Page 1 of 7
CUDA Update to FAHCore_22
Posted: Tue Sep 29, 2020 12:12 am
by agent71
What a difference!
Have been running for over 5 months with a RTX2060 and regularly get 1.1-1.4M PPD - although I tend to pause during working day for heat/noise.
Cuda update landed and just seen first few WUs use Core22 Version 0.0.13 and it's up to 1.8M PPD.
That's a healthy 25-28% increase.
https://foldingathome.org/2020/09/28/fo ... a-support/
Re: CUDA Update
Posted: Tue Sep 29, 2020 12:26 am
by markdotgooley
Just seeing my first new ones just now. PPD estimates for a 2060 plus a 2060 KO are 3.3 million instead of 2.5... the X server settings app on Ubuntu Linux doesn’t work right any more though. Maybe I need to update some software? Or try a reboot at least?
These cards are both on the same cheapish motherboard, and GPU utilization seems to be a bit down as well. Maybe they’re being slightly throttled by limited bandwidth: PCIe 3.0 supposedly but one’s at x8 and the other at x4 because that’s all the motherboard allows.
Re: CUDA Update
Posted: Tue Sep 29, 2020 6:06 am
by sirdakz03
I also have the Core22 0.0.13 version, but I can't seem to run the WU in CUDA. It says CUDA_ERROR_FILE_NOT_FOUND when attempting to create the CUDA context:
Code: Select all
05:24:08:WU01:FS01:0x22:*************************** Core22 Folding@home Core ***************************
05:24:08:WU01:FS01:0x22: Core: Core22
05:24:08:WU01:FS01:0x22: Type: 0x22
05:24:08:WU01:FS01:0x22: Version: 0.0.13
05:24:08:WU01:FS01:0x22: Author: Joseph Coffland <joseph@cauldrondevelopment.com>
05:24:08:WU01:FS01:0x22: Copyright: 2020 foldingathome.org
05:24:08:WU01:FS01:0x22: Homepage: https://foldingathome.org/
05:24:08:WU01:FS01:0x22: Date: Sep 19 2020
05:24:08:WU01:FS01:0x22: Time: 02:35:58
05:24:08:WU01:FS01:0x22: Revision: 571cf95de6de2c592c7c3ed48fcfb2e33e9ea7d3
05:24:08:WU01:FS01:0x22: Branch: core22-0.0.13
05:24:08:WU01:FS01:0x22: Compiler: Visual C++ 2015
05:24:08:WU01:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
05:24:08:WU01:FS01:0x22: -DOPENMM_GIT_HASH="\"189320d0\""
05:24:08:WU01:FS01:0x22: Platform: win32 10
05:24:08:WU01:FS01:0x22: Bits: 64
05:24:08:WU01:FS01:0x22: Mode: Release
05:24:08:WU01:FS01:0x22:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
05:24:08:WU01:FS01:0x22: <peastman@stanford.edu>
05:24:08:WU01:FS01:0x22: Args: -dir 01 -suffix 01 -version 706 -lifeline 28420 -checkpoint 15
05:24:08:WU01:FS01:0x22: -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device
05:24:08:WU01:FS01:0x22: 0 -gpu 0
05:24:08:WU01:FS01:0x22:************************************ libFAH ************************************
05:24:08:WU01:FS01:0x22: Date: Sep 7 2020
05:24:08:WU01:FS01:0x22: Time: 19:09:56
05:24:08:WU01:FS01:0x22: Revision: 44301ed97b996b63fe736bb8073f22209cb2b603
05:24:08:WU01:FS01:0x22: Branch: HEAD
05:24:08:WU01:FS01:0x22: Compiler: Visual C++ 2015
05:24:08:WU01:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
05:24:08:WU01:FS01:0x22: Platform: win32 10
05:24:08:WU01:FS01:0x22: Bits: 64
05:24:08:WU01:FS01:0x22: Mode: Release
05:24:08:WU01:FS01:0x22:************************************ CBang *************************************
