Team Black Miner (ETHW ETC Vertcoin Ravencoin Zilliqa +dual +tripple mining ) - page 20.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: Riptide_NVN on February 09, 2022, 09:11:24 AM

Is there a version coming out that will set intensity automatically so as to avoid having to manually tune?

The intensity setting is pool depended. We can case on the pool and apply optimized defaults. Then you typically will get a few MHASH more on some pools, and less on others, and the difference between reported hashrate and poolhashrate will decrease.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: aesma on February 09, 2022, 09:00:16 AM

Whether I set power limit to 85% or 100% doesn't seem to make a difference, in either case the GPU clocks go down over time. After a night my 3090 was at 930MHz. I restarted windows, reset MSI afterburner, and now it's slowly going down again. Power is also going down and MHs aren't too bad (110) but I'd like to understand what is causing this. Since overclocking is "+XMHz" there is no point in overclocking if the base clock goes down and down and down. Temperatures are good, in fact I had to close my window as the rig couldn't make enough heat to compensate the slight opening.
For the 3080Ti, autotune is happening, so basically it's not mining from my understanding, 3 shares done during the same time the 3090 did 39. I will read up on what it does, I'm guessing once it has run for some time I should take note of the value reached and add it to the script so it doesn't autotune again next time.

Your card is trottleing, and probobly because the temperature of your memory. Download gpuz and you can see the perfcap reason. Lower the tdp will avoid heat to be transfered to the memory and extend the lifeline of your card.

The LHR unlock doesn't work on high power, you need to reduce to f.ex 70%. The autotune in 1.54 has a bug, but a new 1.55 will be released soon.

in msi afterburner, increase the temp limit, reduce the tdp like in the picture. Coreclock doesn't need to be high as well

Riptide_NVN

member

Activity: 294

Merit: 16

Is there a version coming out that will set intensity automatically so as to avoid having to manually tune?

aesma

hero member

Activity: 2548

Merit: 950

fly or die

Keep in mind I'm a noob here (well I mined LTC for a year in 2013 with R9 290s). Although I overclocked CPUs quite a bit in my days, but things were simpler then, no power limits, you just upped voltage, upped cooling, and killed components from time to time, sometimes with smoke and lights.

Whether I set power limit to 85% or 100% doesn't seem to make a difference, in either case the GPU clocks go down over time. After a night my 3090 was at 930MHz. I restarted windows, reset MSI afterburner, and now it's slowly going down again. Power is also going down and MHs aren't too bad (110) but I'd like to understand what is causing this. Since overclocking is "+XMHz" there is no point in overclocking if the base clock goes down and down and down. Temperatures are good, in fact I had to close my window as the rig couldn't make enough heat to compensate the slight opening.

For the 3080Ti, autotune is happening, so basically it's not mining from my understanding, 3 shares done during the same time the 3090 did 39. I will read up on what it does, I'm guessing once it has run for some time I should take note of the value reached and add it to the script so it doesn't autotune again next time.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: Dr_Victor on February 08, 2022, 01:48:16 PM

raveOS + 11.5

but what is the excact config. Are you using the workername in lowercase only?

konanlux

newbie

Activity: 10

Merit: 0

Quote

If you give me some of that hashrate I'll throw you a couple of hints! :-)

But because I'm a nice guy here's a couple quick ones for free:
  Increase your clock to 1600 ~ 1750, it's the only way to mess around with voltages on Linux
  Don't use Cuda 1.15 on linux, for whatever reason it performs much worse than 1.14 -- but in windows 1.15 works much better apparently.
  Lower your intensity a bit, try 3800

there is no place for a lot of heat in the room, 3 x6 rtx 3090 rigs and another x7 rtx 3070 non lhr, thanks for the advice, I will try intensity 3800

Dr_Victor

sr. member

Activity: 954

Merit: 250

Quote from: sp_ on February 08, 2022, 12:46:20 PM

Quote from: Dr_Victor on February 08, 2022, 12:16:48 PM

It does not work with Ethermine - errors:Pool connection disconnected or timeout. Error: Pool reponded(!) with error.

