Using MPS to dramatically increase PPD on big GPUs (Linux guide)

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

Moderators: Site Moderators, FAHC Science Team

Post Reply
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

muziqaz wrote: ↑Wed Oct 15, 2025 8:22 am The test sequence I wrote few posts back will tell you if you are gaining anything from MPS initiative or not
I tried to replicate your test procedure. Unfortunately, it doesn't work and I'm not a experienced Linux user.
I copied the WU and the core to my desktop folder "mpstest0" and mpstest0\00
Then entered the commands:

Code: Select all

myxubuntu@myxubuntu:~/Desktop/mpstest0$ ./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 0 -cuda-device 0 -gpu 0 --log-time=true
./FahCore_22: error while loading shared libraries: libOpenMM.so.7.7: cannot open shared object file: No such file or directory
After chatbotting, I entered the following command:

Code: Select all

myxubuntu@myxubuntu:~/Desktop/mpstest0$ export LD_LIBRARY_PATH=.
After that, I tried the command again:

Code: Select all

myxubuntu@myxubuntu:~/Desktop/mpstest0$ ./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 0 -cuda-device 0 -gpu 0 --log-time=true
WARNING:Ignoring invalid or unsupported argument '-cuda-platform'
WARNING:Ignoring invalid or unsupported argument '0'
20:49:30:Trying to acquire exclusive lock on unit 01
20:49:30:********************** Core22 Folding@home Core ***********************
20:49:30:       Core: Core22
20:49:30:       Type: 0x22
20:49:30:    Version: 0.0.20
20:49:30:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
20:49:30:  Copyright: 2020 foldingathome.org
20:49:30:   Homepage: https://foldingathome.org/
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:52
20:49:30:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie
20:49:30:             -DOPENMM_VERSION="\"7.7.0\""
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
20:49:30:             <peastman@stanford.edu>
20:49:30:       Args: -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 0 -cuda-device
20:49:30:             0 -gpu 0 --log-time=true
20:49:30:******************************* libFAH ********************************
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:22
20:49:30:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:******************************** CBang ********************************
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:00
20:49:30:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:******************************* System ********************************
20:49:30:        CPU: AMD Ryzen 9 5900X 12-Core Processor
20:49:30:     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
20:49:30:       CPUs: 24
20:49:30:     Memory: 31.27GiB
20:49:30:Free Memory: 22.91GiB
20:49:30:    Threads: POSIX_THREADS
20:49:30: OS Version: 6.14
20:49:30:Has Battery: false
20:49:30: On Battery: false
20:49:30: UTC Offset: 2
20:49:30:        PID: 5275
20:49:30:        CWD: /home/myxubuntu/Desktop/mpstest0
20:49:30:******************************* OpenMM ********************************
20:49:30:    Version: 7.7.0
20:49:30:***********************************************************************
20:49:30:Project: 14955 (Run 5, Clone 159, Gen 24)
20:49:30:Reading tar file core.xml
20:49:30:Reading tar file integrator.xml
20:49:30:Reading tar file state.xml
20:49:30:Reading tar file system.xml
20:49:31:Digital signatures verified
It seemed to start, however after "Digital signature verified" it stops and command line is "back"
I deleted the "-cuda-platform 0" option, however the same result:

Code: Select all

myxubuntu@myxubuntu:~/Desktop/mpstest0$ ./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-device 0 -gpu 0 --log-time=true
20:57:26:Trying to acquire exclusive lock on unit 01
20:57:26:********************** Core22 Folding@home Core ***********************
20:57:26:       Core: Core22
20:57:26:       Type: 0x22
20:57:26:    Version: 0.0.20
20:57:26:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
20:57:26:  Copyright: 2020 foldingathome.org
20:57:26:   Homepage: https://foldingathome.org/
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:52
20:57:26:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie
20:57:26:             -DOPENMM_VERSION="\"7.7.0\""
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
20:57:26:             <peastman@stanford.edu>
20:57:26:       Args: -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-device 0 -gpu 0
20:57:26:             --log-time=true
20:57:26:******************************* libFAH ********************************
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:22
20:57:26:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:******************************** CBang ********************************
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:00
20:57:26:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:******************************* System ********************************
20:57:26:        CPU: AMD Ryzen 9 5900X 12-Core Processor
20:57:26:     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
20:57:26:       CPUs: 24
20:57:26:     Memory: 31.27GiB
20:57:26:Free Memory: 22.93GiB
20:57:26:    Threads: POSIX_THREADS
20:57:26: OS Version: 6.14
20:57:26:Has Battery: false
20:57:26: On Battery: false
20:57:26: UTC Offset: 2
20:57:26:        PID: 5417
20:57:26:        CWD: /home/myxubuntu/Desktop/mpstest0
20:57:26:******************************* OpenMM ********************************
20:57:26:    Version: 7.7.0
20:57:26:***********************************************************************
20:57:26:Project: 14955 (Run 5, Clone 159, Gen 24)
20:57:26:Reading tar file core.xml
20:57:26:Reading tar file integrator.xml
20:57:26:Reading tar file state.xml
20:57:27:Reading tar file system.xml
20:57:27:Digital signatures verified
And.. back to command line.

