GPU shows "failed"

It seems that a lot of GPU problems revolve around specific versions of drivers. Though AMD has their own support structure, you can often learn from information reported by others who fold.

Moderators: Site Moderators, FAHC Science Team

Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

GPU shows "failed"

Post by Peter_Hucker »

I have 4 machines with AMD GPUs:

1) R9 280X + RX 560
2) R9 280X
3) R9 280X
4) HD 7970

Two of the 280X cards show "failed" and aren't bothering to get another work unit, and there seems to be no way of telling them to?
One of those cards is perfect as I play games on it, and it's successfully done work units before.
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

I've found a workaround, I hope this doesn't happen too often. I deleted the slot for the GPU, restarted the client, and it detected the card again and got a new task.

Is a Radeon R9 280X too old for FAH?
JimboPalmer
Posts: 2573
Joined: Mon Feb 16, 2009 4:12 am
Location: Greenwood MS USA

Re: GPU shows "failed"

Post by JimboPalmer »

Welcome to Folding@Home!

https://www.techpowerup.com/gpu-specs/r ... 280x.c2398

This 9 year old design does meet the minimums F@H needs: it supports OpenCL 1.2 and does 64 bit floating point math.

It is possible that it is not supported in the newest drivers.

https://www.amd.com/en/support/graphics ... eon-r9-280
Tsar of all the Rushers
I tried to remain childlike, all I achieved was childish.
A friend to those who want no friends
toTOW
Site Moderator
Posts: 6296
Joined: Sun Dec 02, 2007 10:38 am
Location: Bordeaux, France
Contact:

Re: GPU shows "failed"

Post by toTOW »

Pause/Fold on the slot will remove the failed state and resume folding. But you have to find the root cause of errors. Please post the logs from FAHClient.
Image

Folding@Home beta tester since 2002. Folding Forum moderator since July 2008.
MeeLee
Posts: 1375
Joined: Tue Feb 19, 2019 10:16 pm

Re: GPU shows "failed"

Post by MeeLee »

Check temperatures and if the GPU is overclocked or not.
Not the best GPU to be folding on.
A single RX or RTX GPU will outdo all 4 of them, at a much lower energy cost.
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

I have installed the newest drivers available for that card. They aren't updating them any more. But they should work as well as they always did, not that I've ever folded on them before.

The log file doesn't seem to go back far enough to find the problem. It was only 24 hours ago but it's no longer in in c:\programdata\fahclient\log.txt
Joe_H
Site Admin
Posts: 7857
Joined: Tue Apr 21, 2009 4:41 pm
Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2
Location: W. MA

Re: GPU shows "failed"

Post by Joe_H »

By default the previous 16 log files are kept in the logs subdirectory.
Image

iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

Thanks, found it:

Code: Select all

03:22:12:WU03:FS02:0x22:*********************** Log Started 2022-02-16T03:22:12Z ***********************
03:22:12:WU03:FS02:0x22:*************************** Core22 Folding@home Core ***************************
03:22:12:WU03:FS02:0x22:       Core: Core22
03:22:12:WU03:FS02:0x22:       Type: 0x22
03:22:12:WU03:FS02:0x22:    Version: 0.0.20
03:22:12:WU03:FS02:0x22:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
03:22:12:WU03:FS02:0x22:  Copyright: 2020 foldingathome.org
03:22:12:WU03:FS02:0x22:   Homepage: https://foldingathome.org/
03:22:12:WU03:FS02:0x22:       Date: Jan 20 2022
03:22:12:WU03:FS02:0x22:       Time: 01:15:36
03:22:12:WU03:FS02:0x22:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
03:22:12:WU03:FS02:0x22:     Branch: HEAD
03:22:12:WU03:FS02:0x22:   Compiler: Visual C++
03:22:12:WU03:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
03:22:12:WU03:FS02:0x22:             -DOPENMM_VERSION="\"7.7.0\""
03:22:12:WU03:FS02:0x22:   Platform: win32 10
03:22:12:WU03:FS02:0x22:       Bits: 64
03:22:12:WU03:FS02:0x22:       Mode: Release
03:22:12:WU03:FS02:0x22:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
03:22:12:WU03:FS02:0x22:             <peastman@stanford.edu>
03:22:12:WU03:FS02:0x22:       Args: -dir 03 -suffix 01 -version 706 -lifeline 17264 -checkpoint 15
03:22:12:WU03:FS02:0x22:             -opencl-platform 0 -opencl-device 1 -gpu-vendor amd -gpu 1
03:22:12:WU03:FS02:0x22:             -gpu-usage 100
03:22:12:WU03:FS02:0x22:************************************ libFAH ************************************
03:22:12:WU03:FS02:0x22:       Date: Jan 20 2022
03:22:12:WU03:FS02:0x22:       Time: 01:14:17
03:22:12:WU03:FS02:0x22:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
03:22:12:WU03:FS02:0x22:     Branch: HEAD
03:22:12:WU03:FS02:0x22:   Compiler: Visual C++
03:22:12:WU03:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
03:22:12:WU03:FS02:0x22:   Platform: win32 10
03:22:12:WU03:FS02:0x22:       Bits: 64
03:22:12:WU03:FS02:0x22:       Mode: Release
03:22:12:WU03:FS02:0x22:************************************ CBang *************************************
03:22:12:WU03:FS02:0x22:       Date: Jan 20 2022
03:22:12:WU03:FS02:0x22:       Time: 01:13:20
03:22:12:WU03:FS02:0x22:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
03:22:12:WU03:FS02:0x22:     Branch: HEAD
03:22:12:WU03:FS02:0x22:   Compiler: Visual C++
03:22:12:WU03:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
03:22:12:WU03:FS02:0x22:   Platform: win32 10
03:22:12:WU03:FS02:0x22:       Bits: 64
03:22:12:WU03:FS02:0x22:       Mode: Release
03:22:12:WU03:FS02:0x22:************************************ System ************************************
03:22:12:WU03:FS02:0x22:        CPU: AMD Ryzen 9 3900XT 12-Core Processor
03:22:12:WU03:FS02:0x22:     CPU ID: AuthenticAMD Family 23 Model 113 Stepping 0
03:22:12:WU03:FS02:0x22:       CPUs: 24
03:22:12:WU03:FS02:0x22:     Memory: 63.93GiB
03:22:12:WU03:FS02:0x22:Free Memory: 48.91GiB
03:22:12:WU03:FS02:0x22:    Threads: WINDOWS_THREADS
03:22:12:WU03:FS02:0x22: OS Version: 6.2
03:22:12:WU03:FS02:0x22:Has Battery: true
03:22:12:WU03:FS02:0x22: On Battery: false
03:22:12:WU03:FS02:0x22: UTC Offset: 0
03:22:12:WU03:FS02:0x22:        PID: 11780
03:22:12:WU03:FS02:0x22:        CWD: C:\ProgramData\FAHClient\work
03:22:12:WU03:FS02:0x22:************************************ OpenMM ************************************
03:22:12:WU03:FS02:0x22:    Version: 7.7.0