What's your excact launch config? Windows/linux? 11.4 or 11.5?

raveOS + 11.5

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: Dr_Victor on February 08, 2022, 12:16:48 PM

It does not work with Ethermine - errors:Pool connection disconnected or timeout. Error: Pool reponded(!) with error.

What's your excact launch config? Windows/linux? 11.4 or 11.5?

KillrBee

jr. member

Activity: 139

Merit: 3

Quote from: konanlux on February 08, 2022, 11:29:30 AM

Quote from: sp_ on February 07, 2022, 02:11:39 PM

Quote from: KillrBee on February 07, 2022, 01:46:00 PM

Second -- Check out that 3090 hashrate for almost 6 hours of mining!

Nice hashrate. I think the 2060S needs to be clocked lower because you have some rejected shares there. It's the cuda code that is crashing, not the AMD code. I think the 2060S is clocked to high. try 7900 or 7800 memclock.

I could perhaps include a GPU # in the error message that will track wich of the gpu's that crashed the rig.

any advice to improve is welcome

and this rig run v1.50 hiveos, only set --xintensity 4096

whats its the diference v1.54 an v1.54cuda115 ?

If you give me some of that hashrate I'll throw you a couple of hints! :-)

But because I'm a nice guy here's a couple quick ones for free:
  Increase your clock to 1600 ~ 1750, it's the only way to mess around with voltages on Linux
  Don't use Cuda 1.15 on linux, for whatever reason it performs much worse than 1.14 -- but in windows 1.15 works much better apparently.
  Lower your intensity a bit, try 3800

Dr_Victor

sr. member

Activity: 954

Merit: 250

Quote from: sp_ on February 07, 2022, 05:08:00 AM

v1.54
1. If the lhr autotune fails after 50 adjustments, select the best tuning for the rest of the period.
2. Ethermine and flexpool over to a different stratum implementation.

https://github.com/sp-hash/TeamBlackMiner/releases/

TeamBlackMiner_1_54_Ubuntu_18_04_Cuda_11_4.tar.xz
https://www.virustotal.com/gui/file/9f87fa90f5ad1b15f27655dc8accfa80d09d820e56140cfc217f3b0da61dd633?nocache=1

TeamBlackMiner_1_54_Ubuntu_18_04_Cuda_11_5.tar.xz
https://www.virustotal.com/gui/file/25a5c47e825915908af0d333955ba8bc56c05ac8e5b5355df7afe27e47afe592?nocache=1

TeamBlackMiner_1_54_cuda_11_4.7z
https://www.virustotal.com/gui/file/210d13919351303e309150a621526b01157155882a9c3550e692e01e75cab132?nocache=1

TeamBlackMiner_1_54_cuda_11_4.7z
https://www.virustotal.com/gui/file/16cdc2d3b826371303fec26ca35be16f3d33b8c076c4d4c52ba82dfefa2136eb?nocache=1

-----

Quote from: Dr_Victor on February 07, 2022, 04:21:02 AM

It does not work with Ethermine - errors:Pool connection disconnected or timeout. Error: Pool reponded(!) with error.

v1.54 should have a fix on ethermine and flexpool with slightly different stratum implementation.
Make sure that you mine with less than 5% stale shares on these pools, if not you can be thrown off.

Still errors on Ethermine Sad

konanlux

newbie

Activity: 10

Merit: 0

Quote from: sp_ on February 07, 2022, 02:11:39 PM

Quote from: KillrBee on February 07, 2022, 01:46:00 PM

https://i.imgur.com/2mLsNVr.png

Second -- Check out that 3090 hashrate for almost 6 hours of mining!