Here's the log files with two attempts:

Code: Select all

*********************** Log Started 2025-10-16T20:49:30Z ***********************
20:49:30:********************** Core22 Folding@home Core ***********************
20:49:30:       Core: Core22
20:49:30:       Type: 0x22
20:49:30:    Version: 0.0.20
20:49:30:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
20:49:30:  Copyright: 2020 foldingathome.org
20:49:30:   Homepage: https://foldingathome.org/
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:52
20:49:30:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie
20:49:30:             -DOPENMM_VERSION="\"7.7.0\""
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
20:49:30:             <peastman@stanford.edu>
20:49:30:       Args: -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 0 -cuda-device
20:49:30:             0 -gpu 0 --log-time=true
20:49:30:******************************* libFAH ********************************
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:22
20:49:30:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:******************************** CBang ********************************
20:49:30:       Date: Jan 20 2022
20:49:30:       Time: 00:57:00
20:49:30:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
20:49:30:     Branch: HEAD
20:49:30:   Compiler: GNU 9.4.0
20:49:30:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:49:30:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
20:49:30:   Platform: linux 5.11.0-1025-azure
20:49:30:       Bits: 64
20:49:30:       Mode: Release
20:49:30:******************************* System ********************************
20:49:30:        CPU: AMD Ryzen 9 5900X 12-Core Processor
20:49:30:     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
20:49:30:       CPUs: 24
20:49:30:     Memory: 31.27GiB
20:49:30:Free Memory: 22.91GiB
20:49:30:    Threads: POSIX_THREADS
20:49:30: OS Version: 6.14
20:49:30:Has Battery: false
20:49:30: On Battery: false
20:49:30: UTC Offset: 2
20:49:30:        PID: 5275
20:49:30:        CWD: /home/myxubuntu/Desktop/mpstest0
20:49:30:******************************* OpenMM ********************************
20:49:30:    Version: 7.7.0
20:49:30:***********************************************************************
20:49:30:Project: 14955 (Run 5, Clone 159, Gen 24)
20:49:30:Reading tar file core.xml
20:49:30:Reading tar file integrator.xml
20:49:30:Reading tar file state.xml
20:49:30:Reading tar file system.xml
20:49:31:Digital signatures verified
20:49:31:Folding@home GPU Core22 Folding@home Core
20:49:31:Version 0.0.20
20:49:31:  Checkpoint write interval: 50000 steps (5%) [20 total]
20:49:31:  JSON viewer frame write interval: 10000 steps (1%) [100 total]
20:49:31:  XTC frame write interval: 25000 steps (2.5%) [40 total]
20:49:31:  Global context and integrator variables write interval: disabled
20:49:31:No -opencl-device specified; using deprecated -gpu argument as an alias for -opencl-device.
20:49:31:Please consider upgrading your client version.
20:49:31:Error loading plugin libOpenMMCUDA.so: Error loading library /home/myxubuntu/Desktop/mpstest0/libOpenMMCUDA.so: libcufft.so.10: cannot open shared object file: No such file or directory
20:49:31:Error loading plugin libOpenMMCudaCompiler.so: Error loading library /home/myxubuntu/Desktop/mpstest0/libOpenMMCudaCompiler.so: libnvrtc.so.11.2: cannot open shared object file: No such file or directory
20:49:31:There are 3 platforms available.
20:49:31:Platform 0: Reference
20:49:31:Platform 1: CPU
20:49:31:Platform 2: OpenCL
20:49:31:  opencl-device 0 specified
20:49:31:cuda-device was set but CUDA platform could not be found.
20:49:40:Attempting to create OpenCL context:
20:49:40:  Configuring platform OpenCL
20:49:40:Failed to create OpenCL context:
20:49:40:Error initializing context: clCreateContext (-9999)
20:49:40:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
20:49:40:Saving result file ../logfile_01.txt
20:49:40:Saving result file science.log
20:49:40:Folding@home Core Shutdown: BAD_WORK_UNIT