05:24:08:WU01:FS01:0x22: Date: Sep 7 2020
05:24:08:WU01:FS01:0x22: Time: 19:08:30
05:24:08:WU01:FS01:0x22: Revision: 33fcfc2b3ed2195a423606a264718e31e6b3903f
05:24:08:WU01:FS01:0x22: Branch: HEAD
05:24:08:WU01:FS01:0x22: Compiler: Visual C++ 2015
05:24:08:WU01:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
05:24:08:WU01:FS01:0x22: Platform: win32 10
05:24:08:WU01:FS01:0x22: Bits: 64
05:24:08:WU01:FS01:0x22: Mode: Release
05:24:08:WU01:FS01:0x22:************************************ System ************************************
05:24:08:WU01:FS01:0x22: CPU: Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
05:24:08:WU01:FS01:0x22: CPU ID: GenuineIntel Family 6 Model 158 Stepping 9
05:24:08:WU01:FS01:0x22: CPUs: 8
05:24:08:WU01:FS01:0x22: Memory: 39.96GiB
05:24:08:WU01:FS01:0x22:Free Memory: 24.64GiB
05:24:08:WU01:FS01:0x22: Threads: WINDOWS_THREADS
05:24:08:WU01:FS01:0x22: OS Version: 6.2
05:24:08:WU01:FS01:0x22:Has Battery: false
05:24:08:WU01:FS01:0x22: On Battery: false
05:24:08:WU01:FS01:0x22: UTC Offset: 8
05:24:08:WU01:FS01:0x22: PID: 29576
05:24:08:WU01:FS01:0x22: CWD: C:\Users\Dakz-PC\AppData\Roaming\FAHClient\work
05:24:08:WU01:FS01:0x22:************************************ OpenMM ************************************
05:24:08:WU01:FS01:0x22: Revision: 189320d0
05:24:08:WU01:FS01:0x22:********************************************************************************
05:24:08:WU01:FS01:0x22:Project: 17406 (Run 0, Clone 1793, Gen 1)
05:24:08:WU01:FS01:0x22:Unit: 0x00000001cedfaa925f59be2c838c0036
05:24:08:WU01:FS01:0x22:Reading tar file core.xml
05:24:08:WU01:FS01:0x22:Reading tar file integrator.xml.bz2
05:24:08:WU01:FS01:0x22:Reading tar file state.xml.bz2
05:24:08:WU01:FS01:0x22:Reading tar file system.xml.bz2
05:24:08:WU01:FS01:0x22:Digital signatures verified
05:24:08:WU01:FS01:0x22:Folding@home GPU Core22 Folding@home Core
05:24:08:WU01:FS01:0x22:Version 0.0.13
05:24:08:WU01:FS01:0x22: Checkpoint write interval: 25000 steps (2%) [50 total]
05:24:08:WU01:FS01:0x22: JSON viewer frame write interval: 12500 steps (1%) [100 total]
05:24:08:WU01:FS01:0x22: XTC frame write interval: 10000 steps (0.8%) [125 total]
05:24:08:WU01:FS01:0x22: Global context and integrator variables write interval: disabled
05:24:08:WU01:FS01:0x22:There are 4 platforms available.
05:24:08:WU01:FS01:0x22:Platform 0: Reference
05:24:08:WU01:FS01:0x22:Platform 1: CPU
05:24:08:WU01:FS01:0x22:Platform 2: OpenCL
05:24:08:WU01:FS01:0x22: opencl-device 0 specified
05:24:08:WU01:FS01:0x22:Platform 3: CUDA
05:24:08:WU01:FS01:0x22: cuda-device 0 specified
05:24:17:WU01:FS01:0x22:Attempting to create CUDA context:
05:24:17:WU01:FS01:0x22: Configuring platform CUDA
05:24:18:WU01:FS01:0x22:Failed to create CUDA context:
05:24:18:WU01:FS01:0x22:Error loading CUDA module: CUDA_ERROR_FILE_NOT_FOUND (301)
05:24:18:WU01:FS01:0x22:Attempting to create OpenCL context:
05:24:18:WU01:FS01:0x22: Configuring platform OpenCL
05:24:24:WU01:FS01:0x22: Using OpenCL on platformId 0 and gpu 0
I'm on an NVIDIA RTX 2070 Super with the latest drivers (456.55). I've read that I don't need to install the CUDA Toolkit, so I'm wondering what's happening.