Nice hashrate. I think the 2060S needs to be clocked lower because you have some rejected shares there. It's the cuda code that is crashing, not the AMD code. I think the 2060S is clocked to high. try 7900 or 7800 memclock.

I could perhaps include a GPU # in the error message that will track wich of the gpu's that crashed the rig.

any advice to improve is welcome

https://i.imgur.com/EdIVyq5.jpg

https://i.imgur.com/bLHumxq.jpg

and this rig run v1.50 hiveos, only set --xintensity 4096
https://i.imgur.com/oNAr6Wo.jpg
whats its the diference v1.54 an v1.54cuda115 ?

KillrBee

jr. member

Activity: 139

Merit: 3

Quote from: Alireza1773 on February 08, 2022, 07:27:32 AM

Quote from: sp_ on February 08, 2022, 02:05:55 AM

Quote from: Alireza1773 on February 07, 2022, 06:30:42 PM

Hello guys
I have 6 amd rx 580 and 5 rx6600xt
os: hiveos
When disable rx 580 gpu's miner doesnt work and show this error:
source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE
What should i do?

Run tbminer --list-devices to see the device order and use -Y [5,6,7,8,9] parameter to select devices.

I do this but the miner doesn't start and this error still shows: source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE

What version of HiveOS are you using? the 6600 cards need newer drivers / linux kernel to be recognized. Make sure you've updated HiveOS to 0.6-213

Alireza1773

newbie

Activity: 2

Merit: 0

Quote from: sp_ on February 08, 2022, 02:05:55 AM

Quote from: Alireza1773 on February 07, 2022, 06:30:42 PM

Hello guys
I have 6 amd rx 580 and 5 rx6600xt
os: hiveos
When disable rx 580 gpu's miner doesnt work and show this error:
source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE
What should i do?

Run tbminer --list-devices to see the device order and use -Y [5,6,7,8,9] parameter to select devices.

I do this but the miner doesn't start and this error still shows: source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: Alireza1773 on February 07, 2022, 06:30:42 PM

Hello guys
I have 6 amd rx 580 and 5 rx6600xt
os: hiveos
When disable rx 580 gpu's miner doesnt work and show this error:
source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE
What should i do?

Run tbminer --list-devices to see the device order and use -Y [5,6,7,8,9] parameter to select devices.

Quote from: long2905 on February 07, 2022, 09:44:37 PM

whats the current lhr performance for 3080ti, 3080, 3070ti and 3060 lhr card? any plan to implement dual mining or features to compete with others like trex and lolminer?

I suppose the dream of high hashrate for 6900xt cards are no more

511.65 driver:

3080ti around 80MHASH (low power)
3070/3060ti around 45MHASH (low power)
3060 around 35MHASH (low power)

The latest 3080 has not yet been unlocked by anyone but you can test the latest autotune function in v1.54 together with --tweak to see if it unlocks. sometimes it helps to downgrade the driver because NVIDIA is
improving the lock with software in the latest drivers.

The 6900xt can do 69MHASH poolside with TBM, which is a good result with the 256bit bus.

Dual mining... Toncoin will go pos in a month, so a waste of time. Alephium is not listed on any major exchanges yet, so putting that on hold.

Remember that dual mining is using much more power, and limit the lifetime of you graphic cards.

long2905

newbie

Activity: 21

Merit: 0

whats the current lhr performance for 3080ti, 3080, 3070ti and 3060 lhr card? any plan to implement dual mining or features to compete with others like trex and lolminer?

I suppose the dream of high hashrate for 6900xt cards are no more

Alireza1773

newbie

Activity: 2

Merit: 0

Hello guys
I have 6 amd rx 580 and 5 rx6600xt
os: hiveos
When disable rx 580 gpu's miner doesnt work and show this error:
source/sp_workers/cl_gpu.cpp:cl_setup_devices():601: A OpenCL runtime error occurred: CL INVALID DEVICE
What should i do?

