Radeon R7 240 fails to start folding

[FishCZ]

Hi,
on monday I have build a new system. To test it I used F@H. both CPU and GPU were working just fine and then GPU started to fail.
here is the log snip.

is this archaic GPU still supported? or can you spot in the log, what went wrong?

Thanks for an advice

Code: Select all

22:29:56:WU00:FS01:Connecting to assign1.foldingathome.org:80
22:29:57:WU00:FS01:Assigned to work server 54.157.202.86
22:29:57:WU00:FS01:Requesting new work unit for slot 01: gpu:43:0 Hawaii [Radeon R7 200 Series] from 54.157.202.86
22:29:57:WU00:FS01:Connecting to 54.157.202.86:8080
22:29:58:WU00:FS01:Downloading 6.91MiB
22:30:02:WU00:FS01:Download complete
22:30:02:WU00:FS01:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:13446 run:2389 clone:68 gen:0 core:0x22 unit:0x00000044000000000000348600000955
22:30:02:WU00:FS01:Starting
22:30:02:WU00:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:\ProgramData\FAHClient\cores/cores.foldingathome.org/win/64bit/22-0.0.13/Core_22.fah/FahCore_22.exe -dir 00 -suffix 01 -version 706 -lifeline 14180 -checkpoint 15 -opencl-platform 0 -opencl-device 0 -gpu-vendor amd -gpu 0 -gpu-usage 100
22:30:02:WU00:FS01:Started FahCore on PID 2764
22:30:02:WU00:FS01:Core PID:15776
22:30:02:WU00:FS01:FahCore 0x22 started
22:30:03:WU00:FS01:0x22:*********************** Log Started 2021-03-26T22:30:02Z ***********************
22:30:03:WU00:FS01:0x22:*************************** Core22 Folding@home Core ***************************
22:30:03:WU00:FS01:0x22:       Core: Core22
22:30:03:WU00:FS01:0x22:       Type: 0x22
22:30:03:WU00:FS01:0x22:    Version: 0.0.13
22:30:03:WU00:FS01:0x22:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
22:30:03:WU00:FS01:0x22:  Copyright: 2020 foldingathome.org
22:30:03:WU00:FS01:0x22:   Homepage: https://foldingathome.org/
22:30:03:WU00:FS01:0x22:       Date: Sep 19 2020
22:30:03:WU00:FS01:0x22:       Time: 02:35:58
22:30:03:WU00:FS01:0x22:   Revision: 571cf95de6de2c592c7c3ed48fcfb2e33e9ea7d3
22:30:03:WU00:FS01:0x22:     Branch: core22-0.0.13
22:30:03:WU00:FS01:0x22:   Compiler: Visual C++ 2015
22:30:03:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
22:30:03:WU00:FS01:0x22:             -DOPENMM_GIT_HASH="\"189320d0\""
22:30:03:WU00:FS01:0x22:   Platform: win32 10
22:30:03:WU00:FS01:0x22:       Bits: 64
22:30:03:WU00:FS01:0x22:       Mode: Release
22:30:03:WU00:FS01:0x22:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
22:30:03:WU00:FS01:0x22:             <peastman@stanford.edu>
22:30:03:WU00:FS01:0x22:       Args: -dir 00 -suffix 01 -version 706 -lifeline 2764 -checkpoint 15
22:30:03:WU00:FS01:0x22:             -opencl-platform 0 -opencl-device 0 -gpu-vendor amd -gpu 0
22:30:03:WU00:FS01:0x22:             -gpu-usage 100
22:30:03:WU00:FS01:0x22:************************************ libFAH ************************************
22:30:03:WU00:FS01:0x22:       Date: Sep 7 2020
22:30:03:WU00:FS01:0x22:       Time: 19:09:56
22:30:03:WU00:FS01:0x22:   Revision: 44301ed97b996b63fe736bb8073f22209cb2b603
22:30:03:WU00:FS01:0x22:     Branch: HEAD
22:30:03:WU00:FS01:0x22:   Compiler: Visual C++ 2015
22:30:03:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
22:30:03:WU00:FS01:0x22:   Platform: win32 10
22:30:03:WU00:FS01:0x22:       Bits: 64
22:30:03:WU00:FS01:0x22:       Mode: Release
22:30:03:WU00:FS01:0x22:************************************ CBang *************************************
22:30:03:WU00:FS01:0x22:       Date: Sep 7 2020
22:30:03:WU00:FS01:0x22:       Time: 19:08:30
22:30:03:WU00:FS01:0x22:   Revision: 33fcfc2b3ed2195a423606a264718e31e6b3903f
22:30:03:WU00:FS01:0x22:     Branch: HEAD