03:22:12:WU03:FS02:0x22:********************************************************************************
03:22:12:WU03:FS02:0x22:Project: 18201 (Run 6407, Clone 1, Gen 11)
03:22:12:WU03:FS02:0x22:Reading tar file core.xml
03:22:12:WU03:FS02:0x22:Reading tar file integrator.xml
03:22:12:WU03:FS02:0x22:Reading tar file state.xml
03:22:13:WU03:FS02:0x22:Reading tar file system.xml
03:22:14:WU03:FS02:0x22:Digital signatures verified
03:22:14:WU03:FS02:0x22:Folding@home GPU Core22 Folding@home Core
03:22:14:WU03:FS02:0x22:Version 0.0.20
03:22:14:WU03:FS02:0x22:  Checkpoint write interval: 25000 steps (2%) [50 total]
03:22:14:WU03:FS02:0x22:  JSON viewer frame write interval: 12500 steps (1%) [100 total]
03:22:14:WU03:FS02:0x22:  XTC frame write interval: 20000 steps (1.6%) [62 total]
03:22:14:WU03:FS02:0x22:  Global context and integrator variables write interval: disabled
03:22:14:WU03:FS02:0x22:There are 3 platforms available.
03:22:14:WU03:FS02:0x22:Platform 0: Reference
03:22:14:WU03:FS02:0x22:Platform 1: CPU
03:22:14:WU03:FS02:0x22:Platform 2: OpenCL
03:22:14:WU03:FS02:0x22:  opencl-device 1 specified
03:22:42:WU03:FS02:0x22:Attempting to create OpenCL context:
03:22:42:WU03:FS02:0x22:  Configuring platform OpenCL
03:22:42:WU03:FS02:0x22:Failed to create OpenCL context:
03:22:42:WU03:FS02:0x22:Illegal value for DeviceIndex: 1
03:22:42:WU03:FS02:0x22:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
03:22:42:WU03:FS02:0x22:Saving result file ..\logfile_01.txt
03:22:42:WU03:FS02:0x22:Saving result file science.log
03:22:42:WU03:FS02:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
03:22:43:WARNING:WU03:FS02:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
03:22:43:WU03:FS02:Sending unit results: id:03 state:SEND error:FAULTY project:18201 run:6407 clone:1 gen:11 core:0x22 unit:0x000000010000000b0000471900001907
03:22:43:WU03:FS02:Uploading 2.56KiB to 128.252.203.11
03:22:43:WU03:FS02:Connecting to 128.252.203.11:8080
03:22:43:WU03:FS02:Upload complete
03:22:43:WU02:FS02:Connecting to assign1.foldingathome.org:80
03:22:43:WU03:FS02:Server responded WORK_ACK (400)
03:22:43:WU03:FS02:Cleaning up
03:22:44:WU02:FS02:Assigned to work server 54.157.202.86
03:22:44:WU02:FS02:Requesting new work unit for slot 02: gpu:40:0 Tahiti XT [Radeon R9 200/HD 7900/8970] from 54.157.202.86
03:22:44:WU02:FS02:Connecting to 54.157.202.86:8080
03:22:44:WU02:FS02:Downloading 344.66KiB
03:22:44:WU02:FS02:Download complete
03:22:44:WU02:FS02:Received Unit: id:02 state:DOWNLOAD error:NO_ERROR project:13463 run:283 clone:32 gen:0 core:0x22 unit:0x0000002000000000000034970000011b
03:22:44:WU02:FS02:Starting
03:22:44:WU02:FS02:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:\ProgramData\FAHClient\cores/cores.foldingathome.org/win/64bit/22-0.0.20/Core_22.fah/FahCore_22.exe -dir 02 -suffix 01 -version 706 -lifeline 11356 -checkpoint 15 -opencl-platform 0 -opencl-device 1 -gpu-vendor amd -gpu 1 -gpu-usage 100
03:22:44:WU02:FS02:Started FahCore on PID 14852
03:22:44:WU02:FS02:Core PID:18504
03:22:44:WU02:FS02:FahCore 0x22 started
Last edited by Joe_H on Thu Feb 17, 2022 3:25 pm, edited 1 time in total.
Reason: added code tags to log
prcowley
Posts: 23
Joined: Thu Jan 03, 2019 11:03 pm
Hardware configuration: Op Sys: Linux Ubuntu Studio 21.04 LTS
Kernal: 5.11.0-37-lowlatency
Proc: AMD Ryzen 7 1700 - 8-core
Mem: 32 GB
GPU: Nvidia GeForce GTX 1080Ti
Storage: 2 TB
Location: Gisborne, New Zealand
Contact:

Re: GPU shows "failed"

Post by prcowley »

I have had this issue multiple times and it always seems to be related to new Linux headers being updated and sometimes new Nvidia drivers. Something gets out of sync!
The easiest way I have found is to power down and restart until the client finds the GPU again. Sometimes it has taken two restart before sanity returns. It is a pain in the ass though. Fortunately it does not happen very often but it has taken a long time of searching and stumbling around to find this solution.
Pete Cowley, Gisborne, New Zealand. The first city to see the light of the new day. :D
Image
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

Yes, the computer in the log above is now behaving.

However I have a machine with a 7970 that looks like it's running but uses virtually no GPU and hardly progresses. Should a 7970 be ok? Same Tahiti chip, just marginally (5%) slower RAM.

This is it starting a task:

Code: Select all

10:04:57:WU02:FS02:0x22:*********************** Log Started 2022-02-17T10:04:57Z ***********************
10:04:57:WU02:FS02:0x22:*************************** Core22 Folding@home Core ***************************
10:04:57:WU02:FS02:0x22:       Core: Core22
10:04:57:WU02:FS02:0x22:       Type: 0x22
10:04:57:WU02:FS02:0x22:    Version: 0.0.20
10:04:57:WU02:FS02:0x22:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
10:04:57:WU02:FS02:0x22:  Copyright: 2020 foldingathome.org
10:04:57:WU02:FS02:0x22:   Homepage: https://foldingathome.org/
10:04:57:WU02:FS02:0x22:       Date: Jan 20 2022
10:04:57:WU02:FS02:0x22:       Time: 01:15:36
10:04:57:WU02:FS02:0x22:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
10:04:57:WU02:FS02:0x22:     Branch: HEAD
10:04:57:WU02:FS02:0x22:   Compiler: Visual C++
10:04:57:WU02:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
10:04:57:WU02:FS02:0x22:             -DOPENMM_VERSION="\"7.7.0\""
10:04:57:WU02:FS02:0x22:   Platform: win32 10
10:04:57:WU02:FS02:0x22:       Bits: 64
10:04:57:WU02:FS02:0x22:       Mode: Release
10:04:57:WU02:FS02:0x22:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
10:04:57:WU02:FS02:0x22:             <peastman@stanford.edu>
10:04:57:WU02:FS02:0x22:       Args: -dir 02 -suffix 01 -version 706 -lifeline 8036 -checkpoint 15
10:04:57:WU02:FS02:0x22:             -opencl-platform 1 -opencl-device 0 -gpu-vendor amd -gpu 0
10:04:57:WU02:FS02:0x22:             -gpu-usage 100
10:04:57:WU02:FS02:0x22:************************************ libFAH ************************************
10:04:57:WU02:FS02:0x22:       Date: Jan 20 2022
10:04:57:WU02:FS02:0x22:       Time: 01:14:17
10:04:57:WU02:FS02:0x22:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
10:04:57:WU02:FS02:0x22:     Branch: HEAD
10:04:57:WU02:FS02:0x22:   Compiler: Visual C++
10:04:57:WU02:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
10:04:57:WU02:FS02:0x22:   Platform: win32 10
10:04:57:WU02:FS02:0x22:       Bits: 64
10:04:57:WU02:FS02:0x22:       Mode: Release
10:04:57:WU02:FS02:0x22:************************************ CBang *************************************
10:04:57:WU02:FS02:0x22:       Date: Jan 20 2022
10:04:57:WU02:FS02:0x22:       Time: 01:13:20
10:04:57:WU02:FS02:0x22:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
10:04:57:WU02:FS02:0x22:     Branch: HEAD
10:04:57:WU02:FS02:0x22:   Compiler: Visual C++
10:04:57:WU02:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Zc:throwingNew /MT