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: Nikneyime on February 07, 2022, 01:15:43 PM

over 6 hours it showed I sent 577 shares, however 2Miners stated that only 500 shares had been submitted.

What setting did you run on and which version did you test?

6 hours is not long enough. you should mine at least 5000 shares. The web statistics are inaccurate, they only update every 15 minutes and they only show the last 6 hours of mining. The share statistics are also aligned to a whole hour boundary, so if you started mining in the middle of the hour you risk loosing as much as 1 hour in the last 6 hours share rate statistics. 577 shares in 6 hours is around 346,2MHASH with give a poolside hashrate of 69.24 MhASH per card and this is an excellent result.

If you make a new test that start on the even hour. it will be synced with the pool statistics. and easier to compare.

Share statistics: Last an hour (02/07/2022 7:00 PM - 07:51 PM)

----------------------------------------------------

Quote from: KillrBee on February 07, 2022, 01:46:00 PM

Second -- Check out that 3090 hashrate for almost 6 hours of mining!

Nice hashrate. I think the 2060S needs to be clocked lower because you have some rejected shares there. It's the cuda code that is crashing, not the AMD code. I think the 2060S is clocked to high. try 7900 or 7800 memclock.

I could perhaps include a GPU # in the error message that will track wich of the gpu's that crashed the rig.

KillrBee

jr. member

Activity: 139

Merit: 3

Three notes here in one picture:

First is, when my CUDA cards fail under Linux, I don't get any information about which card has failed, and the error messages are extremely vague. Overclocking controls are pretty limited in Linux for Nvidia cards so I can only assume that it's the memory throwing a fault and causing the card to crash). Is there any way to indicate which card failed (and once one fails, the other one goes next, so I always get two of the same failure messages)

Second -- Check out that 3090 hashrate for almost 6 hours of mining!

Third -- These AMD cards are killing me -- they're so inconsistent in their real hashing speed. Are there any upcoming improvements in the kernel for them?

KillrBee

jr. member

Activity: 139

Merit: 3

Quote from: Nikneyime on February 07, 2022, 01:15:43 PM

I tried 5x RTX 3070 non-lhr against 2Miners pool, over 6 hours, I had no stales, but the received shares vs reported was way off. TBM was stating each 3070 was doing 65MH/s and over 6 hours it showed I sent 577 shares, however 2Miners stated that my average was less than 300MH/s, (less than 60 per 3070), and than only 500 shares had been submitted. If the Dev took the other 77 shares, then that is a dev fee of over 13%, not the claimed .5%. I do not know if the software was overstating or the Dev was taking more than claimed, either way i don’t like it, so I went back to T-Rex with that rig

2Miners use a 6 hour rotating snapshot window for displaying the number of shares, if you started your miner near the end of that window then your 77 shares would be reflected in the previous snapshot and reset to 0 for the next 6 hour window. If you look at the graphs for the miners, you can see if there were any invalid or stale shares provided by your miner for a longer timeframe (the yellow lines would show some peaks). Their various un-coordinated data analysis windows are a genuine pain in the ass to keep track of, and I will only really trust the averages on 2Miners after 13 hours of mining.

Nikneyime

newbie

Activity: 2

Merit: 0

I tried 5x RTX 3070 non-lhr against 2Miners pool, over 6 hours, I had no stales, but the received shares vs reported was way off. TBM was stating each 3070 was doing 65MH/s and over 6 hours it showed I sent 577 shares, however 2Miners stated that my average was less than 300MH/s, (less than 60 per 3070), and than only 500 shares had been submitted. If the Dev took the other 77 shares, then that is a dev fee of over 13%, not the claimed .5%. I do not know if the software was overstating or the Dev was taking more than claimed, either way i don’t like it, so I went back to T-Rex with that rig

Topic: Team Black Miner (ETHW ETC Vertcoin Ravencoin Zilliqa +dual +tripple mining ) - page 20. (Read 35049 times)