Pages:
Author

Topic: Better efficiency coming to your S19 from Bitmain - page 2. (Read 853 times)

copper member
Activity: 76
Merit: 72
Bitmain released their new firmware for S19J Pro models earlier today.
I installed it on my S19J Pro 104T (has beaglebone board). Here's the side by side comparison of performance between the old FW and the two modes on new FW.


Standard power mode on new FW seems identical to old firmware.
The low power mode does reach the 25.5 W/Th that bitmain promised but lowers hashrate by about 20%.
It runs at fixed frequency,  there's no additional autotuning downtime. The miner boots right up and starts hashing.
Since the miner only pull about 8.7 amps on low power mode, people might be able to squeeze two miners on 20A circuits (I know it breaks the 80% rule, but also know that some people will still do it)
I also ran this test with AMLOGIC control board on the same miner and got identical results for all 3 columns.
If you have a AMLOGIC board on the miner, you don't need to replace it to be able to underclock now.

This new firmware isn't as customizable as Vnish or Braiins, but happy to see Bitmain provide additional options to miners.




Really nice data, thank you.

Do you have any data on dB before and after LPM? Seeing as the fans RPM go down significantly I'm wondering how much change in noise.

I did some sound measurements on S19J Pro 104T. These are measured on a smartphone, placed 3 ft away from the miner exhaust fan.
New Firmware Standard power mode - Avg Fan speed: 4800 rpm, 66 decibels
New Firmware Low Power mode - Avg Fan speed: 3750 rpm, 58 decibels
newbie
Activity: 25
Merit: 24
In theory this would be a win win if we hit wide scale adoption.   Drop in hash for efficiency netting a lower difficulty.   Oh wait miners would just add more miners.   Roll Eyes
newbie
Activity: 13
Merit: 7
Bitmain released their new firmware for S19J Pro models earlier today.
I installed it on my S19J Pro 104T (has beaglebone board). Here's the side by side comparison of performance between the old FW and the two modes on new FW.
https://i.imgur.com/HrCoD6H.jpg

Standard power mode on new FW seems identical to old firmware.
The low power mode does reach the 25.5 W/Th that bitmain promised but lowers hashrate by about 20%.
It runs at fixed frequency,  there's no additional autotuning downtime. The miner boots right up and starts hashing.
Since the miner only pull about 8.7 amps on low power mode, people might be able to squeeze two miners on 20A circuits (I know it breaks the 80% rule, but also know that some people will still do it)
I also ran this test with AMLOGIC control board on the same miner and got identical results for all 3 columns.
If you have a AMLOGIC board on the miner, you don't need to replace it to be able to underclock now.

This new firmware isn't as customizable as Vnish or Braiins, but happy to see Bitmain provide additional options to miners.




Really nice data, thank you.

Do you have any data on dB before and after LPM? Seeing as the fans RPM go down significantly I'm wondering how much change in noise.
legendary
Activity: 4102
Merit: 7765
'The right to privacy matters'
I wonder if this upgrade will block running other fw again... Someone did 20.9 J/T with Braiins OS+.

I put all the user-submitted BraiinsOS+ performance data on a chart.
This data is from bos_x19_reports.txt file that has been floating around in BraiinsOS Telegram.



Bitmain low power mode isn't any better than Braiins' efficiency performance for similar hash rates.
Still important for folks who don't want to tinker with aftermarket firmware.


.

 usually it is no better but for many under warranty they prefer it.

I have one for braiins and one for bitmain.

after comparing I will run some with each.

I have 8 s19's on hand. looking for 25-30 more.

this firmware will save me in the summer.

I will run 4ph in the winter say 38 at  100-110th power =  118kwatts
I will run 3.2ph in the summer say 38 at 80-90th. power =   82kwatts

I may avoid getting liquid cooling.
copper member
Activity: 76
Merit: 72
I wonder if this upgrade will block running other fw again... Someone did 20.9 J/T with Braiins OS+.

I put all the user-submitted BraiinsOS+ performance data on a chart.
This data is from bos_x19_reports.txt file that has been floating around in BraiinsOS Telegram.