Re: CUDA Update
Posted: Tue Sep 29, 2020 7:31 am
by PaulIngram
I'm seeing 4M ppd from a pair of Gigabyte 2060 supers. Up from ~ 3M yesterday.
Also seeing an increase on a 1060SC from ~480K PPD to ~750K PPD.
Re: CUDA Update
Posted: Tue Sep 29, 2020 7:32 am
by PantherX
Welcome to the F@H Forum sirdakz03,
Can you please post the first 100 lines which will inform us of what the system configuration is and what the client settings are. If you require guidance, please view this topic: viewtopic.php?f=24&t=26036
Based on the log file for covering FahCore_22 version 0.0.13, it has show that while CUDA failed, it is using OpenCL (assuming that no other error messages were thrown since the log stopped there). I would suggest that you download a fresh copy from Nvidia's website and then during installation, select a clean install. I am also assuming that your Windows 10 system is fully patched.
Re: CUDA Update
Posted: Tue Sep 29, 2020 7:58 am
by sirdakz03
PantherX wrote:Welcome to the F@H Forum sirdakz03,
Can you please post the first 100 lines which will inform us of what the system configuration is and what the client settings are. If you require guidance, please view this topic: viewtopic.php?f=24&t=26036
Based on the log file for covering FahCore_22 version 0.0.13, it has show that while CUDA failed, it is using OpenCL (assuming that no other error messages were thrown since the log stopped there). I would suggest that you download a fresh copy from Nvidia's website and then during installation, select a clean install. I am also assuming that your Windows 10 system is fully patched.
Thanks! I've been folding since 2012 but I've just registered to the official f@h forums specifically to ask this.
Here's the first 100 something lines:
Code: Select all
*********************** Log Started 2020-09-29T07:55:16Z ***********************
07:55:16:Trying to access database...
07:55:16:Successfully acquired database lock
07:55:16:Read GPUs.txt
07:55:16:Enabled folding slot 01: PAUSED gpu:0:TU104 [GeForce RTX 2070 SUPER] 8218 (by user)
07:55:16:Enabled folding slot 02: PAUSED cpu:7 (by user)
07:55:16:****************************** FAHClient ******************************
07:55:16: Version: 7.6.13
07:55:16: Author: Joseph Coffland <joseph@cauldrondevelopment.com>
07:55:16: Copyright: 2020 foldingathome.org
07:55:16: Homepage: https://foldingathome.org/
07:55:16: Date: Apr 27 2020
07:55:16: Time: 21:21:01
07:55:16: Revision: 5a652817f46116b6e135503af97f18e094414e3b
07:55:16: Branch: master
07:55:16: Compiler: Visual C++ 2008
07:55:16: Options: /TP /nologo /EHa /wd4297 /wd4103 /Ox /MT
07:55:16: Platform: win32 10
07:55:16: Bits: 32
07:55:16: Mode: Release
07:55:16: Args: --open-web-control
07:55:16: Config: C:\Users\Dakz-PC\AppData\Roaming\FAHClient\config.xml
07:55:16:******************************** CBang ********************************
07:55:16: Date: Apr 24 2020
07:55:16: Time: 17:07:55
07:55:16: Revision: ea081a3b3b0f4a37c4d0440b4f1bc184197c7797
