Strange warning: No openc-device specified; using deprecate

[prcowley]

I have this strange warning message come up saying (in then end) please consider upgrading your client version.
Then it seems to find CUDA and off it goes.

For some totally unfathomable reason, the log file fails to mention the client version is 7.6.21 which is (I believe) the latest

Here is a snippet from the log:

Code: Select all

06:51:33:WU00:FS01:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:17324 run:0 clone:1004 gen:4 core:0x22 unit:0x000003ec00000004000043ac00000000
06:51:33:WU00:FS01:Starting
06:51:33:WU00:FS01:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/cores.foldingathome.org/lin/64bit/22-0.0.13/Core_22.fah/FahCore_22 -dir 00 -suffix 01 -version 706 -lifeline 1876 -checkpoint 15 -cuda-device 0 -gpu-vendor nvidia -gpu -1 -gpu-usage 100
06:51:33:WU00:FS01:Started FahCore on PID 14858
06:51:33:WU00:FS01:Core PID:14862
06:51:33:WU00:FS01:FahCore 0x22 started
06:51:34:WU00:FS01:0x22:*********************** Log Started 2021-02-04T06:51:33Z ***********************
06:51:34:WU00:FS01:0x22:*************************** Core22 Folding@home Core ***************************
06:51:34:WU00:FS01:0x22:       Core: Core22
06:51:34:WU00:FS01:0x22:       Type: 0x22
06:51:34:WU00:FS01:0x22:    Version: 0.0.13
06:51:34:WU00:FS01:0x22:     Author: Joseph Coffland <[email protected]>
06:51:34:WU00:FS01:0x22:  Copyright: 2020 foldingathome.org
06:51:34:WU00:FS01:0x22:   Homepage: https://foldingathome.org/
06:51:34:WU00:FS01:0x22:       Date: Sep 19 2020
06:51:34:WU00:FS01:0x22:       Time: 01:10:35
06:51:34:WU00:FS01:0x22:   Revision: 571cf95de6de2c592c7c3ed48fcfb2e33e9ea7d3
06:51:34:WU00:FS01:0x22:     Branch: core22-0.0.13
06:51:34:WU00:FS01:0x22:   Compiler: GNU 4.8.2 20140120 (Red Hat 4.8.2-15)
06:51:34:WU00:FS01:0x22:    Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections -O3
06:51:34:WU00:FS01:0x22:             -funroll-loops -DOPENMM_GIT_HASH="\"189320d0\""
06:51:34:WU00:FS01:0x22:   Platform: linux2 4.19.76-linuxkit
06:51:34:WU00:FS01:0x22:       Bits: 64
06:51:34:WU00:FS01:0x22:       Mode: Release
06:51:34:WU00:FS01:0x22:Maintainers: John Chodera <[email protected]> and Peter Eastman
06:51:34:WU00:FS01:0x22:             <[email protected]>
06:51:34:WU00:FS01:0x22:       Args: -dir 00 -suffix 01 -version 706 -lifeline 14858 -checkpoint 15
06:51:34:WU00:FS01:0x22:             -cuda-device 0 -gpu-vendor nvidia -gpu -1 -gpu-usage 100
06:51:34:WU00:FS01:0x22:************************************ libFAH ************************************
06:51:34:WU00:FS01:0x22:       Date: Sep 15 2020
06:51:34:WU00:FS01:0x22:       Time: 05:14:43
06:51:34:WU00:FS01:0x22:   Revision: 44301ed97b996b63fe736bb8073f22209cb2b603
06:51:34:WU00:FS01:0x22:     Branch: HEAD
06:51:34:WU00:FS01:0x22:   Compiler: GNU 4.8.2 20140120 (Red Hat 4.8.2-15)
06:51:34:WU00:FS01:0x22:    Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections -O3
06:51:34:WU00:FS01:0x22:             -funroll-loops
06:51:34:WU00:FS01:0x22:   Platform: linux2 4.19.76-linuxkit
06:51:34:WU00:FS01:0x22:       Bits: 64
06:51:34:WU00:FS01:0x22:       Mode: Release
06:51:34:WU00:FS01:0x22:************************************ CBang *************************************
06:51:34:WU00:FS01:0x22:       Date: Sep 15 2020
06:51:34:WU00:FS01:0x22:       Time: 05:11:04