10:04:57:WU02:FS02:0x22:   Platform: win32 10
10:04:57:WU02:FS02:0x22:       Bits: 64
10:04:57:WU02:FS02:0x22:       Mode: Release
10:04:57:WU02:FS02:0x22:************************************ System ************************************
10:04:57:WU02:FS02:0x22:        CPU: Intel(R) Core(TM) i5-8600K CPU @ 3.60GHz
10:04:57:WU02:FS02:0x22:     CPU ID: GenuineIntel Family 6 Model 158 Stepping 10
10:04:57:WU02:FS02:0x22:       CPUs: 6
10:04:57:WU02:FS02:0x22:     Memory: 15.84GiB
10:04:57:WU02:FS02:0x22:Free Memory: 12.06GiB
10:04:57:WU02:FS02:0x22:    Threads: WINDOWS_THREADS
10:04:57:WU02:FS02:0x22: OS Version: 6.2
10:04:57:WU02:FS02:0x22:Has Battery: false
10:04:57:WU02:FS02:0x22: On Battery: false
10:04:57:WU02:FS02:0x22: UTC Offset: 0
10:04:57:WU02:FS02:0x22:        PID: 8404
10:04:57:WU02:FS02:0x22:        CWD: C:\ProgramData\FAHClient\work
10:04:57:WU02:FS02:0x22:************************************ OpenMM ************************************
10:04:57:WU02:FS02:0x22:    Version: 7.7.0
10:04:57:WU02:FS02:0x22:********************************************************************************
10:04:57:WU02:FS02:0x22:Project: 13463 (Run 337, Clone 19, Gen 0)
10:04:57:WU02:FS02:0x22:Digital signatures verified
10:04:57:WU02:FS02:0x22:Folding@home GPU Core22 Folding@home Core
10:04:57:WU02:FS02:0x22:Version 0.0.20
10:04:57:WU02:FS02:0x22:  Checkpoint write interval: 50000 steps (5%) [20 total]
10:04:57:WU02:FS02:0x22:  JSON viewer frame write interval: 10000 steps (1%) [100 total]
10:04:57:WU02:FS02:0x22:  XTC frame write interval: 250000 steps (25%) [4 total]
10:04:57:WU02:FS02:0x22:  Global context and integrator variables write interval: 2500 steps (0.25%) [400 total]
10:04:57:WU02:FS02:0x22:There are 3 platforms available.
10:04:57:WU02:FS02:0x22:Platform 0: Reference
10:04:57:WU02:FS02:0x22:Platform 1: CPU
10:04:57:WU02:FS02:0x22:Platform 2: OpenCL
10:04:57:WU02:FS02:0x22:  opencl-device 0 specified
10:04:58:WU02:FS02:0x22:Attempting to create OpenCL context:
10:04:58:WU02:FS02:0x22:  Configuring platform OpenCL
10:05:13:WU02:FS02:0x22:  Using OpenCL on platformId 1 and gpu 0
10:05:13:WU02:FS02:0x22:Completed 0 out of 1000000 steps (0%)
10:05:13:WU02:FS02:0x22:Checkpoint completed at step 0
10:05:24:Removing old file 'configs/config-20220215-233639.xml'
10:05:24:Saving configuration to config.xml
10:05:24:<config>
10:05:24:  <!-- Network -->
10:05:24:  <proxy v=':8080'/>
10:05:24:
10:05:24:  <!-- Remote Command Server -->
10:05:24:  <password v='*****'/>
10:05:24:
10:05:24:  <!-- Slot Control -->
10:05:24:  <power v='FULL'/>
10:05:24:
10:05:24:  <!-- User Information -->
10:05:24:  <passkey v='*****'/>
10:05:24:  <team v='224497'/>
10:05:24:  <user v='PeterHucker_1HK9mWMp2xTK3f7fjowi1mCCbczu2EgFyR'/>
10:05:24:
10:05:24:  <!-- Folding Slots -->
10:05:24:  <slot id='0' type='CPU'>
10:05:24:    <cpus v='5'/>
10:05:24:    <paused v='true'/>
10:05:24:  </slot>
10:05:24:  <slot id='1' type='GPU'>
10:05:24:    <gpu-beta v='True'/>
10:05:24:    <opencl-index v='0'/>
10:05:24:    <paused v='true'/>
10:05:24:    <pci-bus v='0'/>
10:05:24:    <pci-slot v='2'/>
10:05:24:  </slot>
10:05:24:  <slot id='2' type='GPU'>
10:05:24:    <pci-bus v='1'/>
10:05:24:    <pci-slot v='0'/>
10:05:24:  </slot>
10:05:24:</config>
Last edited by Joe_H on Fri Feb 18, 2022 2:39 am, edited 1 time in total.
Reason: added code tags to log
gunnarre
Posts: 567
Joined: Sun May 24, 2020 7:23 pm
Location: Norway