07:55:16: Branch: master
07:55:16: Compiler: Visual C++ 2008
07:55:16: Options: /TP /nologo /EHa /wd4297 /wd4103 /Ox /MT
07:55:16: Platform: win32 10
07:55:16: Bits: 32
07:55:16: Mode: Release
07:55:16:******************************* System ********************************
07:55:16: CPU: Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
07:55:16: CPU ID: GenuineIntel Family 6 Model 158 Stepping 9
07:55:16: CPUs: 8
07:55:16: Memory: 39.96GiB
07:55:16: Free Memory: 23.00GiB
07:55:16: Threads: WINDOWS_THREADS
07:55:16: OS Version: 6.2
07:55:16: Has Battery: false
07:55:16: On Battery: false
07:55:16: UTC Offset: 8
07:55:16: PID: 1156
07:55:16: CWD: C:\Users\Dakz-PC\AppData\Roaming\FAHClient
07:55:16: Win32 Service: false
07:55:16: OS: Windows 10 Enterprise
07:55:16: OS Arch: AMD64
07:55:16: GPUs: 1
07:55:16: GPU 0: Bus:1 Slot:0 Func:0 NVIDIA:8 TU104 [GeForce RTX 2070 SUPER]
07:55:16: 8218
07:55:16: CUDA Device 0: Platform:0 Device:0 Bus:1 Slot:0 Compute:7.5 Driver:11.1
07:55:16:OpenCL Device 0: Platform:0 Device:0 Bus:1 Slot:0 Compute:1.2 Driver:456.55
07:55:16:******************************* libFAH ********************************
07:55:16: Date: Apr 15 2020
07:55:16: Time: 14:53:14
07:55:16: Revision: 216968bc7025029c841ed6e36e81a03a316890d3
07:55:16: Branch: master
07:55:16: Compiler: Visual C++ 2008
07:55:16: Options: /TP /nologo /EHa /wd4297 /wd4103 /Ox /MT
07:55:16: Platform: win32 10
07:55:16: Bits: 32
07:55:16: Mode: Release
07:55:16:***********************************************************************
07:55:16:<config>
07:55:16: <!-- Network -->
07:55:16: <proxy v=':8080'/>
07:55:16:
07:55:16: <!-- Slot Control -->
07:55:16: <power v='full'/>
07:55:16:
07:55:16: <!-- User Information -->
07:55:16: <passkey v='*****'/>
07:55:16: <team v='167809'/>
07:55:16: <user v='sirdakz03'/>
07:55:16:
07:55:16: <!-- Folding Slots -->
07:55:16: <slot id='1' type='GPU'>
07:55:16: <paused v='true'/>
07:55:16: </slot>
07:55:16: <slot id='2' type='CPU'>
07:55:16: <paused v='true'/>
07:55:16: </slot>
07:55:16:</config>
07:55:16:WU01:FS01:Sending unit results: id:01 state:SEND error:NO_ERROR project:17406 run:0 clone:1793 gen:1 core:0x22 unit:0x00000001cedfaa925f59be2c838c0036
07:55:16:WU01:FS01:Uploading 20.96MiB to 206.223.170.146
07:55:16:WU01:FS01:Connecting to 206.223.170.146:8080
07:55:19:6:127.0.0.1:New Web session
07:55:52:WU01:FS01:Upload 0.60%
07:55:52:WARNING:WU01:FS01:Exception: Failed to send results to work server: Transfer failed
07:55:52:WU01:FS01:Sending unit results: id:01 state:SEND error:NO_ERROR project:17406 run:0 clone:1793 gen:1 core:0x22 unit:0x00000001cedfaa925f59be2c838c0036
07:55:52:WU01:FS01:Uploading 20.96MiB to 206.223.170.146
07:55:52:WU01:FS01:Connecting to 206.223.170.146:8080
07:56:28:WU01:FS01:Upload 0.60%
07:56:28:WARNING:WU01:FS01:Exception: Failed to send results to work server: Transfer failed
07:56:39:FS01:Unpaused
07:56:39:WU02:FS01:Starting