06:51:34:WU00:FS01:0x22:   Revision: 33fcfc2b3ed2195a423606a264718e31e6b3903f
06:51:34:WU00:FS01:0x22:     Branch: HEAD
06:51:34:WU00:FS01:0x22:   Compiler: GNU 4.8.2 20140120 (Red Hat 4.8.2-15)
06:51:34:WU00:FS01:0x22:    Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections -O3
06:51:34:WU00:FS01:0x22:             -funroll-loops -fPIC
06:51:34:WU00:FS01:0x22:   Platform: linux2 4.19.76-linuxkit
06:51:34:WU00:FS01:0x22:       Bits: 64
06:51:34:WU00:FS01:0x22:       Mode: Release
06:51:34:WU00:FS01:0x22:************************************ System ************************************
06:51:34:WU00:FS01:0x22:        CPU: AMD Ryzen 7 1700 Eight-Core Processor
06:51:34:WU00:FS01:0x22:     CPU ID: AuthenticAMD Family 23 Model 1 Stepping 1
06:51:34:WU00:FS01:0x22:       CPUs: 16
06:51:34:WU00:FS01:0x22:     Memory: 31.36GiB
06:51:34:WU00:FS01:0x22:Free Memory: 1.37GiB
06:51:34:WU00:FS01:0x22:    Threads: POSIX_THREADS
06:51:34:WU00:FS01:0x22: OS Version: 5.4
06:51:34:WU00:FS01:0x22:Has Battery: false
06:51:34:WU00:FS01:0x22: On Battery: false
06:51:34:WU00:FS01:0x22: UTC Offset: 13
06:51:34:WU00:FS01:0x22:        PID: 14862
06:51:34:WU00:FS01:0x22:        CWD: /var/lib/fahclient/work
06:51:34:WU00:FS01:0x22:************************************ OpenMM ************************************
06:51:34:WU00:FS01:0x22:   Revision: 189320d0
06:51:34:WU00:FS01:0x22:********************************************************************************
06:51:34:WU00:FS01:0x22:Project: 17324 (Run 0, Clone 1004, Gen 4)
06:51:34:WU00:FS01:0x22:Unit: 0x00000000000000000000000000000000
06:51:34:WU00:FS01:0x22:Reading tar file core.xml
06:51:34:WU00:FS01:0x22:Reading tar file integrator.xml.bz2
06:51:34:WU00:FS01:0x22:Reading tar file state.xml.bz2
06:51:34:WU00:FS01:0x22:Reading tar file system.xml.bz2
06:51:34:WU00:FS01:0x22:Digital signatures verified
06:51:34:WU00:FS01:0x22:Folding@home GPU Core22 Folding@home Core
06:51:34:WU00:FS01:0x22:Version 0.0.13
06:51:34:WU00:FS01:0x22:  Checkpoint write interval: 25000 steps (2%) [50 total]
06:51:34:WU00:FS01:0x22:  JSON viewer frame write interval: 12500 steps (1%) [100 total]
06:51:34:WU00:FS01:0x22:  XTC frame write interval: 250000 steps (20%) [5 total]
06:51:34:WU00:FS01:0x22:  Global context and integrator variables write interval: disabled
06:51:34:WU00:FS01:0x22:[b]No -opencl-device specified; using deprecated -gpu argument as an alias for -opencl-device.[/b]
06:51:34:WU00:FS01:0x22:[b]Please consider upgrading your client version.[/b]
06:51:34:WU00:FS01:0x22:There are 4 platforms available.
06:51:34:WU00:FS01:0x22:Platform 0: Reference
06:51:34:WU00:FS01:0x22:Platform 1: CPU
06:51:34:WU00:FS01:0x22:[color=#FF0000]Platform 2: OpenCL
06:51:34:WU00:FS01:0x22: [b] opencl-device -1 specified[/b]
06:51:34:WU00:FS01:0x22:Platform 3: CUDA
06:51:34:WU00:FS01:0x22:  cuda-device 0 specified[/color]
06:51:43:WU02:FS01:Upload complete
06:51:43:WU02:FS01:Server responded WORK_ACK (400)
06:51:43:WU02:FS01:Final credit estimate, 222934.00 points
06:51:43:WU02:FS01:Cleaning up
06:51:44:WU00:FS01:0x22:Attempting to create CUDA context:
06:51:44:WU00:FS01:0x22:  Configuring platform CUDA
06:51:54:WU00:FS01:0x22:  Using CUDA and gpu 0
06:51:54:WU00:FS01:0x22:Completed 0 out of 1250000 steps (0%)
06:51:55:WU00:FS01:0x22:Checkpoint completed at step 0
06:53:04:WU00:FS01:0x22:Completed 12500 out of 1250000 steps (1%)

