Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux)

[darkneorus]

no success with P104/P106 GPUs on 425.31 driver/Win10 1709.
with -strap 1 the miner runs for several seconds then hashrate for all cards drops to zero, and the rig can reboot with a dpc_watchdog_violation BSoD.
with -strap 2 - incorrect shares then MEMORY_MANAGEMENT BSoD.

am I doing something wrong?

[ganzocrypt]

How do you remove the driver installed by EthDcr...exe ?

[darkneorus]

How do you remove the driver installed by EthDcr...exe ?

Code:

-driver uninstall

[rednoW]

2 Claymore

gtx1080tis are here, driver 425.31

-strap 1 - low 40+ hashrate, but system is stable
-strap 2 - need to apply only with std mem clock. Can raise mem clock after miner started and reach fantastic 56.5mhz but if miner closes then system crashes. Need to lower mem clock before exit.
I have my P2 cuda states disabled by the way to prevent memory clock spikes between different P-states switches.

-strap 3 - system crashes even with -502 mem clock.

So, it looks promising but unusable. Get back to EthPill and 55mhs stable and no crashes when miner starts or exits

[Larsinist]

i get more hash with phoenix miner, so could be that some of my settings are wrong, or?

[badworm]

Both v14.2 do BSOD with Vega and 19.5.1 when I use -strap.

PAGE_FAULT_IN_NONPAGED_AREA
cmdrv64.sys

Strap Drivers are already in use when you load another instance of the miner. Here is log from b711 version :

01:37:11:329   fcc   Driver is not installed! Trying to install it...
01:37:11:373   fcc   Driver is in use, try to start miner with "-driver uninstall" parameter.
01:37:11:792   fcc   SSL: Imported 76 certificates from local storage

Rgds

[nitrobg]

No love for Linux?

[adaseb]

I have tried this version and its my first time returning to mining ETH since last November 2018. I must say that this version is much more stable than the previous versions.

Basically did nothing except replace the pools and config file and it worked right way. I got decent speeds without having to do much tweeking with the straps or memory clock. Most important I've never gotten a system hang before which was frequent before. Especially if it was a computer I used to browse the web.

[mu_enrico]

I tried v14.2 in Windows 10 and made my Windows 10 running in test mode...
My cards are RX550 4GB and RX560 4GB with custom bios, tried with -strap 1 and got worse hashrate
So I disabled -strap and it seems that this version run as stable as v12.0
I'll give you update after 24 hours or so.

[rednoW]

gtx1070 micron from 32.8 to 35 mhs - nice!

[Marvell2]

gtx1070 micron from 32.8 to 35 mhs - nice!

wow thats impressive, what memory setting though and what power limit

In my case i gave up on vegas with this build , cant get it to detect my 56s and add straps , 570 microns and Hynix are another story

Im getting 30mhs on stock bios with strap 3 and also seem to be able to lower core and mem voltage more than on my custom bioses.

I have 5 570s all using 90 or less watts mesured at the wall !!

i think i can go even lower , posibly 70-80 watts

[rednoW]

gtx1070 micron from 32.8 to 35 mhs - nice!

wow thats impressive, what memory setting though and what power limit

Those are old palit gamerock and jetstream gtx 1070 cards with micron memory.
They don't like aggressive downvolt with Curve (MSI Afterburner) below 775mv.
So I need to use classic power target and gpu overclock to gain more efficiency   ~1600+gpu clock, +550 mem, ~800mV, ~120watt according nvidia smi.
settings are the same before was 32.8mhs now with claymore 35. Resulting gpu clock and voltage became lower automatically with high load ...

[rednoW]

2 Claymore

gtx1080tis are here, driver 425.31

-strap 1 - low 40+ hashrate, but system is stable
-strap 2 - need to apply only with std mem clock. Can raise mem clock after miner started and reach fantastic 56.5mhz but if miner closes then system crashes. Need to lower mem clock before exit.
I have my P2 cuda states disabled by the way to prevent memory clock spikes between different P-states switches.

-strap 3 - system crashes even with -502 mem clock.

So, it looks promising but unusable. Get back to EthPill and 55mhs stable and no crashes when miner starts or exits

2 Claymore.
I think you should rework the way you do miner exit on gtx1080ti's with straps applied. I found a trick how to fix your timings and than run your miner without strap management.
It keeps high speed of modded timing and doesn't crash system on exit.