07:56:39:WU02:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:\Users\Dakz-PC\AppData\Roaming\FAHClient\cores/cores.foldingathome.org/win/64bit/22-0.0.13/Core_22.fah/FahCore_22.exe -dir 02 -suffix 01 -version 706 -lifeline 1156 -checkpoint 15 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device 0 -gpu 0
07:56:39:WU02:FS01:Started FahCore on PID 17736
07:56:39:WU02:FS01:Core PID:21244
07:56:39:WU02:FS01:FahCore 0x22 started
07:56:39:WU02:FS01:0x22:*********************** Log Started 2020-09-29T07:56:39Z ***********************
07:56:39:WU02:FS01:0x22:*************************** Core22 Folding@home Core ***************************
07:56:39:WU02:FS01:0x22: Core: Core22
07:56:39:WU02:FS01:0x22: Type: 0x22
07:56:39:WU02:FS01:0x22: Version: 0.0.13
07:56:39:WU02:FS01:0x22: Author: Joseph Coffland <joseph@cauldrondevelopment.com>
07:56:39:WU02:FS01:0x22: Copyright: 2020 foldingathome.org
07:56:39:WU02:FS01:0x22: Homepage: https://foldingathome.org/
07:56:39:WU02:FS01:0x22: Date: Sep 19 2020
07:56:39:WU02:FS01:0x22: Time: 02:35:58
07:56:39:WU02:FS01:0x22: Revision: 571cf95de6de2c592c7c3ed48fcfb2e33e9ea7d3
07:56:39:WU02:FS01:0x22: Branch: core22-0.0.13
07:56:39:WU02:FS01:0x22: Compiler: Visual C++ 2015
07:56:39:WU02:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
07:56:39:WU02:FS01:0x22: -DOPENMM_GIT_HASH="\"189320d0\""
07:56:39:WU02:FS01:0x22: Platform: win32 10
07:56:39:WU02:FS01:0x22: Bits: 64
07:56:39:WU02:FS01:0x22: Mode: Release
07:56:39:WU02:FS01:0x22:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
07:56:39:WU02:FS01:0x22: <peastman@stanford.edu>
07:56:39:WU02:FS01:0x22: Args: -dir 02 -suffix 01 -version 706 -lifeline 17736 -checkpoint 15
07:56:39:WU02:FS01:0x22: -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device
07:56:39:WU02:FS01:0x22: 0 -gpu 0
07:56:39:WU02:FS01:0x22:************************************ libFAH ************************************
07:56:39:WU02:FS01:0x22: Date: Sep 7 2020
07:56:39:WU02:FS01:0x22: Time: 19:09:56
07:56:39:WU02:FS01:0x22: Revision: 44301ed97b996b63fe736bb8073f22209cb2b603
07:56:39:WU02:FS01:0x22: Branch: HEAD
07:56:39:WU02:FS01:0x22: Compiler: Visual C++ 2015
07:56:39:WU02:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
07:56:39:WU02:FS01:0x22: Platform: win32 10
07:56:39:WU02:FS01:0x22: Bits: 64
07:56:39:WU02:FS01:0x22: Mode: Release
07:56:39:WU02:FS01:0x22:************************************ CBang *************************************
07:56:39:WU02:FS01:0x22: Date: Sep 7 2020
07:56:39:WU02:FS01:0x22: Time: 19:08:30
07:56:39:WU02:FS01:0x22: Revision: 33fcfc2b3ed2195a423606a264718e31e6b3903f
07:56:39:WU02:FS01:0x22: Branch: HEAD
07:56:39:WU02:FS01:0x22: Compiler: Visual C++ 2015
07:56:39:WU02:FS01:0x22: Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
07:56:39:WU02:FS01:0x22: Platform: win32 10
07:56:39:WU02:FS01:0x22: Bits: 64
07:56:39:WU02:FS01:0x22: Mode: Release
07:56:39:WU02:FS01:0x22:************************************ System ************************************