So it is not an error as such but still is weird as I do have open cl installed according to clinfo:

Code: Select all

pete@pete-System-4:~$ clinfo
Number of platforms                               1
  Platform Name                                   NVIDIA CUDA
  Platform Vendor                                 NVIDIA Corporation
  Platform Version                                OpenCL 1.2 CUDA 11.2.109
  Platform Profile                                FULL_PROFILE
  Platform Extensions                             cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_nv_create_buffer cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_device_uuid
  Platform Extensions function suffix             NV

  Platform Name                                   NVIDIA CUDA
Number of devices                                 1
  Device Name                                     GeForce GTX 1080 Ti
  Device Vendor                                   NVIDIA Corporation
  Device Vendor ID                                0x10de
  Device Version                                  OpenCL 1.2 CUDA
  Driver Version                                  460.32.03
  Device OpenCL C Version                         OpenCL C 1.2 
  Device Type                                     GPU
  Device Topology (NV)                            PCI-E, 09:00.0
  Device Profile                                  FULL_PROFILE
  Device Available                                Yes
  Compiler Available                              Yes
  Linker Available                                Yes
  Max compute units                               28
  Max clock frequency                             1670MHz
  Compute Capability (NV)                         6.1
  Device Partition                                (core)
    Max number of sub-devices                     1
    Supported partition types                     None
    Supported affinity domains                    (n/a)
  Max work item dimensions                        3
  Max work item sizes                             1024x1024x64
  Max work group size                             1024
  Preferred work group size multiple              32
  Warp size (NV)                                  32
  Preferred / native vector sizes                 
    char                                                 1 / 1       
    short                                                1 / 1       
    int                                                  1 / 1       
    long                                                 1 / 1       
    half                                                 0 / 0        (n/a)
    float                                                1 / 1       
    double                                               1 / 1        (cl_khr_fp64)
  Half-precision Floating-point support           (n/a)
  Single-precision Floating-point support         (core)
    Denormals                                     Yes
    Infinity and NANs                             Yes
    Round to nearest                              Yes
    Round to zero                                 Yes
    Round to infinity                             Yes
    IEEE754-2008 fused multiply-add               Yes
    Support is emulated in software               No
    Correctly-rounded divide and sqrt operations  Yes
  Double-precision Floating-point support         (cl_khr_fp64)
    Denormals                                     Yes
    Infinity and NANs                             Yes
    Round to nearest                              Yes
    Round to zero                                 Yes
    Round to infinity                             Yes
    IEEE754-2008 fused multiply-add               Yes
    Support is emulated in software               No
  Address bits                                    64, Little-Endian
  Global memory size                              11713052672 (10.91GiB)
  Error Correction support                        No
  Max memory allocation                           2928263168 (2.727GiB)
  Unified memory for Host and Device              No
  Integrated memory (NV)                          No
  Minimum alignment for any data type             128 bytes
  Alignment of base address                       4096 bits (512 bytes)
  Global Memory cache type                        Read/Write
  Global Memory cache size                        1376256 (1.312MiB)
  Global Memory cache line size                   128 bytes
  Image support                                   Yes
    Max number of samplers per kernel             32
    Max size for 1D images from buffer            268435456 pixels
    Max 1D or 2D image array size                 2048 images
    Max 2D image size                             16384x32768 pixels
    Max 3D image size                             16384x16384x16384 pixels
    Max number of read image args                 256
    Max number of write image args                16
  Local memory type                               Local
  Local memory size                               49152 (48KiB)
  Registers per block (NV)                        65536
  Max number of constant args                     9
  Max constant buffer size                        65536 (64KiB)
  Max size of kernel argument                     4352 (4.25KiB)
  Queue properties                                
    Out-of-order execution                        Yes
    Profiling                                     Yes
  Prefer user sync for interop                    No
  Profiling timer resolution                      1000ns
  Execution capabilities                          
    Run OpenCL kernels                            Yes
    Run native kernels                            No
    Kernel execution timeout (NV)                 Yes
  Concurrent copy and kernel execution (NV)       Yes
    Number of async copy engines                  2
  printf() buffer size                            1048576 (1024KiB)
  Built-in kernels                                (n/a)
  Device Extensions                               cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_nv_create_buffer cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_device_uuid