*********************** Log Started 2025-10-16T20:57:26Z ***********************
20:57:26:********************** Core22 Folding@home Core ***********************
20:57:26:       Core: Core22
20:57:26:       Type: 0x22
20:57:26:    Version: 0.0.20
20:57:26:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
20:57:26:  Copyright: 2020 foldingathome.org
20:57:26:   Homepage: https://foldingathome.org/
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:52
20:57:26:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie
20:57:26:             -DOPENMM_VERSION="\"7.7.0\""
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
20:57:26:             <peastman@stanford.edu>
20:57:26:       Args: -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-device 0 -gpu 0
20:57:26:             --log-time=true
20:57:26:******************************* libFAH ********************************
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:22
20:57:26:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:******************************** CBang ********************************
20:57:26:       Date: Jan 20 2022
20:57:26:       Time: 00:57:00
20:57:26:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
20:57:26:     Branch: HEAD
20:57:26:   Compiler: GNU 9.4.0
20:57:26:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
20:57:26:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
20:57:26:   Platform: linux 5.11.0-1025-azure
20:57:26:       Bits: 64
20:57:26:       Mode: Release
20:57:26:******************************* System ********************************
20:57:26:        CPU: AMD Ryzen 9 5900X 12-Core Processor
20:57:26:     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
20:57:26:       CPUs: 24
20:57:26:     Memory: 31.27GiB
20:57:26:Free Memory: 22.93GiB
20:57:26:    Threads: POSIX_THREADS
20:57:26: OS Version: 6.14
20:57:26:Has Battery: false
20:57:26: On Battery: false
20:57:26: UTC Offset: 2
20:57:26:        PID: 5417
20:57:26:        CWD: /home/myxubuntu/Desktop/mpstest0
20:57:26:******************************* OpenMM ********************************
20:57:26:    Version: 7.7.0
20:57:26:***********************************************************************
20:57:26:Project: 14955 (Run 5, Clone 159, Gen 24)
20:57:26:Reading tar file core.xml
20:57:26:Reading tar file integrator.xml
20:57:26:Reading tar file state.xml
20:57:27:Reading tar file system.xml
20:57:27:Digital signatures verified
20:57:27:Folding@home GPU Core22 Folding@home Core
20:57:27:Version 0.0.20
20:57:28:  Checkpoint write interval: 50000 steps (5%) [20 total]
20:57:28:  JSON viewer frame write interval: 10000 steps (1%) [100 total]
20:57:28:  XTC frame write interval: 25000 steps (2.5%) [40 total]
20:57:28:  Global context and integrator variables write interval: disabled
20:57:28:No -opencl-device specified; using deprecated -gpu argument as an alias for -opencl-device.
20:57:28:Please consider upgrading your client version.
20:57:28:Error loading plugin libOpenMMCUDA.so: Error loading library /home/myxubuntu/Desktop/mpstest0/libOpenMMCUDA.so: libcufft.so.10: cannot open shared object file: No such file or directory
20:57:28:Error loading plugin libOpenMMCudaCompiler.so: Error loading library /home/myxubuntu/Desktop/mpstest0/libOpenMMCudaCompiler.so: libnvrtc.so.11.2: cannot open shared object file: No such file or directory
20:57:28:There are 3 platforms available.
20:57:28:Platform 0: Reference
20:57:28:Platform 1: CPU
20:57:28:Platform 2: OpenCL
20:57:28:  opencl-device 0 specified
20:57:28:cuda-device was set but CUDA platform could not be found.
20:57:37:Attempting to create OpenCL context:
20:57:37:  Configuring platform OpenCL
20:57:37:Failed to create OpenCL context:
20:57:37:Error initializing context: clCreateContext (-9999)
20:57:37:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
20:57:37:Saving result file ../logfile_01.txt
20:57:37:Saving result file science.log
20:57:37:Folding@home Core Shutdown: BAD_WORK_UNIT
Here's the logfile for the next WU I'm folding:

Code: Select all

*********************** Log Started 2025-10-16T21:12:23Z ***********************
*************************** Core27 Folding@home Core ***************************
       Core: Core27
       Type: 0x27
    Version: 8.2.1
     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
  Copyright: 2022 foldingathome.org
   Homepage: https://foldingathome.org/
       Date: Mar 27 2025
       Time: 04:28:24
   Revision: 84019442387121f2bed4ed52e05c53b7b7b52a72
     Branch: HEAD
   Compiler: GNU 7.5.0
    Options: -faligned-new -std=c++14 -fsigned-char -ffunction-sections
             -fdata-sections -O3 -funroll-loops -fno-pie
             -DOPENMM_VERSION="\"8.2.0\""
   Platform: linux 6.8.0-1021-azure
       Bits: 64
       Mode: Release
Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
             <peastman@stanford.edu>
       Args: -dir _O2QuDg4RbgBbI1oQDMsFV2ufMai1KqX3t8QYfzlTAc -suffix 01
             -version 8.4.9 -lifeline 5759 -gpu-uuid
             82ebfe35-4873-868a-4b21-5d782ea14f54 -gpu-platform cuda -gpu-vendor
             nvidia -cuda-platform 0 -cuda-device 0
************************************ libFAH ************************************
       Date: Mar 27 2025
       Time: 04:23:18
   Revision: c7d2824a47eb025fa8cda8968c7a5e971585d90c
     Branch: HEAD
   Compiler: GNU 7.5.0
    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
             -fdata-sections -O3 -funroll-loops -fno-pie
   Platform: linux 6.8.0-1021-azure
       Bits: 64
       Mode: Release
************************************ CBang *************************************
    Version: 1.7.2
     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
        Org: Cauldron Development LLC
  Copyright: Cauldron Development LLC, 2003-2024
   Homepage: https://cauldrondevelopment.com/
    License: LGPL-2.1-or-later
       Date: Mar 27 2025
       Time: 04:22:53
   Revision: f1cd4c791e8c40a35dcfeab3ab85d910949cc0cb
     Branch: HEAD
   Compiler: GNU 7.5.0
    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
   Platform: linux 6.8.0-1021-azure
       Bits: 64
       Mode: Release
************************************ System ************************************
        CPU: AMD Ryzen 9 5900X 12-Core Processor
     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
       CPUs: 24
     Memory: 31.27GiB
Free Memory: 25.17GiB
 OS Version: 6.14
Has Battery: false
 On Battery: false
   Hostname: myxubuntu
 UTC Offset: 2
        PID: 5769
        CWD: /var/lib/fah-client/work
       Exec: /var/lib/fah-client/cores/openmm-core-27/centos-7.9.2009-64bit/release/fahcore-27-centos-7.9.2009-64bit-release-8.2.1/FahCore_27
************************************ OpenMM ************************************
    Version: 8.2.0