07:56:39:WU02:FS01:0x22: CPU: Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
07:56:39:WU02:FS01:0x22: CPU ID: GenuineIntel Family 6 Model 158 Stepping 9
07:56:39:WU02:FS01:0x22: CPUs: 8
07:56:39:WU02:FS01:0x22: Memory: 39.96GiB
07:56:39:WU02:FS01:0x22:Free Memory: 22.97GiB
07:56:39:WU02:FS01:0x22: Threads: WINDOWS_THREADS
07:56:39:WU02:FS01:0x22: OS Version: 6.2
07:56:39:WU02:FS01:0x22:Has Battery: false
07:56:39:WU02:FS01:0x22: On Battery: false
07:56:39:WU02:FS01:0x22: UTC Offset: 8
07:56:39:WU02:FS01:0x22: PID: 21244
07:56:39:WU02:FS01:0x22: CWD: C:\Users\Dakz-PC\AppData\Roaming\FAHClient\work
07:56:40:WU02:FS01:0x22:************************************ OpenMM ************************************
07:56:40:WU02:FS01:0x22: Revision: 189320d0
07:56:40:WU02:FS01:0x22:********************************************************************************
07:56:40:WU02:FS01:0x22:Project: 16921 (Run 18, Clone 50, Gen 2)
07:56:40:WU02:FS01:0x22:Unit: 0x000000020002894c5f5bf3573a5391c8
07:56:40:WU02:FS01:0x22:Digital signatures verified
07:56:40:WU02:FS01:0x22:Folding@home GPU Core22 Folding@home Core
07:56:40:WU02:FS01:0x22:Version 0.0.13
07:56:40:WU02:FS01:0x22: Checkpoint write interval: 250000 steps (5%) [20 total]
07:56:40:WU02:FS01:0x22: JSON viewer frame write interval: 50000 steps (1%) [100 total]
07:56:40:WU02:FS01:0x22: XTC frame write interval: 250000 steps (5%) [20 total]
07:56:40:WU02:FS01:0x22: Global context and integrator variables write interval: disabled
07:56:40:WU02:FS01:0x22:There are 4 platforms available.
07:56:40:WU02:FS01:0x22:Platform 0: Reference
07:56:40:WU02:FS01:0x22:Platform 1: CPU
07:56:40:WU02:FS01:0x22:Platform 2: OpenCL
07:56:40:WU02:FS01:0x22: opencl-device 0 specified
07:56:40:WU02:FS01:0x22:Platform 3: CUDA
07:56:40:WU02:FS01:0x22: cuda-device 0 specified
07:56:40:WU02:FS01:0x22:Attempting to create CUDA context:
07:56:40:WU02:FS01:0x22: Configuring platform CUDA
07:56:40:WU02:FS01:0x22:Failed to create CUDA context:
07:56:40:WU02:FS01:0x22:Error loading CUDA module: CUDA_ERROR_FILE_NOT_FOUND (301)
07:56:40:WU02:FS01:0x22:Attempting to create OpenCL context:
07:56:40:WU02:FS01:0x22: Configuring platform OpenCL
07:56:44:WU02:FS01:0x22: Using OpenCL on platformId 0 and gpu 0
07:56:44:WU02:FS01:0x22:Completed 0 out of 5000000 steps (0%)
07:56:44:WU02:FS01:0x22:Checkpoint completed at step 0
07:56:52:WU01:FS01:Sending unit results: id:01 state:SEND error:NO_ERROR project:17406 run:0 clone:1793 gen:1 core:0x22 unit:0x00000001cedfaa925f59be2c838c0036
07:56:52:WU01:FS01:Uploading 20.96MiB to 206.223.170.146
07:56:52:WU01:FS01:Connecting to 206.223.170.146:8080
07:57:12:WU02:FS01:0x22:Completed 50000 out of 5000000 steps (1%)
And yes, you are correct in that I do have the Windows 10 May Update (2004) and is fully patched. I did a clean install, but that didn't do anything. I will try to use DDU to completely uninstall and then reinstall the drivers again.
