed. The fans were at auto when i was getting those mega numbers and it sounded like a wind tunnel in here and was mighty warm.
I set them to 10 after sitting next to the machines to take my screenshots and compose my post.

As for firmware I am using 2.6.7 and there is auto fan in there (see screenshots) :

https://dl.dropboxusercontent.com/u/201234/gallery/Firmware.jpg

https://dl.dropboxusercontent.com/u/201234/gallery/Fan.jpg

Here's my Acis stat screen :

https://dl.dropboxusercontent.com/u/201234/gallery/acis.jpg

Tomorrow these are being moved into the cold room rather than my lounge, that should keep the temps down far more than this ambient room temp.

I'm also toying with spinning the fan around (test done) and sticking it in front of a portable AC unit thats in there running and seeing if having the fans run 10 on and it sucking in cold AC air can get some numbers crunched at a quieter level.

https://dl.dropboxusercontent.com/u/201234/gallery/fan_reverse.jpg

I was just trying to understand why when i'd not made any changed to the SP20 that the hash rate figures rose so high, after all the only thing i'd done was to change the pool it was mining.

I have asked ST direct via a support call if the hashing is linked to fan speed via voltage as if the fan speed being on nuts has the system draw more power and thus crunch the numbers higher then having a fan reversed sucking in cold AC air and the fans set low isn't going to do anything, surely then i'm just going to have a rig running at 1.4TH/s just really cooly.

Given what you've said about chips throttling due to heat, i'm actually hoping having these cool by the AC and sucking in the air that the hash rates will be even higher as the chips can continue to work harder without getting hotter. Either way we'll know tomorrow is ST don't reply today to that question via a ticket.
LeeC (OP)
Newbie
*
Offline Offline

Activity: 17
Merit: 0


View Profile
February 15, 2015, 03:40:20 PM
 #5

As an addendum to my last reply, it's important to note that i'm really happy so far with the SP20's and the hash rates they are putting out.
I'm just curious about finding how and why they tick and finding the balance between power draw and hashing rates.

Given the progress i made from advice in another thread running 2 SP20's from 2 PSU's and finding a happy medium that'll be the way forward, it's simple curiosity that's got me asking how and why these things work and what they could do with excessive cooling. Especially until i need to pay for the power! At that time, it'll be low and slow Smiley
macgyver007
Member
**
Offline Offline

Activity: 105
Merit: 10


View Profile
February 15, 2015, 04:04:22 PM
 #6

OK i've thrown in a bit of a red herring in the fan speed. The fans were at auto when i was getting those mega numbers and it sounded like a wind tunnel in here and was mighty warm.
I set them to 10 after sitting next to the machines to take my screenshots and compose my post.

As for firmware I am using 2.6.7 and there is auto fan in there (see screenshots) :

Here's my Acis stat screen :



Tomorrow these are being moved into the cold room rather than my lounge, that should keep the temps down far more than this ambient room temp.

I'm also toying with spinning the fan around (test done) and sticking it in front of a portable AC unit thats in there running and seeing if having the fans run 10 on and it sucking in cold AC air can get some numbers crunched at a quieter level.
I was just trying to understand why when i'd not made any changed to the SP20 that the hash rate figures rose so high, after all the only thing i'd done was to change the pool it was mining.

I have asked ST direct via a support call if the hashing is linked to fan speed via voltage as if the fan speed being on nuts has the system draw more power and thus crunch the numbers higher then having a fan reversed sucking in cold AC air and the fans set low isn't going to do anything, surely then i'm just going to have a rig running at 1.4TH/s just really cooly.

Given what you've said about chips throttling due to heat, i'm actually hoping having these cool by the AC and sucking in the air that the hash rates will be even higher as the chips can continue to work harder without getting hotter. Either way we'll know tomorrow is ST don't reply today to that question via a ticket.

You should re-read my previous post and pay attention to the screenshot...I already commented on your firmware version and what to do about that, and notice that no A/C is going to bring your inlet temp down to 10C as in the screenshot I posted where my miners are cooled by Canadian winter which beats any A/C (its -40 today but miners are in 5C environment) Tongue

As I thought, none of your chips are even using higher than .73v (actually only 1 is and its not very happy about it as you can see in YELLOW) and the rest are borderline throttling /at their temp max.
the fan setting of 10 should not be used with stock settings under any means, and that is why the default is NOT 10
Remove the red herring from your setup and post back your results afterwards...and use manual selection to choose firmware 2.6.14 as I mentioned earlier 2.6.7 is buggy.
I don't see how you are testing for excessive cooling when you are cooking your chips  Huh
I think paying closer attention to the fine details will do a lot in helping you..if Im not mistaken you ended up with the SP20 buying them thinking they mine scrypt coins?
LeeC (OP)
Newbie
*
Offline Offline

Activity: 17
Merit: 0


View Profile
February 15, 2015, 05:06:27 PM
 #7

Well that turned into something unexpected , but it all helps.