********************************************************************************
Project: 15303 (Run 195, Clone 0, Gen 20)
Reading tar file core.xml
Reading tar file integrator.xml
Reading tar file state.xml
Reading tar file system.xml
Digital signatures verified
Folding@home GPU Core27 Folding@home Core
Version 8.2.1
  GPU info: Platform: CUDA
  GPU info: PlatformIndex: 0
  GPU info: Device: NVIDIA GeForce RTX 5080
  GPU info: DeviceIndex: 0
  GPU info: Vendor: 0x10de
  GPU info: PCI: 43:00:00
  GPU info: Compute: 12.0
  GPU info: Driver: 13.0
  GPU info: GPU: true
  Checkpoint write interval: 125000 steps (5%) [20 total]
  JSON viewer frame write interval: 25000 steps (1%) [100 total]
  XTC frame write interval: 50000 steps (2%) [50 total]
  TRR frame write interval: disabled
  Global context and integrator variables write interval: disabled
There are 4 platforms available.
Platform 0: Reference
Platform 1: CPU
Platform 2: OpenCL
Platform 3: CUDA
  cuda-device 0 specified
Attempting to create CUDA context:
  Configuring platform CUDA
  Using CUDA on CUDA Platform and gpu 0
  GPU info: Platform: CUDA
  GPU info: PlatformIndex: 0
  GPU info: Device: NVIDIA GeForce RTX 5080
  GPU info: DeviceIndex: 0
  GPU info: Vendor: 0x10de
  GPU info: PCI: 43:00:00
  GPU info: Compute: 12.0
  GPU info: Driver: 13.0
  GPU info: GPU: true
Completed 0 out of 2500000 steps (0%)
Checkpoint completed at step 0
Completed 25000 out of 2500000 steps (1%)
Completed 50000 out of 2500000 steps (2%)
Completed 75000 out of 2500000 steps (3%)
Completed 100000 out of 2500000 steps (4%)
Completed 125000 out of 2500000 steps (5%)
Checkpoint completed at step 125000
Completed 150000 out of 2500000 steps (6%)
Completed 175000 out of 2500000 steps (7%)
Completed 200000 out of 2500000 steps (8%)
Completed 225000 out of 2500000 steps (9%)
Completed 250000 out of 2500000 steps (10%)
Checkpoint completed at step 250000
Completed 275000 out of 2500000 steps (11%)
Completed 300000 out of 2500000 steps (12%)
Completed 325000 out of 2500000 steps (13%)
Completed 350000 out of 2500000 steps (14%)
Completed 375000 out of 2500000 steps (15%)
Checkpoint completed at step 375000
Completed 400000 out of 2500000 steps (16%)
Completed 425000 out of 2500000 steps (17%)
Completed 450000 out of 2500000 steps (18%)
Completed 475000 out of 2500000 steps (19%)
Completed 500000 out of 2500000 steps (20%)
Checkpoint completed at step 500000
Completed 525000 out of 2500000 steps (21%)
Completed 550000 out of 2500000 steps (22%)
Completed 575000 out of 2500000 steps (23%)
Completed 600000 out of 2500000 steps (24%)
Completed 625000 out of 2500000 steps (25%)
Checkpoint completed at step 625000
Completed 650000 out of 2500000 steps (26%)
Last edited by foldinghomealone on Thu Oct 16, 2025 9:26 pm, edited 1 time in total.
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