[davescream]

14.0 - 14.2 versions do not detect NVIDIA GPUs on mixed rigs (AMD + NVIDIA)

[rednoW]

14.0 - 14.2 versions do not detect NVIDIA GPUs on mixed rigs (AMD + NVIDIA)

Not true

[davescream]

14.0 - 14.2 versions do not detect NVIDIA GPUs on mixed rigs (AMD + NVIDIA)

14.2 No NVIDIA cards in the list, NVML library will not be used.

Code:

2:55:53:410 10dc Check and remove old log files...
12:55:53:410 10dc args: -logfile c:\local\FARM6\Miner.log -mode 1 -epool us1.ethermine.org:4444 -ewal 0x********************************.Farm6 -epsw ****** -mpsw ******* -r 1 -ethi 8 -etha 0 -ftime 5 -y 1 -altnum 1 -minspeed 140 -tt 73 -ttdcr 76 -ttli 81 -tstop 85 -fanmin 15 -asm 2
12:55:53:410 10dc
12:55:53:410 10dc ЙHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH»
12:55:53:410 10dc є              Claymore's Dual GPU Miner - v14.2 Beta            є
12:55:53:410 10dc є              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             є
12:55:53:410 10dc є                       Supercharged Edition                     є
12:55:53:410 10dc ИHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHj
12:55:53:410 10dc
12:55:53:410 10dc b715
12:55:53:629 10dc ETH: 13 pools are specified
12:55:53:629 10dc Main Ethereum pool is us1.ethermine.org:4444
12:55:54:457 10dc OpenCL platform: NVIDIA CUDA
12:55:54:457 10dc OpenCL platform: AMD Accelerated Parallel Processing
12:55:54:457 10dc OpenCL initializing...

12:55:54:457 10dc AMD Cards available: 3
12:55:54:457 10dc GPU #1: Ellesmere (Radeon (TM) RX 470 Graphics), 4096 MB available, 32 compute units (pci bus 1:0:0)
12:55:54:457 10dc GPU #1 recognized as Radeon RX 470/570
12:55:54:457 10dc GPU #2: Ellesmere (Radeon (TM) RX 470 Graphics), 4096 MB available, 32 compute units (pci bus 2:0:0)
12:55:54:457 10dc GPU #2 recognized as Radeon RX 470/570
12:55:54:457 10dc GPU #3: Ellesmere (Radeon (TM) RX 480 Graphics), 4096 MB available, 36 compute units (pci bus 3:0:0)
12:55:54:457 10dc GPU #3 recognized as Radeon RX 480/580
12:55:54:457 10dc POOL/SOLO version
12:55:54:457 10dc Platform: Windows
12:55:54:535 10dc start building OpenCL program for GPU 1...
12:55:55:035 10dc done
12:55:55:254 10dc start building OpenCL program for GPU 2...
12:55:55:739 10dc done
12:55:55:957 10dc start building OpenCL program for GPU 3...
12:55:56:442 10dc done
12:55:56:707 10dc GPU #1: algorithm ASM 2
12:55:56:754 10dc GPU #2: algorithm ASM 2
12:55:56:817 10dc GPU #3: algorithm ASM 2
12:55:56:832 10dc cudaGetDeviceCount failed (35, CUDA driver version is insufficient for CUDA runtime version), probably no CUDA devices
12:55:56:832 10dc No NVIDIA CUDA GPUs detected.
12:55:56:832 10dc Total cards: 3
12:55:56:832 10dc
Please consider using "-strap" option to increase mining speed, check Readme for details.

12:55:58:864 10dc
You can use "+" and "-" keys to achieve best ETH speed, see "FINE TUNING" section in Readme for details.

12:55:59:879 10dc No NVIDIA cards in the list, NVML library will not be used.
12:55:59:879 10dc All AMD cards use Compute Mode already
12:55:59:879 10dc CrossFire is disabled already
12:56:00:926 10dc SSL: Imported 31 certificates from local storage

12.0 - ok  12:52:51:975   1248   NVIDIA Cards available: 2

Code:

12:52:47:663 1248 Check and remove old log files...
12:52:47:663 1248 args: -logfile c:\local\FARM6\Miner.log -mode 1 -epool us1.ethermine.org:4444 -ewal 0x*******.Farm6 -epsw ********** -mpsw ******** -r 1 -ethi 8 -etha 0 -ftime 5 -y 1 -altnum 1 -minspeed 140 -tt 73 -ttdcr 76 -ttli 81 -tstop 85 -fanmin 15 -asm 2
12:52:47:663 1248
12:52:47:663 1248 ЙHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH»
12:52:47:663 1248 є                Claymore's Dual GPU Miner - v12.0               є
12:52:47:663 1248 є              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             є
12:52:47:663 1248 ИHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHj
12:52:47:663 1248
12:52:47:663 1248 b615
12:52:47:928 1248 ETH: 13 pools are specified
12:52:47:928 1248 Main Ethereum pool is us1.ethermine.org:4444
12:52:48:616 1248 OpenCL platform: NVIDIA CUDA
12:52:48:616 1248 OpenCL platform: AMD Accelerated Parallel Processing
12:52:48:616 1248 OpenCL initializing...