If your Canadian winters can't cool the units then it's saved me bothering to even try and test these tomorrow, i'll drop them in the room and put them back to auto.

Not so surprising is moving the fans back to auto after doing the firmware upgrade temperatures are down and voltage it up.

As for the mining so they were bought for bitcoin, it was my looking into something i should have left alone until i at least understood the basics of what these do.

Learning in public can and is appearing to be painful..
macgyver007
Member
**
Offline Offline

Activity: 105
Merit: 10


View Profile
February 15, 2015, 05:48:20 PM
Last edit: February 15, 2015, 06:21:11 PM by macgyver007
 #8

Well that turned into something unexpected , but it all helps.

If your Canadian winters can't cool the units then it's saved me bothering to even try and test these tomorrow, i'll drop them in the room and put them back to auto.

Not so surprising is moving the fans back to auto after doing the firmware upgrade temperatures are down and voltage it up.

As for the mining so they were bought for bitcoin, it was my looking into something i should have left alone until i at least understood the basics of what these do.

Learning in public can and is appearing to be painful..
I never said my units have a problem with cooling at all...I just mentioned that they are running in a very cold environment so you would understand why your inlet temp is 33C and mine is 10C and the resulting outlet temp differences....I am fortunate to have a cool location to house my miners naturally. You can run the miners and achieve similar results regardless of where you put them really at this voltage...any extreme cooling benefits would only result in lower fan speeds as the chips all adjust voltages
I will be posting a review of SP20 in the next week
As I said 2.6.7 is old test version so anything is possible. Keep in mind that even though you set the upper voltage limit to .75v (default) none of your chips are using even close to that in the screenshot you posted...that's why I suggested upper limit of .72v, that and also the extra 100 or so GH/s u get above that is around 1W/GH which is not worth it
Learning is something we all do every day but it helps to pay attention....would you buy a hot rod and floor it out of the dealership parking lot without putting it through all the gears and not expect to hit a wall or 2? Cheesy Cool
wh00per
Hero Member
*****
Offline Offline

Activity: 572
Merit: 500



View Profile
February 15, 2015, 06:41:19 PM
Last edit: February 15, 2015, 06:58:21 PM by wh00per
 #9


Right now, all SP20's have their fans set to auto and i can see they are all pulling somewhat the same mining rate. Their temperatures aren't very different but over the last 5 minutes I have units reporting 1428, 1586 and 1865 . This is quite a difference and i see it reporting that rate to MRR too [...]

The only thing i've changed of late was my pool , but this was done at MRR and not locally. I switched the 'when not mining' pool to CKPool From BTCGuide as a test. Would this pool change have had any effect on the local hash rate of the machine (i don't think so) If not any ideas ?

The hashing speed as seen by the pool is a function of the high difficulty (as set by the pool) shares solved in a given amount of time, with respect to the theoretical time in which a share of that difficulty would be solved in ideal conditions (i.e how long it takes to solve a share).

The hashing speed as seen by the mining unit is a function of the low difficulty shares solved by the unit determined with respect to the logged clock speeds, hardware errors etc.

This is why you see your SP20 determined hashrate at 1.5TH and on the pool "calculated" at 1.8TH. That means your miner was rather lucky over the interval monitored by the pool and solved a little more high difficulty shares during that time. Sometimes the pool speed is lower than  what your miner says, and that means you miner solved less high difficulty shares over the averaging time ..

Now, if you switched pools, different pools have different running averages .. in short term you might see fluctuations on the speed as detected by the pool. They cancel out over a large time span, if you keep your settings constant.


If you look at Burnside's calculator (http://ltc.kattare.com/calc.php) you can reverse the equations to tell you the hashrate as a function of time ..
There, they calculate the value of the shares based on the theoretical difficulty (block find speed).

Code:
                    fee        =  elem(coin + "_fee").value / 100.0,
                    gh_per_sec =  elem(coin + "_mhs").value / 1000000.0,
                    coin_usd   =  elem(coin + "_usd").value,
                    coin_btc   =  elem(coin + "_btc").value,
                    usd_kwh    =  elem(coin + "_usd_kwh").value,
                    watts      =  elem(coin + "_watts").value,
                    cost_day     =  (watts / 1000) * 24  * usd_kwh,

                    // Calculated Values
                    coin_per_share  =  (1 / diff) * (value - (value * fee)),
                    shares_per_day =  gh_per_sec * (1 / gh_per_share) * sec_per_day,
                    coin_per_day    =  coin_per_share * shares_per_day,
                    time_hour      =  ((gh_per_share * diff) / gh_per_sec) / 3600,
                    time_day       =  time_hour / 24,
                    usd_profit         =  (coin_per_day * coin_usd) - cost_day;
                    usd_gross         =  (coin_per_day * coin_usd);
                    btc_gross         =  (coin_per_day * coin_btc);