Bitmain low power mode isn't any better than Braiins' efficiency performance for similar hash rates.
Still important for folks who don't want to tinker with aftermarket firmware.

legendary
Activity: 4102
Merit: 7765
'The right to privacy matters'
please note my 4 hour power test clocks me at 24.11 watts a th
copper member
Activity: 76
Merit: 72
I wonder if this upgrade will block running other fw again... Someone did 20.9 J/T with Braiins OS+.

I can confirm that this new firmware from Bitmain does not block BraiinsOS.
When I insert BraiinsOS+ SD card, the miner boot into BraiinsOS even with the new Bitmain firmware on the control board.
I tested this for both Beaglebone and Xilinx boards.
legendary
Activity: 1988
Merit: 1561
CLEAN non GPL infringing code made in Rust lang
Bitmain released their new firmware for S19J Pro models earlier today.
I installed it on my S19J Pro 104T (has beaglebone board). Here's the side by side comparison of performance between the old FW and the two modes on new FW.

Thank you for answering my original question in this thread, i would have preferred an answer here than twitter, but thanks for verifying.
legendary
Activity: 4102
Merit: 7765
'The right to privacy matters'
got good  water meter setup

78.30 kwatts and 2:23 pm doing 77-80th on low speed


will check watt use in an hour or so.



the fan comes today will set up tonight see how quiet.

may do a video .



no complaints on the firmware at this time.

I used this unit from kaboomracks.

it arrived on Friday new sealed cost 2000
member
Activity: 129
Merit: 16
I just fired up my 100th pro on low


the freq is 405 vs 525

fans are lower.

I am looking to do the following for sound control

one of these

https://www.ebay.com/itm/285066554125?var=586911646248


four of these
https://www.ebay.com/itm/203764567923

two of these
https://www.ebay.com/itm/304119146264



my 100th unit does 78th on low and 3000 rpm fans

I have not measured power use but should be under 1800 watts as 21.5 x 78 = 1677 watts

besides any other issue this feature allow me to drop close to 1000 watts per miner which comes in handy in the summer.


Edit now at 80th




So you plan to replace the oem fans with just 1 inline fan?

A legit question.

 the answer is I plan to test this to see if it works.

That fan says it can do:
  
Model........Size.........Frequency...Voltage......Current....Power.........Spe ed.............Noise........Air Flow CFM......Pressure
HF100......4 Inch.......60HZ........110V.........0.55A........60W..........2400r/min........<31dB.......200..................150 Pa
HF150......6 Inch.......60HZ........110V.........0.68A........75W..........2450r/min........<37dB.......395..................300 Pa

HF200.....8 Inch......60HZ.......110V.......1.73A.....190W.......2500r/min......<40dB.........735........380 Pa


as much as 735 cfm.

Now these four fans in place do 225+225 = 450 then the out flow do 225+225 which is 450

but when fans are in series they do not add to 450 + 450= 900 they are more likely to be in the 500-600 cfm when at top speed.

thus this 8 inch blower should be more than good enough set to top speed of 2500 rpm and doing 735cfm

I am hoping to set to 2000 rpm and maybe it works for high hash.

I am hoping to set to 1500 rpm and maybe it works for low hash.


very nice, you've given us an idea! we may do the same!

Let us know the results, we're only aiming for reduced noise, temps or watts is not an issue, only noise!

<40dB sounds awesome to be honest
legendary
Activity: 4102
Merit: 7765
'The right to privacy matters'
I just fired up my 100th pro on low


the freq is 405 vs 525

fans are lower.

I am looking to do the following for sound control

one of these

https://www.ebay.com/itm/285066554125?var=586911646248


four of these
https://www.ebay.com/itm/203764567923

two of these
https://www.ebay.com/itm/304119146264



my 100th unit does 78th on low and 3000 rpm fans

I have not measured power use but should be under 1800 watts as 21.5 x 78 = 1677 watts

besides any other issue this feature allow me to drop close to 1000 watts per miner which comes in handy in the summer.


Edit now at 80th




So you plan to replace the oem fans with just 1 inline fan?

A legit question.

 the answer is I plan to test this to see if it works.

That fan says it can do:
  
Model........Size.........Frequency...Voltage......Current....Power.........Spe ed.............Noise........Air Flow CFM......Pressure
HF100......4 Inch.......60HZ........110V.........0.55A........60W..........2400r/min........<31dB.......200..................150 Pa
HF150......6 Inch.......60HZ........110V.........0.68A........75W..........2450r/min........<37dB.......395..................300 Pa

HF200.....8 Inch......60HZ.......110V.......1.73A.....190W.......2500r/min......<40dB.........735........380 Pa


as much as 735 cfm.