NULL platform behavior
  clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...)  NVIDIA CUDA
  clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...)   Success [NV]
  clCreateContext(NULL, ...) [default]            Success [NV]
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT)  No platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU)  No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU)  No platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR)  No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM)  Invalid device type for platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL)  No platform

ICD loader properties
  ICD loader Name                                 OpenCL ICD Loader
  ICD loader Vendor                               OCL Icd free software
  ICD loader Version                              2.2.11
  ICD loader Profile                              OpenCL 2.1
pete@pete-System-4:~$ 

Anyone have any idea of what the issue is and how to resolve it - if it in fact needs to be resolved.
It is by no means a show stopper!

Thanks in advance for any help.
Cheers
Pete

Mod Edit: Added Code Tags - PantherX

[Whompithian]

The message is a generic indication that something went wrong with Folding@home's attempt to interrogate the state of OpenCL on your system. If it proceeds to fold, using CUDA or any other means, then there is no issue and you can ignore the message.

The client version is only logged when the client first starts up. If you have been folding for a few days, then it is likely that part of the log was rotated out. You may be able to find the version string in one of the backups under the "logs" subdirectory of the FAHClient working directory. It appears in a section like the following:

Code: Select all

*********************** Log Started 2021-02-05T07:59:55Z ***********************
2021-02-05:07:59:55:************************** libFAH **************************
2021-02-05:07:59:55:           Date: Oct 20 2020
2021-02-05:07:59:55:           Time: 20:36:41
2021-02-05:07:59:55:       Revision: 5ca109d295a6245e2a2f590b3d0085ad5e567aeb
2021-02-05:07:59:55:         Branch: master
2021-02-05:07:59:55:       Compiler: GNU 4.9.4
2021-02-05:07:59:55:        Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections
2021-02-05:07:59:55:                 -O3 -funroll-loops
2021-02-05:07:59:55:       Platform: linux2 5.8.0-1-amd64
2021-02-05:07:59:55:           Bits: 64
2021-02-05:07:59:55:           Mode: Release
2021-02-05:07:59:55:************************ FAHClient *************************
2021-02-05:07:59:55:        Version: 7.6.21
2021-02-05:07:59:55:         Author: Joseph Coffland <[email protected]>
2021-02-05:07:59:55:      Copyright: 2020 foldingathome.org
2021-02-05:07:59:55:       Homepage: https://foldingathome.org/
2021-02-05:07:59:55:           Date: Oct 20 2020
2021-02-05:07:59:55:           Time: 20:38:59
2021-02-05:07:59:55:       Revision: 6efbf0e138e22d3963e6a291f78dcb9c6422a278
2021-02-05:07:59:55:         Branch: master
2021-02-05:07:59:55:       Compiler: GNU 4.9.4
2021-02-05:07:59:55:        Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections
2021-02-05:07:59:55:                 -O3 -funroll-loops
2021-02-05:07:59:55:       Platform: linux2 5.8.0-1-amd64
2021-02-05:07:59:55:           Bits: 64
2021-02-05:07:59:55:           Mode: Release
2021-02-05:07:59:55:           Args: --config=/etc/fahclient/config.xml --chdir=/var/lib/fahclient/
2021-02-05:07:59:55:         Config: /etc/fahclient/config.xml

Also, please wrap log content, commands, and command output in "[ code ] [ /code ]" tags (no spaces) to make it easier to read, especially when there is a lot of text. You can edit your original post to apply the tags retroactively.

I hope you find some of this information helpful.