CSA/cUL Certified Power Distribution Panels - Basic, Switched, Metered. 1-3 phases. Up to 600V. NMC:N4F9qvHz11BHcc4nh1LCJFsrZhA1EWgVwj
klondike_bar
Legendary
*
Offline Offline

Activity: 2128
Merit: 1005

ASIC Wannabe


View Profile
February 16, 2015, 11:49:09 PM
 #10


As I thought, none of your chips are even using higher than .73v (actually only 1 is and its not very happy about it as you can see in YELLOW) and the rest are borderline throttling /at their temp max.
the fan setting of 10 should not be used with stock settings under any means, and that is why the default is NOT 10

+1. its obvious that all the chips are operating at the high end of their temperature range, which likely has an impact on the operating efficiency compared to the ideal 85-105C readings

you should use at least 25% fan for the settings you have

24" PCI-E cables with 16AWG wires and stripped ends - great for server PSU mods, best prices https://bitcointalk.org/index.php?topic=563461
No longer a wannabe - now an ASIC owner!
GigaBit
Hero Member
*****
Offline Offline

Activity: 1736
Merit: 513


Leading Crypto Sports Betting & Casino Platform


View Profile WWW
February 17, 2015, 06:51:31 PM
 #11

To give a real life example, picture your Bitcoin miner as a D10 bulldozer.

As you bulldoze, you accelerate, cruise and decelerate, when doing ripping work.

Picture your hashing power as the speed your miner is made to run at... for a bulldozer, it's about 1-5 mp/h, your bitcoin miner is 1TH/s.

When you start working, it takes time to get to optimal speed but you can never "cruise" forever or you'd fall off a cliff...

You also hit some harder spots than others and start track digging... both are things that will slow you down.

On the flip side, you can get soft spots that will make you go faster than what the miner (dozer) was made for. (Going downhill)

That's why you should #HashRate rather than #HashPower because you can never run at the same speed all the time.

Your hashpower should be consistent on an average basis; some tiny discrepancies with pools may exist.

On a 1TH/s you can expect around 200GH/s of fluctuation; Your internet speed is also a factor in some parts of the world; including myself.

I have to run on an LTE connection, believe it or not and I did hit bandwitdth caps with mining during peak hours.

Let's not forget the temperature, the cooler the miner (without freezing it) the steadier your hashrate should be.

If you are mining in a pool, you have to consider other people's mining situations; many miners use junk hardware or have huge temp fluctuations.

Both of those will effect what is showing from your pool side; during winter, one can tell since hashing power is a lot steadier =D

There's a lot more to consider but those are the "every miner should know" column.

..Stake.com..   ▄████████████████████████████████████▄
   ██ ▄▄▄▄▄▄▄▄▄▄            ▄▄▄▄▄▄▄▄▄▄ ██  ▄████▄
   ██ ▀▀▀▀▀▀▀▀▀▀ ██████████ ▀▀▀▀▀▀▀▀▀▀ ██  ██████
   ██ ██████████ ██      ██ ██████████ ██   ▀██▀
   ██ ██      ██ ██████  ██ ██      ██ ██    ██
   ██ ██████  ██ █████  ███ ██████  ██ ████▄ ██
   ██ █████  ███ ████  ████ █████  ███ ████████
   ██ ████  ████ ██████████ ████  ████ ████▀
   ██ ██████████ ▄▄▄▄▄▄▄▄▄▄ ██████████ ██
   ██            ▀▀▀▀▀▀▀▀▀▀            ██ 
   ▀█████████▀ ▄████████████▄ ▀█████████▀
  ▄▄▄▄▄▄▄▄▄▄▄▄███  ██  ██  ███▄▄▄▄▄▄▄▄▄▄▄▄
 ██████████████████████████████████████████
▄▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▄
█  ▄▀▄             █▀▀█▀▄▄
█  █▀█             █  ▐  ▐▌
█       ▄██▄       █  ▌  █
█     ▄██████▄     █  ▌ ▐▌
█    ██████████    █ ▐  █
█   ▐██████████▌   █ ▐ ▐▌
█    ▀▀██████▀▀    █ ▌ █
█     ▄▄▄██▄▄▄     █ ▌▐▌
█                  █▐ █
█                  █▐▐▌
█                  █▐█
▀▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▀█
▄▄█████████▄▄
▄██▀▀▀▀█████▀▀▀▀██▄
▄█▀       ▐█▌       ▀█▄
██         ▐█▌         ██
████▄     ▄█████▄     ▄████
████████▄███████████▄████████
███▀    █████████████    ▀███
██       ███████████       ██
▀█▄       █████████       ▄█▀
▀█▄    ▄██▀▀▀▀▀▀▀██▄  ▄▄▄█▀
▀███████         ███████▀
▀█████▄       ▄█████▀
▀▀▀███▄▄▄███▀▀▀
..PLAY NOW..
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!