My bad.
Give me a second, I will edit my instruction post with additional info. You nearly got there, though :D
FAH Omega tester
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

OK, I added few bits, changed example core to core22.
Basically what I forgot was replace problematic libstdc++.so.6 with symlink to same library (but newer) which resides in your system.
Second thing which I forgot is to expose your test folder content to the core.
In my head, everything now should work :D key word being: should
FAH Omega tester
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

Thanks. However, it still doesn't work
Again: the unkown argument: -cuda-platform 0
Then it stops as before:

Code: Select all

*********************** Log Started 2025-10-16T22:06:11Z ***********************
22:06:11:********************** Core22 Folding@home Core ***********************
22:06:11:       Core: Core22
22:06:11:       Type: 0x22
22:06:11:    Version: 0.0.20
22:06:11:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
22:06:11:  Copyright: 2020 foldingathome.org
22:06:11:   Homepage: https://foldingathome.org/
22:06:11:       Date: Jan 20 2022
22:06:11:       Time: 00:57:52
22:06:11:   Revision: 3f211b8a4346514edbff34e3cb1c0e0ec951373c
22:06:11:     Branch: HEAD
22:06:11:   Compiler: GNU 9.4.0
22:06:11:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
22:06:11:             -fdata-sections -O3 -funroll-loops -fno-pie
22:06:11:             -DOPENMM_VERSION="\"7.7.0\""
22:06:11:   Platform: linux 5.11.0-1025-azure
22:06:11:       Bits: 64
22:06:11:       Mode: Release
22:06:11:Maintainers: John Chodera <john.chodera@choderalab.org> and Peter Eastman
22:06:11:             <peastman@stanford.edu>
22:06:11:       Args: -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 0 -cuda-device
22:06:11:             0 -gpu 0 --log-time=true
22:06:11:******************************* libFAH ********************************
22:06:11:       Date: Jan 20 2022
22:06:11:       Time: 00:57:22
22:06:11:   Revision: 9f4ad694e75c2350d4bb6b8b5b769ba27e483a2f
22:06:11:     Branch: HEAD
22:06:11:   Compiler: GNU 9.4.0
22:06:11:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
22:06:11:             -fdata-sections -O3 -funroll-loops -fno-pie
22:06:11:   Platform: linux 5.11.0-1025-azure
22:06:11:       Bits: 64
22:06:11:       Mode: Release
22:06:11:******************************** CBang ********************************
22:06:11:       Date: Jan 20 2022
22:06:11:       Time: 00:57:00
22:06:11:   Revision: ab023d155b446906d55b0f6c9a1eedeea04f7a1a
22:06:11:     Branch: HEAD
22:06:11:   Compiler: GNU 9.4.0
22:06:11:    Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
22:06:11:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
22:06:11:   Platform: linux 5.11.0-1025-azure
22:06:11:       Bits: 64
22:06:11:       Mode: Release
22:06:11:******************************* System ********************************
22:06:11:        CPU: AMD Ryzen 9 5900X 12-Core Processor
22:06:11:     CPU ID: AuthenticAMD Family 25 Model 33 Stepping 0
22:06:11:       CPUs: 24
22:06:11:     Memory: 31.27GiB
22:06:11:Free Memory: 24.90GiB
22:06:11:    Threads: POSIX_THREADS
22:06:11: OS Version: 6.14
22:06:11:Has Battery: false
22:06:11: On Battery: false
22:06:11: UTC Offset: 2
22:06:11:        PID: 13076
22:06:11:        CWD: /home/myxubuntu/Desktop/mps_test
22:06:11:******************************* OpenMM ********************************
22:06:11:    Version: 7.7.0
22:06:11:***********************************************************************
22:06:11:Project: 14955 (Run 5, Clone 159, Gen 24)
22:06:11:Reading tar file core.xml
22:06:11:Reading tar file integrator.xml
22:06:11:Reading tar file state.xml
22:06:11:Reading tar file system.xml
22:06:12:Digital signatures verified
22:06:12:Folding@home GPU Core22 Folding@home Core
22:06:12:Version 0.0.20
22:06:12:  Checkpoint write interval: 50000 steps (5%) [20 total]
22:06:12:  JSON viewer frame write interval: 10000 steps (1%) [100 total]
22:06:12:  XTC frame write interval: 25000 steps (2.5%) [40 total]
22:06:12:  Global context and integrator variables write interval: disabled
22:06:12:No -opencl-device specified; using deprecated -gpu argument as an alias for -opencl-device.
22:06:12:Please consider upgrading your client version.
22:06:12:There are 4 platforms available.
22:06:12:Platform 0: Reference
22:06:12:Platform 1: CPU
22:06:12:Platform 2: OpenCL
22:06:12:  opencl-device 0 specified
22:06:12:Platform 3: CUDA
22:06:12:  cuda-device 0 specified
22:06:21:Attempting to create CUDA context:
22:06:21:  Configuring platform CUDA
22:06:21:Failed to create CUDA context:
22:06:21:The requested CUDA device could not be loaded
22:06:21:Attempting to create OpenCL context:
22:06:21:  Configuring platform OpenCL
22:06:21:Failed to create OpenCL context:
22:06:21:Error initializing context: clCreateContext (-9999)
22:06:21:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
22:06:21:Saving result file ../logfile_01.txt
22:06:21:Saving result file science.log
22:06:21:Folding@home Core Shutdown: BAD_WORK_UNIT
In the log it states that platform 3 is CUDA.