EDIT: Aaaaaand it's still the same T_T. Did the DDU, clean installed and it still used OpenCL over CUDA, with the same CUDA error.
Re: CUDA Update
Posted: Tue Sep 29, 2020 9:56 am
by PantherX
As a quick sanity check, can you please run GPU-Z and verify that you can see CUDA ticked? Since you're the first person to report this issue, I am hoping to collect enough data to pass it on to the right people which can hopefully result in a fix
Re: CUDA Update
Posted: Tue Sep 29, 2020 10:03 am
by gunnarre
13426 on 1660 Super: 1.24 M PPD, TPF 1:56. (CUDA)
13426 on 1050 Ti: 476k PPD, TPF 3:41 (CUDA)
13426 on RX580: 447k PPD, TPF 3:50 (OpenCL)
Project 14487 TPF:
1:40 1660 Super OpenCL
1:09 1660 Super CUDA
Re: CUDA Update
Posted: Tue Sep 29, 2020 10:19 am
by sirdakz03
PantherX wrote:As a quick sanity check, can you please run GPU-Z and verify that you can see CUDA ticked? Since you're the first person to report this issue, I am hoping to collect enough data to pass it on to the right people which can hopefully result in a fix
CUDA is ticked as per the latest GPU-Z. Let me know what else information is needed from me and I'll be happy to provide them.
Maybe I missed something, however the article states that I shouldn't be doing anything else except install the latest graphics driver. I also just installed the latest CUDA Toolkit in the hopes that it might run the WU's to CUDA, but to no avail.
Re: CUDA Update
Posted: Tue Sep 29, 2020 1:42 pm
by Neil-B
Iirc folder in another thread had to remove cuda dev kit before could get cuda to work
Re: CUDA Update
Posted: Tue Sep 29, 2020 3:44 pm
by sirdakz03
Neil-B wrote:Iirc folder in another thread had to remove cuda dev kit before could get cuda to work
Confirmed, this fixed my issue. I did have an older version of the CUDA toolkit (10.1) installed and I thought I cleared that previously before I attempted to install the newest CUDA Toolkit version. I have removed and uninstalled the toolkits both old and new, and now have the WU's running on CUDA. Yay!
On a side note, it had to work on the WU from scratch, as it didn't seem to recognize the OpenCL checkpoint.
Re: CUDA Update
Posted: Tue Sep 29, 2020 3:59 pm
by foldy
That maybe shows another issue: If user have any CUDA toolkit installed then FAH seems to use that instead of the nvidia driver. That is not a good idea, FAH should always run using the nvidia driver for CUDA. viewtopic.php?f=80&t=36167
Re: CUDA Update
Posted: Tue Sep 29, 2020 6:05 pm
by gunnarre
sirdakz03 wrote:On a side note, it had to work on the WU from scratch, as it didn't seem to recognize the OpenCL checkpoint.
This is expected behavior, and shouldn't happen more than once when you get the new core. But the other problem of failing to fold if another CUDA toolkit is installed, is more critical.
Re: CUDA Update
Posted: Tue Sep 29, 2020 6:28 pm
by HaloJones
I'm seeing massively better CUDA performance on Linux than Windows10. And I'm meaning up to 40% on identical cards.
Re: CUDA Update
Posted: Tue Sep 29, 2020 6:41 pm
by foldy
For project 13426 I see bus usage on pcie 2.0 x16 is only 16% that is very good. With OpenCL bus usage was much higher on Windows which can lead to bottleneck on fast GPUs. PPD went from OpenCL 1.5M PPD to CUDA 2.0M PPD with gtx 1080ti on WIndows.
Just speculation but maybe the CUDA speedup to OpenCL is related to bus usage? Maybe OpenCL could be optimized when bus usage gets reduced somehow?