22:30:03:WU00:FS01:0x22:   Compiler: Visual C++ 2015
22:30:03:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
22:30:03:WU00:FS01:0x22:   Platform: win32 10
22:30:03:WU00:FS01:0x22:       Bits: 64
22:30:03:WU00:FS01:0x22:       Mode: Release
22:30:03:WU00:FS01:0x22:************************************ System ************************************
22:30:03:WU00:FS01:0x22:        CPU: AMD Ryzen 5 3600 6-Core Processor
22:30:03:WU00:FS01:0x22:     CPU ID: AuthenticAMD Family 23 Model 113 Stepping 0
22:30:03:WU00:FS01:0x22:       CPUs: 12
22:30:03:WU00:FS01:0x22:     Memory: 15.93GiB
22:30:03:WU00:FS01:0x22:Free Memory: 10.92GiB
22:30:03:WU00:FS01:0x22:    Threads: WINDOWS_THREADS
22:30:03:WU00:FS01:0x22: OS Version: 6.2
22:30:03:WU00:FS01:0x22:Has Battery: false
22:30:03:WU00:FS01:0x22: On Battery: false
22:30:03:WU00:FS01:0x22: UTC Offset: 1
22:30:03:WU00:FS01:0x22:        PID: 15776
22:30:03:WU00:FS01:0x22:        CWD: C:\ProgramData\FAHClient\work
22:30:03:WU00:FS01:0x22:************************************ OpenMM ************************************
22:30:03:WU00:FS01:0x22:   Revision: 189320d0
22:30:03:WU00:FS01:0x22:********************************************************************************
22:30:03:WU00:FS01:0x22:Project: 13446 (Run 2389, Clone 68, Gen 0)
22:30:03:WU00:FS01:0x22:Unit: 0x00000000000000000000000000000000
22:30:03:WU00:FS01:0x22:Reading tar file core.xml
22:30:03:WU00:FS01:0x22:Reading tar file integrator.xml.bz2
22:30:03:WU00:FS01:0x22:Reading tar file state.xml.bz2
22:30:03:WU00:FS01:0x22:Reading tar file system.xml.bz2
22:30:03:WU00:FS01:0x22:Digital signatures verified
22:30:03:WU00:FS01:0x22:Folding@home GPU Core22 Folding@home Core
22:30:03:WU00:FS01:0x22:Version 0.0.13
22:30:03:WU00:FS01:0x22:  Checkpoint write interval: 50000 steps (5%) [20 total]
22:30:03:WU00:FS01:0x22:  JSON viewer frame write interval: 10000 steps (1%) [100 total]
22:30:03:WU00:FS01:0x22:  XTC frame write interval: 250000 steps (25%) [4 total]
22:30:03:WU00:FS01:0x22:  Global context and integrator variables write interval: 25000 steps (2.5%) [40 total]
22:30:03:WU00:FS01:0x22:There are 3 platforms available.
22:30:03:WU00:FS01:0x22:Platform 0: Reference
22:30:03:WU00:FS01:0x22:Platform 1: CPU
22:30:03:WU00:FS01:0x22:Platform 2: OpenCL
22:30:03:WU00:FS01:0x22:  opencl-device 0 specified
22:30:16:WU00:FS01:0x22:Attempting to create OpenCL context:
22:30:16:WU00:FS01:0x22:  Configuring platform OpenCL
22:30:21:WU00:FS01:0x22:Failed to create OpenCL context:
22:30:21:WU00:FS01:0x22:Error compiling kernel: "C:\Users\RyJan\AppData\Local\Temp\OCL15776T24.cl", line 21: warning: OpenCL
22:30:21:WU00:FS01:0x22:          extension is now part of core
22:30:21:WU00:FS01:0x22:  #pragma OPENCL EXTENSION cl_khr_fp64 : enable
22:30:21:WU00:FS01:0x22:                           ^
22:30:21:WU00:FS01:0x22:
22:30:21:WU00:FS01:0x22:Error:E010:Irreducible ControlFlow Detected
22:30:21:WU00:FS01:0x22:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
22:30:21:WU00:FS01:0x22:Saving result file ..\logfile_01.txt
22:30:21:WU00:FS01:0x22:Saving result file science.log
22:30:21:WU00:FS01:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
22:30:21:WARNING:WU00:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
22:30:21:WU00:FS01:Sending unit results: id:00 state:SEND error:FAULTY project:13446 run:2389 clone:68 gen:0 core:0x22 unit:0x00000044000000000000348600000955
22:30:21:WU00:FS01:Uploading 2.74KiB to 54.157.202.86
22:30:21:WU00:FS01:Connecting to 54.157.202.86:8080
22:30:21:WU02:FS01:Connecting to assign1.foldingathome.org:80
22:30:21:WU00:FS01:Upload complete
22:30:21:WU00:FS01:Server responded WORK_ACK (400)
22:30:21:WU00:FS01:Cleaning up