I have to go to bed now. Hope, I can further test tomorrow.
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

Try:
export LD_LIBRARY_PATH="$PWD"
./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -cuda-platform 1 -cuda-device 1 -gpu 1 --log-time=true
It would be good time for my Nvidia wielding colleagues to chime in :D
FAH Omega tester
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

It doesn't work and I don't have any idea.
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

foldinghomealone wrote: ↑Fri Oct 17, 2025 4:16 pm It doesn't work and I don't have any idea.
Post full system specs. How many GPUs you have, which PCI e slots they are slotted in. I'm running blind here
FAH Omega tester
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

OK, 3rd time is the charm:
./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device 0 -gpu 0 --log-time=true
FAH Omega tester
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

Thank you for this question, now I'm puzzled...
It should be: RTX 5080 4.0x16

However, PCIe connection seems to be different:

Code: Select all

myxubuntu@myxubuntu:/sys/bus/pci/devices$ cat /sys/bus/pci/devices/0000:2b:00.0/current_link_speed
16.0 GT/s PCIe
myxubuntu@myxubuntu:/sys/bus/pci/devices$ cat /sys/bus/pci/devices/0000:2b:00.0/current_link_width
4
myxubuntu@myxubuntu:/sys/bus/pci/devices$ cat /sys/bus/pci/devices/0000:2b:00.0/max_link_speed
32.0 GT/s PCIe
myxubuntu@myxubuntu:/sys/bus/pci/devices$ cat /sys/bus/pci/devices/0000:2b:00.0/max_link_width
16

Code: Select all

System:
  Host: myxubuntu Kernel: 6.14.0-33-generic arch: x86_64 bits: 64
  Desktop: Xfce v: 4.18.1 Distro: Xubuntu 24.04.3 LTS (Noble Numbat)
Machine:
  Type: Desktop Mobo: Micro-Star model: MPG B550 GAMING EDGE WIFI (MS-7C91)
    v: 1.0 serial: <superuser required> UEFI: American Megatrends LLC. v: 1.K1
    date: 09/10/2025
CPU:
  Info: 12-core model: AMD Ryzen 9 5900X bits: 64 type: MT MCP cache:
    L2: 6 MiB
  Speed (MHz): avg: 2042 min/max: 550/4954 cores: 1: 1731 2: 3603 3: 1731
    4: 1731 5: 1731 6: 1731 7: 1731 8: 1731 9: 1731 10: 1731 11: 1731 12: 1731
    13: 1731 14: 3609 15: 1731 16: 1731 17: 1731 18: 1731 19: 3597 20: 1731
    21: 3600 22: 1731 23: 1731 24: 1731
Graphics:
  Device-1: NVIDIA driver: nvidia v: 580.65.06
  Device-2: Logitech HD Webcam C525 driver: snd-usb-audio,uvcvideo type: USB
  Display: x11 server: X.Org v: 21.1.11 driver: X: loaded: nvidia
    unloaded: fbdev,modesetting,nouveau,vesa
    gpu: nv_platform,nvidia,nvidia-nvswitch resolution: 2560x1440~180Hz
  API: EGL v: 1.5 drivers: nouveau,nvidia,swrast
    platforms: gbm,x11,surfaceless,device
  API: OpenGL v: 4.6.0 compat-v: 4.5 vendor: nvidia mesa v: 580.65.06
    renderer: NVIDIA GeForce RTX 5080/PCIe/SSE2
Audio:
  Device-1: NVIDIA driver: snd_hda_intel
  Device-2: AMD Starship/Matisse HD Audio driver: snd_hda_intel
  Device-3: Logitech HD Webcam C525 driver: snd-usb-audio,uvcvideo type: USB
  API: ALSA v: k6.14.0-33-generic status: kernel-api
  Server-1: PipeWire v: 1.0.5 status: active
Network:
  Device-1: Intel Wi-Fi 6 AX200 driver: iwlwifi
  IF: wlo1 state: down mac: a8:7e:ea:34:0a:0d
  Device-2: Realtek RTL8125 2.5GbE driver: r8169
  IF: enp42s0 state: up speed: 1000 Mbps duplex: full mac: 2c:f0:5d:55:f3:1a
Bluetooth:
  Device-1: Intel AX200 Bluetooth driver: btusb type: USB
  Report: hciconfig ID: hci0 state: up address: A8:7E:EA:34:0A:11 bt-v: 5.2
Drives:
  Local Storage: total: 1.86 TiB used: 19.97 GiB (1.0%)
  ID-1: /dev/nvme0n1 vendor: Kingston model: SKC3000D2048G size: 1.86 TiB
Partition:
  ID-1: / size: 105.39 GiB used: 19.94 GiB (18.9%) fs: ext4
    dev: /dev/nvme0n1p7
  ID-2: /boot/efi size: 196 MiB used: 37 MiB (18.9%) fs: vfat
    dev: /dev/nvme0n1p1
Swap:
  Alert: No swap data was found.
Sensors:
  System Temperatures: cpu: 44.5 C mobo: N/A gpu: nvidia temp: 41 C
  Fan Speeds (rpm): N/A gpu: nvidia fan: 0%
Info:
  Memory: total: 32 GiB available: 31.27 GiB used: 2.33 GiB (7.4%)
  Processes: 466 Uptime: 1m Shell: Bash inxi: 3.3.34
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