[JimboPalmer]

CUDA is a Nvidia specific platform they update regularly and is optimized for each new generation of Nvidia GPUs. OpenCL is a vendor independent platform that Nvidia made a driver for in 2012 and so uses the features of the graphics cards of that era.

When F@H code uses CUDA on Nvidia, it gets a 40% to 70% increase in speed over F@H speeds in OpenCL, so if both are available it will choose CUDA by default. (F@H uses CUDA commands available in Kelper and later GPUs, so some late model Fermi GPUs can use OpenCL but not CUDA)

Since AMD only uses OpenCL, they keep their OpenCL driver updated much better than Nvidia. There is no incentive for AMD to lock you in to a proprietary language.

If you really do not want to use CUDA, you can configure F@H to use OpenCL when both are available, but your Points Per Day will drop substantially.

[bruce]

Your GeForce GTX 1080 Ti is capable of running either OpenCL or CUDA provided you've installed the drivers directly from NVidia. (The client assumed you installed those drivers.) If you choose to install those drivers, it will successfully probe both. As long as CUDA works, there's really no particular need for OpenCL, but if, for some unpredictable reason, CUDA fails to set itself up properly, it's nice to have OpenCL as a backup -- although it's going to be slower as mentioned above.

[prcowley]

The message is a generic indication that something went wrong with Folding@home's attempt to interrogate the state of OpenCL on your system. If it proceeds to fold, using CUDA or any other means, then there is no issue and you can ignore the message.

The client version is only logged when the client first starts up. If you have been folding for a few days, then it is likely that part of the log was rotated out. You may be able to find the version string in one of the backups under the "logs" subdirectory of the FAHClient working directory. It appears in a section like the following:

Code: Select all

*********************** Log Started 2021-02-05T07:59:55Z ***********************
2021-02-05:07:59:55:************************** libFAH **************************
2021-02-05:07:59:55:           Date: Oct 20 2020
2021-02-05:07:59:55:           Time: 20:36:41
2021-02-05:07:59:55:       Revision: 5ca109d295a6245e2a2f590b3d0085ad5e567aeb
2021-02-05:07:59:55:         Branch: master
2021-02-05:07:59:55:       Compiler: GNU 4.9.4
2021-02-05:07:59:55:        Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections
2021-02-05:07:59:55:                 -O3 -funroll-loops
2021-02-05:07:59:55:       Platform: linux2 5.8.0-1-amd64
2021-02-05:07:59:55:           Bits: 64
2021-02-05:07:59:55:           Mode: Release
2021-02-05:07:59:55:************************ FAHClient *************************
2021-02-05:07:59:55:        Version: 7.6.21
2021-02-05:07:59:55:         Author: Joseph Coffland <[email protected]>
2021-02-05:07:59:55:      Copyright: 2020 foldingathome.org
2021-02-05:07:59:55:       Homepage: https://foldingathome.org/
2021-02-05:07:59:55:           Date: Oct 20 2020
2021-02-05:07:59:55:           Time: 20:38:59
2021-02-05:07:59:55:       Revision: 6efbf0e138e22d3963e6a291f78dcb9c6422a278
2021-02-05:07:59:55:         Branch: master
2021-02-05:07:59:55:       Compiler: GNU 4.9.4
2021-02-05:07:59:55:        Options: -std=c++11 -fsigned-char -ffunction-sections -fdata-sections
2021-02-05:07:59:55:                 -O3 -funroll-loops
2021-02-05:07:59:55:       Platform: linux2 5.8.0-1-amd64
2021-02-05:07:59:55:           Bits: 64
2021-02-05:07:59:55:           Mode: Release
2021-02-05:07:59:55:           Args: --config=/etc/fahclient/config.xml --chdir=/var/lib/fahclient/
2021-02-05:07:59:55:         Config: /etc/fahclient/config.xml

Also, please wrap log content, commands, and command output in "[ code ] [ /code ]" tags (no spaces) to make it easier to read, especially when there is a lot of text. You can edit your original post to apply the tags retroactively.

I hope you find some of this information helpful.

Hi Whompithian

Thank you for that info. It does make sense but was not obvious. I would imagine it would be a lot less alarming, to me and others, if the Client version was written every time it it started a new log file, but it is a minor point and I am sure the developers have many other more important things to do. I leave my machine on 24 x 7 so it is only when I do a reboot that the client get a restart.

