OpenCL 1.2 finally in Mesa - please test

[PantherX]

When it comes to GPU detection in Windows VS Linux, I have a vague recollection that Linux might be disabled by default since there's more work involved in it than Windows... this was the assumption many years ago so I am not sure if that's the reason why the default behavior between Windows and Linux stems from. Of course, I could have an over-active imagination and this was purely fictional so take it with a decent amount of salt

[bruce]

Quick test: I installed FAH 7.6.21 on a Win 10 laptop with 2 GPUs (Intel + Nvidia). Expert, start manually...

Do we know whether that system has OpenCL 1.1 or >=1.2 on it?

[ajm]

OpenCL 1.2 (according to Geeks3D GPU Caps Viewer)

[Foliant]

The only other thing I can think of is the driver. [...] so you may need to drop back a version or two with the Windows driver.

After several restarts it suddenly worked. No dirver changes, but maybe some silent windows updates.
I noticed it working by poor performance in Solid Edge CAD and let it finish one WU.

The 7970 was showing around 350.000 PPD while being a really hot piece of circuit board.

[Rev]

Putting my hand up, to show that I am also a Linux-user with mesa drivers and installed opencl-icd, and still are unable to fold with the GPU part of my AMD 2400G.

I waited one year after my initial post about it, and it seems nothing happened to make use of all the integrated AMD graphics users on a linux (default - mesa) system. Seems no one on their team is giving a male chicken's donkey about it.

Prerequisites should be fine, just folding isn't:

mesa-opencl-icd 21.2~git2105081800.d81a6e~oibaf~f
ocl-icd-libopencl1 2.2.11-1ubuntu1
ocl-icd-libopencl1:i386 2.2.11-1ubuntu1
opencl-c-headers 2.2~2019.08.06-g0d5f18c-1

clinfo:

$ clinfo
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 21.2.0-devel (git-d81a6e5 2021-05-08 focal-oibaf-ppa)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA

Platform Name Clover
Number of devices 1
Device Name AMD Radeon(TM) Vega 11 Graphics (RAVEN, DRM 3.40.0, 5.12.2-051202-generic, LLVM 12.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 21.2.0-devel (git-d81a6e5 2021-05-08 focal-oibaf-ppa)
Driver Version 21.2.0-devel
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Max compute units 11
Max clock frequency 1250MHz
Max work item dimensions 3
Max work item sizes 256x256x256
Max work group size 256
Preferred / native vector sizes
char 16 / 16
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 0 (n/a)
float 4 / 4
double 2 / 2 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
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 3221225472 (3GiB)
Error Correction support No
Max memory allocation 2254857830 (2.1GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 32768 bits (4096 bytes)
Global Memory cache type None
Image support No
Local memory type Local
Local memory size 32768 (32KiB)
Max number of constant args 16
Max constant buffer size 67108864 (64MiB)
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Profiling timer resolution 0ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Device Extensions cl_khr_byte_addressable_store 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_fp64 cl_khr_extended_versioning

NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) Clover
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Success [MESA]
clCreateContext(NULL, ...) [default] Success [MESA]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name AMD Radeon(TM) Vega 11 Graphics (RAVEN, DRM 3.40.0, 5.12.2-051202-generic, LLVM 12.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name AMD Radeon(TM) Vega 11 Graphics (RAVEN, DRM 3.40.0, 5.12.2-051202-generic, LLVM 12.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) Success (1)
Platform Name Clover
Device Name AMD Radeon(TM) Vega 11 Graphics (RAVEN, DRM 3.40.0, 5.12.2-051202-generic, LLVM 12.0.0)

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

[toTOW]

clinfo:

$ clinfo
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 21.2.0-devel (git-d81a6e5 2021-05-08 focal-oibaf-ppa)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA

Platform Name Clover
Number of devices 1
Device Name AMD Radeon(TM) Vega 11 Graphics (RAVEN, DRM 3.40.0, 5.12.2-051202-generic, LLVM 12.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 21.2.0-devel (git-d81a6e5 2021-05-08 focal-oibaf-ppa)
Driver Version 21.2.0-devel
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Profile FULL_PROFILE

Same player, try again ...

FAH only supports proprietary drivers downloaded from the manufacturer (AMD or nVidia) website.

[bruce]

and FAH requires OpenCL 1.2. OpenCL 1.1 is insufficient.

[Foliant]

and FAH requires OpenCL 1.2.

I can only write about my GPU that supports1.2 and is running within windows.

The Mesamatrix https://mesamatrix.net/#OpenCL1.2 is listing most extensions build in.
But I also get "1.1" by using clinfo.

Could someone from the FAH Team proof if the build in extensions are enough to use FAH?
If someone declares all components should meet the requirements Im willing to contact the Mesa developers.

[JimboPalmer]

[I am just a folder like you, this is not an official policy, just my opinion]

Let’s decide to learn form the Past.

There is an Nvidia architecture called Fermi, none of the Fermi GPUs do all of OpenCL 1.2

However the phase 2* Fermi chips do part of OpenCL 1.2, let say 75% of proteins complete and 25% fail.

The problems with this are many.

“I have a Fermi GPU and he has a Fermi GPU and HE gets to fold, and I don’t”

“F@H is ‘broken’ because my GPU fails so often”

“I lose my bonus points because I did not complete 80% of my Work Units”

It is hard to black list a GPU which is still mostly working.

Projects would complete faster if they rejected Fermi GPUs at startup, rather than waiting for a slow chip to fail later.

Because Nvidia uses multiple GPUs in a card they name the same, some GT 730s are first phase Fermi, which never works, some are Kepler, which always works. The folder feels we are picking on him.

F@H still suffers from Fermi, I refuse to help make phase 2 Fermi work, I don’t want F@H to work badly. And I can’t make it right.

https://en.wikipedia.org/wiki/Fermi_(microarchitecture)
* chips named GF11x where x is some number

Until Mesa fully supports OpenCL 1.2, we have a similar problem, fortunately, we have an ‘easy’ solution, use the proprietary driver which DOES work.

[bruce]

I do understand your frustration, but unfortunately, drivers are the responsibility company that supports your hardware.

Mesa support must be obtained from AMD.com.

FAH does not support any GPU drivers. All we can do is t tell you to get the proprietary drivers and refer you to AMD .com or NVIdia.com or Intel.com or Mac.com

We do have other volunteers who visit this site who may have worked through such problems who may (or may not) be able to help.