muziqaz wrote: ↑Fri Oct 17, 2025 9:07 pm OK, 3rd time is the charm:
./FahCore_22 -dir 00 -suffix 01 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device 0 -gpu 0 --log-time=true
Unfortunately, still not working.

I appreciate your effort. Maybe it's not that important to spend so much time with this topic.
But whenever you give me something to test, I will test
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

Which slot is GPU in? Main, or secondary?
What does the log tells you after unsuccessful attempt?
FAH Omega tester
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

The GPU is installed in the Main slot (checked)

I posted the log before. It's always the same error.
Attempting to create CUDA context --> failed...
then the same with Open_CL...
then: Failed to create ... OpenMM Context

Code: Select all

22:06:21:Attempting to create CUDA context:
22:06:21:  Configuring platform CUDA
22:06:21:Failed to create CUDA context:
22:06:21:The requested CUDA device could not be loaded
22:06:21:Attempting to create OpenCL context:
22:06:21:  Configuring platform OpenCL
22:06:21:Failed to create OpenCL context:
22:06:21:Error initializing context: clCreateContext (-9999)
22:06:21:ERROR:125: Failed to create a GPU-enabled OpenMM Context.
22:06:21:Saving result file ../logfile_01.txt
22:06:21:Saving result file science.log
22:06:21:Folding@home Core Shutdown: BAD_WORK_UNIT
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

Run (inside of your Desktop/mps_test dir):
export LD_LIBRARY_PATH="$PWD"
and
ldd FahCore_22
post the output of ldd FahCore_22 here, please
FAH Omega tester
Image
Top
foldinghomealone
Posts: 164
Joined: Wed Feb 01, 2017 7:07 pm
Hardware configuration: 5900x + 5080

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by foldinghomealone »

here you go:

Code: Select all

export LD_LIBRARY_PATH="$PWD"
ldd FahCore_22
	linux-vdso.so.1 (0x00007b12104a8000)
	libOpenCL.so.1 => /home/myxubuntu/Desktop/mps_test/libOpenCL.so.1 (0x00007b1210200000)
	libOpenMM.so.7.7 => /home/myxubuntu/Desktop/mps_test/libOpenMM.so.7.7 (0x00007b120fe00000)
	libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007b121048a000)
	libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007b1210485000)
	libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007b120fd17000)
	libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007b120fa00000)
	/lib64/ld-linux-x86-64.so.2 (0x00007b12104aa000)
	librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007b121047e000)
	libstdc++.so.6 => /home/myxubuntu/Desktop/mps_test/./libstdc++.so.6 (0x00007b120f600000)
	libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007b1210450000
Image
Top
muziqaz
Posts: 2131
Joined: Sun Dec 16, 2007 6:22 pm
Hardware configuration: 9950x, 9950x3D, 5950x, 5800x3D
7900xtx, RX9070, Radeon 7, 5700xt, 6900xt, Intel B580
Location: London
Contact:
Contact muziqaz

Re: Using MPS to dramatically increase PPD on big GPUs (Linux guide)

Post by muziqaz »

ok, your ldd seems fine.
so last couple of tries:
change opencl-platform, opencl-device, cuda-device and GPU to 1 instead 0 in the fahcore22 launch command. remember to run LD_LIBRARY_PATH command before that.
If 1 does not help, then run this command instead:
./FahCore_22 -dir 00 -suffix 01 -opencl-platform 0 -opencl-device 0 -gpu-vendor nvidia -gpu 0 --log-time=true
This should run your GPUs in opencl mode, but it still would be able to show advantage of MPS
FAH Omega tester
Image
Top
Post Reply

Return to “Problems with NVidia drivers”