[Joe_H]

Welcome to the folding support forum.

The R7 240 is still listed as supported, but towards the low end of the scale for a GPU. There are two possible issues that tI can think of that would cause the processing of the WU to fail. First is that the Project 13446 WUs are using a bit different set of simulation options, so sometimes fail a bit more often. The second may be related to the video driver. If Windows updated it through its update, then it will not have the OpenCL support properly installed, or possibly removed. Or the latest driver version may not support OpenCL processing on this old of a card.

[bruce]

I've seen a number of WUs with the following error messages

Code: Select all

OCL15776T24.cl", line 21: warning: OpenCL extension is now part of core
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
Error:E010:Irreducible ControlFlow Detected

In this timeframe, AMD released some GPUs that failed to support FP64 (hardware Double Precision) and I suspect the R7-240 was one of them. FAH does require that hardware instruction set and the only way I can think of to resove this issue is to reclassify that GPU as unsupported. I'm not an expert here so I'll wait for somebody with in-depth knowledge of the Mixed Precision code.

I think the distinction may be the difference between GPUs supporting Terrascale vs. GPUs supporting GCN. What are the lspci codes associated with your GPU and which architecture is it based on?

To quote from techpowerup's description:

Even though it supports DirectX 12, the feature level is only 11_1, which can be problematic with newer DirectX 12 titles.

[Joe_H]

The R7 240 is supposed to be based on a GCN 1st generation chip, and should support FP64. But there are several variants released by AMD and the other video card makers using AMD chips, not all of them have FP64 notations in their specs on TechPowerUp.com

[FishCZ]

I've seen a number of WUs with the following error messages

Code: Select all

OCL15776T24.cl", line 21: warning: OpenCL extension is now part of core
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
Error:E010:Irreducible ControlFlow Detected

In this timeframe, AMD released some GPUs that failed to support FP64 (hardware Double Precision) and I suspect the R7-240 was one of them. FAH does require that hardware instruction set and the only way I can think of to resove this issue is to reclassify that GPU as unsupported. I'm not an expert here so I'll wait for somebody with in-depth knowledge of the Mixed Precision code.

I think the distinction may be the difference between GPUs supporting Terrascale vs. GPUs supporting GCN. What are the lspci codes associated with your GPU and which architecture is it based on?

To quote from techpowerup's description:

Even though it supports DirectX 12, the feature level is only 11_1, which can be problematic with newer DirectX 12 titles.

The card was folding just fine for couple of days. so this is not gonnna be the root cause

Welcome to the folding support forum.

The R7 240 is still listed as supported, but towards the low end of the scale for a GPU. There are two possible issues that tI can think of that would cause the processing of the WU to fail. First is that the Project 13446 WUs are using a bit different set of simulation options, so sometimes fail a bit more often. The second may be related to the video driver. If Windows updated it through its update, then it will not have the OpenCL support properly installed, or possibly removed. Or the latest driver version may not support OpenCL processing on this old of a card.

I have installed drivers through AMD utility, yet I cant guarantee that Windows didnt messed it up. Please, can you point me to a guide on how to install/fix OpenCL? Or shall I just reinstall latest AMD drivers? Im noob in this area.
thanks

[bruce]

The card was folding just fine for couple of days. so this is not gonnna be the root cause

Thanks for the information. That's useful. Even knowing I was wrong is important to figuring out what was really going on.

[iero]

Could it be the age of the drivers and the issues that come with em?
viewtopic.php?f=19&t=36509

[Joe_H]

Usually reinstalling drivers downloaded from AMD will fix up any changes made by MS Windows Update. Occasionally the latest driver has a problem, so using the alternate driver links and selecting one just prior to that will work.

There are settings to keep the Windows updater from updating video drivers, a google search should find those. However the major Windows updates can reset those and start including drivers again.

[Foliant]

I cant help but I want to point out:

Code: Select all

[...]
22:30:03:WU00:FS01:0x22:Project: 13446 (Run 2389, Clone 68, Gen 0)
[...]
22:30:16:WU00:FS01:0x22:  Configuring platform OpenCL
22:30:21:WU00:FS01:0x22:Failed to create OpenCL context:
22:30:21:WU00:FS01:0x22:Error compiling kernel: "C:\Users\RyJan\AppData\Local\Temp\OCL15776T24.cl", line 21: warning: OpenCL
22:30:21:WU00:FS01:0x22:          extension is now part of core
22:30:21:WU00:FS01:0x22:  #pragma OPENCL EXTENSION cl_khr_fp64 : enable
22:30:21:WU00:FS01:0x22:                           ^
22:30:21:WU00:FS01:0x22:
22:30:21:WU00:FS01:0x22:Error:E010:Irreducible ControlFlow Detected
22:30:21:WU00:FS01:0x22:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
[...]

Seems to be "old AMD"-specific.
Im getting that too: viewtopic.php?f=19&t=36931

[FishCZ]

So, apparently it is folding again. Dunno why tho
What I did in past days:
I have checked the drivers through the AMD software which said that I am on latest drivers, however I have found that there is some OpenCL compatibility pack available from Windows store. Installation of this package did not help immediately. I did several restarts since its installation and the "Failed" error was returning over and over again. What was different the last time I shut it off, was that I have turned PC off without stopping F@H first. After a day off F@H started up with Windows logon and started folding right away on both CPU and GPU slots.
No idea whether it was that package or the turn it on and off again routine

[Foliant]

So, apparently it is folding again. Dunno why tho

I would bet its crunching not a WU of the following: 13446, 17433, 17434, 17435


R7 240
HD 7970
Whatever "Graphics/Compute: GFX6" means its on both of our GPUs. Theyre also both GCN 1.0
Maybe there is simply a bug or missing parts in the instruction set of GCN 1.0