Sorry about forgetting to put the Code tags in. Will try to remember that next time.

Cheers
Pete

[prcowley]

CUDA is a Nvidia specific platform they update regularly and is optimized for each new generation of Nvidia GPUs. OpenCL is a vendor independent platform that Nvidia made a driver for in 2012 and so uses the features of the graphics cards of that era.

When F@H code uses CUDA on Nvidia, it gets a 40% to 70% increase in speed over F@H speeds in OpenCL, so if both are available it will choose CUDA by default. (F@H uses CUDA commands available in Kelper and later GPUs, so some late model Fermi GPUs can use OpenCL but not CUDA)

Since AMD only uses OpenCL, they keep their OpenCL driver updated much better than Nvidia. There is no incentive for AMD to lock you in to a proprietary language.

If you really do not want to use CUDA, you can configure F@H to use OpenCL when both are available, but your Points Per Day will drop substantially.

Thanks for this

It seemed to me that the order that options were tried in, i.e. Open Cl before CUDA, that this this was the preferred method you use if available. I certainly don't want to use something that generates less points but it just seems pretty odd that 1- it can't find Open Cl anyhow and 2 - it is slower.

I could not find anything out about it so your post was most welcome and very informative.

Cheers
Pete Cowley

[bruce]

The client interrogates the hardware when FAHClient is (re-)started although that can be disable with the "gpu" = "false" setting.

Every time the GPU core starts a new WU, it will use that information to decide which platform to use, but only ones which were detected earlier. Mine looks like this:

Code: Select all

18:26:43:WU01:FS01:0x22:There are 4 platforms available.
18:26:43:WU01:FS01:0x22:Platform 0: Reference
18:26:43:WU01:FS01:0x22:Platform 1: CPU
18:26:43:WU01:FS01:0x22:Platform 2: OpenCL
18:26:43:WU01:FS01:0x22:  opencl-device 0 specified
18:26:43:WU01:FS01:0x22:Platform 3: CUDA
18:26:43:WU01:FS01:0x22:  cuda-device 0 specified

It then proceeds to set up a cuda-device if it can and then reverts to an opencl-device if CUDA doesn't set up properly. I gets more complicated if there are multiple GPUs, but the same concepts apply.

Depending on both the GPU and the WU, a CUDA device is somewhat faster than an OpenCL device using the same GPU.

[prcowley]

Hi Bruce

Thanks for that!

Cheers
Pete

[JimboPalmer]

It seemed to me that the order that options were tried in, i.e. Open Cl before CUDA, that this this was the preferred method you use if available.

OpenCL is the baseline, if it won't do OpenCl 1.2, it won't fold at all. Then it checks for CUDA, and if that is compatible, it chooses it by default.

[PantherX]

BTW, can you please post the first 100 lines from your log file? In theory, if you're using the latest client version (V7.6.21), then that message shouldn't be printed unless you have something specific in your configuration which can be now be removed. Just wondering, was this a fresh install or an update from previous version(s)?

[bruce]

It seemed to me that the order that options were tried in, i.e. Open Cl before CUDA, that this this was the preferred method you use if available.

OpenCL is the baseline, if it won't do OpenCl 1.2, it won't fold at all. Then it checks for CUDA, and if that is compatible, it chooses it by default.

I'm not sure if the word "baseline" really applies. Technically you are talking about two different components of the FAH software.

When FAHClient was originally written, OpenCL was all that was available. It became the standard for detecting GPUs. I don't think its baseline logic has changed. FAHClient detects GPUs which have support for OpenCL 1.2 In fact, any device that supports CUDA also supports OpenCL 1.2 so it seems reasonable to have software that verifies only OpenCL devices.

Much later, a version of FAHCore_22 was developed that could use CUDA. Those two packages were developed independently and very likely by different individuals. FAHCore_22 prefers CUDA (since it's somewhat faster) but uses OpenCL if CUDA doesn't happen to be supported (e.g.- GPUs from AMD or Intel).

In fact, the special case of a CUDA device for which an OpenCL driver has not been installed isn't covered.