Re: GPU shows "failed"

Post by gunnarre »

Tahiti is GCN 1.0. AMD has dropped OpenCL support for GCN 1.0 cards, even in Windows - those cards have been unsupported in Linux for years. You might have to roll back the AMD drivers to an older version, and even then I don't think the performance will be good. I retired my Radeon 7770 HD (Cape Verde, also GCN 1.0 architecture) due to this reason, and because it was being outperformed by even CPUs now.
Image
Online: GTX 1660 Super, GTX 1080, GTX 1050 Ti 4G OC, RX580 + occasional CPU folding in the cold.
Offline: Radeon HD 7770, GTX 960, GTX 950
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

Dropped support? I don't care, all that means is I get no updates. The latest version of the driver for that card should work fine. And seems to half the time on FAH.

And I actually just bought the 7970, because it's the best double precision card you can get! I ain't having any of this 1:16 or 1:32 nonsense ratios!

I can run Milkyway at Home on it about 80 times faster than a decent modern CPU.
Joe_H
Site Admin
Posts: 7857
Joined: Tue Apr 21, 2009 4:41 pm
Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2
Location: W. MA

Re: GPU shows "failed"

Post by Joe_H »

Peter_Hucker wrote:Dropped support? I don't care, all that means is I get no updates. The latest version of the driver for that card should work fine. And seems to half the time on FAH.

And I actually just bought the 7970, because it's the best double precision card you can get! I ain't having any of this 1:16 or 1:32 nonsense ratios!
AMD has had a bad habit of "latest driver" for a card sometimes having buggy support for OpenCL. Going back a version or two has sometimes been required. The last driver updates for Tahiti based GPUs were last year.

As for double precision, F@h uses a relatively small amount. Most calculations are done in single precision, so the 7970 only provides a slightly faster calculation.
Image

iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Peter_Hucker
Posts: 306
Joined: Wed Feb 16, 2022 1:18 am

Re: GPU shows "failed"

Post by Peter_Hucker »

I don't have the time to mess around with versions. As long as something runs on it I'm just leaving it. I've not had a problem with any Boinc GPU program using OpenCL.

I don't understand this double precision thing - FAH requires it, but you say it's hardly used? If it's hardly used, you can do double calculations using three single ones. Also, I've never heard of a card not having double, they just have varying speeds of it.
Joe_H
Site Admin
Posts: 7857
Joined: Tue Apr 21, 2009 4:41 pm
Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2
Location: W. MA

Re: GPU shows "failed"

Post by Joe_H »

Double precision is used in a few spots where needed to carry enough precision in critical calculations. They are a fraction of the total calculations done in the F@h GPU Core_22. You can emulate the double precision using singles, but it takes more than just three. Emulating the double precision work has been tried in the past, it was a lot slower than native double precision even with GPUs that process it at a small fraction of their single precision rate.

As for cards without double precision, go back 10-12 years and many didn't have it. And with the integrated GPUs that was even more recent. It was only a few years ago that AMD included double precision support in their iGPUs. Intel had it in their iGPUs 10 years ago, but it is only the ones released starting about 6 years ago that were fast enough for low end GPU work for F@h. Intel has since dropped double precision support in hardware on their latest generation iGPUs. With discrete GPUs Nvidia went to all cards having double precision before AMD. It wasn't until the Rn 300 series cards from AMD that the entire series included double precision, and even then that was not the case for the mobile versions.
Image

iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Post Reply