Now these four fans in place do 225+225 = 450 then the out flow do 225+225 which is 450

but when fans are in series they do not add to 450 + 450= 900 they are more likely to be in the 500-600 cfm when at top speed.

thus this 8 inch blower should be more than good enough set to top speed of 2500 rpm and doing 735cfm

I am hoping to set to 2000 rpm and maybe it works for high hash.

I am hoping to set to 1500 rpm and maybe it works for low hash.
copper member
Activity: 76
Merit: 72
Bitmain released their new firmware for S19J Pro models earlier today.
I installed it on my S19J Pro 104T (has beaglebone board). Here's the side by side comparison of performance between the old FW and the two modes on new FW.


Standard power mode on new FW seems identical to old firmware.
The low power mode does reach the 25.5 W/Th that bitmain promised but lowers hashrate by about 20%.
It runs at fixed frequency,  there's no additional autotuning downtime. The miner boots right up and starts hashing.
Since the miner only pull about 8.7 amps on low power mode, people might be able to squeeze two miners on 20A circuits (I know it breaks the 80% rule, but also know that some people will still do it)
I also ran this test with AMLOGIC control board on the same miner and got identical results for all 3 columns.
If you have a AMLOGIC board on the miner, you don't need to replace it to be able to underclock now.

This new firmware isn't as customizable as Vnish or Braiins, but happy to see Bitmain provide additional options to miners.


sr. member
Activity: 801
Merit: 293
Created AutoTune to saved the planet! ~USA
@Artemis3 I think we can all agree that your firmware provides a significant advantage over OEM currently in regards to performance. I can see how you can and should boast to say hey, we did it first and even better. I get it. Not discounting your contribution.

There is never a need to convince anyone of anything, I agree fully. The intention of my post was not to insult you or anyone else. I was simply stating that by running the new oem firmware it may offer a similar advantage to non-oem firmware which in fact may cut into the financials of pools and companies like yours. Not personal only my assessment based on logic.
I'm only looking at it from the business perspective and how I perceive the fiscal impact (good and bad) that may result from this new firmware release by Bitmain.

Actually we did it first and better. We invented auto-tune Smiley
Also yes this will 100% effect their profits and ours but with our new update we are not worried

Well, also, important point to remember, that when you use 3rd party firmware, they are taking some % of your hash rate also.

So if you do such measurements above as altair_tech has done, you have to remove the % off your TH performance numbers,
since some at least don't do the fee mining all at the same time as pool mining, but rather take the fee mining away at various times.

This is why we provide a widget that realtime calculates the fee and doesn't use encrypted traffic to hide the work as they could possibly be stealing hash. Also the reason you said is why you need to do 1 hour or 24 hour average for hashrate. 15min will not be accurate and usually under report.

Now back onto the topic

I love how no one believed me in telegram that it would reduce hashrate. We have had this firmware for months now. We have an insider at a bitmain mining facility that has high power costs and it is what they were running.

That being said since braiins is shilling their shit I am about to do a mic drop Smiley

This is from a client testing our autotune in Air in Canada.

110 TH 2,475 watts

22.5 watts per TH

Very high performing S19 assumed to be a unicorn at this point as it is new record holder for best efficiency at stock/stock+small oc range thus far.


sr. member
Activity: 801
Merit: 293
Created AutoTune to saved the planet! ~USA
[
@Artemis3 I think we can all agree that your firmware provides a significant advantage over OEM currently in regards to performance. I can see how you can and should boast to say hey, we did it first and even better. I get it. Not discounting your contribution.

There is never a need to convince anyone of anything, I agree fully. The intention of my post was not to insult you or anyone else. I was simply stating that by running the new oem firmware it may offer a similar advantage to non-oem firmware which in fact may cut into the financials of pools and companies like yours. Not personal only my assessment based on logic.
I'm only looking at it from the business perspective and how I perceive the fiscal impact (good and bad) that may result from this new firmware release by Bitmain.