12:52:48:616 1248 AMD Cards available: 3
12:52:48:616 1248 GPU #1: Ellesmere (Radeon (TM) RX 470 Graphics), 4096 MB available, 32 compute units (pci bus 1:0:0)
12:52:48:616 1248 GPU #1 recognized as Radeon RX 470/570
12:52:48:616 1248 GPU #2: Ellesmere (Radeon (TM) RX 470 Graphics), 4096 MB available, 32 compute units (pci bus 2:0:0)
12:52:48:616 1248 GPU #2 recognized as Radeon RX 470/570
12:52:48:616 1248 GPU #3: Ellesmere (Radeon (TM) RX 480 Graphics), 4096 MB available, 36 compute units (pci bus 3:0:0)
12:52:48:616 1248 GPU #3 recognized as Radeon RX 480/580
12:52:48:616 1248 POOL/SOLO version
12:52:48:616 1248 Platform: Windows
12:52:48:756 1248 start building OpenCL program for GPU 1...
12:52:49:241 1248 done
12:52:49:475 1248 start building OpenCL program for GPU 2...
12:52:49:959 1248 done
12:52:50:178 1248 start building OpenCL program for GPU 3...
12:52:50:694 1248 done
12:52:50:959 1248 GPU #1: algorithm ASM 2
12:52:51:022 1248 GPU #2: algorithm ASM 2
12:52:51:084 1248 GPU #3: algorithm ASM 2
12:52:51:975 1248 CUDA initializing...

12:52:51:975 1248 NVIDIA Cards available: 2
12:52:51:975 1248 CUDA Driver Version/Runtime Version: 8.0/8.0
12:52:52:053 1248 GPU #4: GeForce GTX 1070, 8192 MB available, 15 compute units, capability: 6.1  (pci bus 5:0:0)
12:52:52:069 1248 GPU #5: GeForce GTX 1070, 8192 MB available, 15 compute units, capability: 6.1  (pci bus 7:0:0)
12:52:52:069 1248 Total cards: 5
12:52:53:303 1248
You can use "+" and "-" keys to achieve best ETH speed, see "FINE TUNING" section in Readme for details.

12:52:54:381 1248 NVML version: 8.382.53
12:52:54:413 1248 All AMD cards use Compute Mode already
12:52:54:413 1248 CrossFire is disabled already
12:52:56:006 1248 SSL: Imported 31 certificates from local storage

[Claymore]

2 Claymore.
I think you should rework the way you do miner exit on gtx1080ti's with straps applied. I found a trick how to fix your timings and than run your miner without strap management.
It keeps high speed of modded timing and doesn't crash system on exit.

Ok I will restore settings at exit.

14.0 - 14.2 versions do not detect NVIDIA GPUs on mixed rigs (AMD + NVIDIA)

Install latest drivers or use files from "cuda8" folder (redownload v14.2 if you don't have this folder).

[dwu]

I read several pages and I still don't know if I should switch to his straps. I made a custom BIOS a year ago and I have around 29-30MH (Reported: mostly 30-31.5) for my RX480/580 4GB cards.
Probably my timings are not the most stable and memory errors are occurring on some cards too (I still do not fully know if it is actually bad or not), because I tried to push it to the limit. Sometimes the miner restarts once a day.

What do you suggest?
- Should I flash the original BIOS back ...
- ... and use Claymore/Eliovp's Mem tweak?
- ... or can I still use my modded BIOS with Claymore?
- Do i still need to apply the pixel-clock-patcher after installing a new AMD driver?


Thanks for any help!

[rednoW]

2 Claymore

One more thing. Think you have a bug with NVIDIA timings. When I  try to apply strap 1 to one card and strap 2 to another. But they both have the strap 2 as result.

[trustfarm]

Can you support TethashV1 (upgrade ethash)

1. Project description (pools and rteo parity modified node)
  https://github.com/tao-foundation/rteo


here's algorithm EIP-1485
1-1.  https://github.com/ethereum/EIPs/blob/master/EIPS/eip-1485.md#4-real-implementation-for-fnv1a

1-2.  running pool lists
   -  1) teopool.tao.foundation
   -  2) teopool2.tao.foundation
   -  3) pool1.tao.foundation

1-3.  efficiency of algorithm compared than ethash
   - 1) more randomizing on MIX parts.
   - 2) simple more multiplication loops makes less hashrates 5~10% , compared than ethash
   - 3) simple more multiplication loops makes hard to overclock for the meet fpga / asic timing meets.
   - 4) more fair speed on cpu / gpu / fpga / asic .


It has upgrade ethash MIX parts, for prevent ETH-ASIC .

deprecated fnv0 upgrade to fnv1a

applied source patched things here.

2. teominer fork of  ethminer
https://github.com/tao-foundation/teominer

2-1. reference opencl code :  https://github.com/tao-foundation/teominer/blob/master/libethash-cl/kernels/cl/ethash.cl
2-2. reference cuda code :  https://github.com/tao-foundation/teominer/blob/master/libethash-cuda/fnv.cuh

2-3. patched ethash cpu code : https://github.com/tao-foundation/tethashv1
 
If you are interested on this, you can add easily, because small modification of ethash.
And, please let me know email me : cpplover@trustfarm.net

Thanks.