Actually we did it first and better. We invented auto-tune Smiley
Also yes this will 100% effect their profits and ours but with our new update we are not worried

Now back onto the topic

I love how no one believed me in telegram that it would reduce hashrate. We have had this firmware for months now. We have an insider at a bitmain mining facility that has high power costs and it is what they were running.

That being said since braiins is shilling their shit I am about to do a mic drop Smiley

This is from a client testing our autotune in Air in Canada.

110 TH 2,475 watts

22.5 watts per TH

Very high performing S19 assumed to be a unicorn at this point as it is new record holder for best efficiency at stock/stock+small oc range thus far.

member
Activity: 129
Merit: 16
I just fired up my 100th pro on low


the freq is 405 vs 525

fans are lower.

I am looking to do the following for sound control

one of these

https://www.ebay.com/itm/285066554125?var=586911646248


four of these
https://www.ebay.com/itm/203764567923

two of these
https://www.ebay.com/itm/304119146264



my 100th unit does 78th on low and 3000 rpm fans

I have not measured power use but should be under 1800 watts as 21.5 x 78 = 1677 watts

besides any other issue this feature allow me to drop close to 1000 watts per miner which comes in handy in the summer.


Edit now at 80th




So you plan to replace the oem fans with just 1 inline fan?
legendary
Activity: 4102
Merit: 7765
'The right to privacy matters'
I just fired up my 100th pro on low


the freq is 405 vs 525

fans are lower.

I am looking to do the following for sound control

one of these

https://www.ebay.com/itm/285066554125?var=586911646248


four of these
https://www.ebay.com/itm/203764567923

two of these
https://www.ebay.com/itm/304119146264



my 100th unit does 78th on low and 3000 rpm fans

I have not measured power use but should be under 1800 watts as 21.5 x 78 = 1677 watts

besides any other issue this feature allow me to drop close to 1000 watts per miner which comes in handy in the summer.


Edit now at 80th


member
Activity: 129
Merit: 16
Could you share details about RPM and dB on this 20.9J/T? Would it work out the box with an s19j pro 104?

I've seen the bashing posts and threads this wasn't meant to be one of them and apologies if that was the way you read it.

Not trying to disrespect you or anyone else here on this forum as I understand the significant efforts and strides you and many others have made for the good of the community.

I'm only looking at it from the business perspective and how I perceive the fiscal impact (good and bad) that may result from this new firmware release by Bitmain.

No problem, none taken. And Kano is also correct, the "Real Hashrate" as reported in the miner UI is not what reaches the pool but 97.5% of that, so it does make perfect sense to balance the features. Doing this correction, Sadnudbi's 20.88 W/TH becomes 21.41 W/TH.

Has anyone achieved quiet or near silent operation with braiins? say 1000 RPM?
legendary
Activity: 1988
Merit: 1561
CLEAN non GPL infringing code made in Rust lang
Could you share details about RPM and dB on this 20.9J/T? Would it work out the box with an s19j pro 104?

I've seen the bashing posts and threads this wasn't meant to be one of them and apologies if that was the way you read it.

Not trying to disrespect you or anyone else here on this forum as I understand the significant efforts and strides you and many others have made for the good of the community.

I'm only looking at it from the business perspective and how I perceive the fiscal impact (good and bad) that may result from this new firmware release by Bitmain.

No problem, none taken. And Kano is also correct, the "Real Hashrate" as reported in the miner UI is not what reaches the pool but 97.5% of that, so it does make perfect sense to balance the features. Doing this correction, Sadnudbi's 20.88 W/TH becomes 21.41 W/TH.
member
Activity: 129
Merit: 16
Temps dropped by a degree or two, but fan rpm dropped from ~3800 rpm to ~2800 rpm with new low-power firmware.
My wattage measurements are at the PDU, so quite accurate.



I did some rough calculations and the new low power mode is more profitable for power rate >$0.06/kwh


Interesting, thanks!


I wonder if this upgrade will block running other fw again... Someone did 20.9 J/T with Braiins OS+.

Could you share details about RPM and dB on this 20.9J/T? Would it work out the box with an s19j pro 104?
legendary
Activity: 3304
Merit: 1842
Curmudgeonly hardware guy
Probably dropped the core voltage by a few percent to do that. Decreasing chip temps would be a compounding effect, and then shaving off a few more watts with less fan.
Pages:
Jump to: