One option worth trying for OpenCL would be to install the packaged version from the AMDGPU-PRO download. That version runs over PAL rather than ROCm but should work OK with an otherwise upstream-based driver stack. Not sure about GL/CL interop with Mesa though...
Announcement
Collapse
No announcement yet.
AMD Ryzen 4000 Mobile Series "Renoir" Graphics No Longer Experimental With Linux 5.5
Collapse
X
-
Originally posted by bridgman View PostOne option worth trying for OpenCL would be to install the packaged version from the AMDGPU-PRO download. That version runs over PAL rather than ROCm but should work OK with an otherwise upstream-based driver stack. Not sure about GL/CL interop with Mesa though...
Platform Abstraction Library?
Platform Abstraction Library. Contribute to GPUOpen-Drivers/pal development by creating an account on GitHub.
It's hard to keep track of AMD and all changes.
Is ROCm being deprecated?
Is OpenCL also deprecated and one should look at CUDA over HIP?
Phoronix: AMD's GPU Performance API Library 3.5 Drops ROCm/HSA Support Released on Friday was a new version of AMD's GPU Performance API "GPUPerfAPI" project under the GPUOpen umbrella. This is the AMD library used by CodeXL, Radeon Compute Profiler, and others for tapping GPU performance counters and to help in
Originally posted by Nille_kungen View Posthttps://www.olcf.ornl.gov/wp-content..._specsheet.pdf
OpenCL is a dead end and the future seems to be CUDA over HIP so developers should start using HIP instead.
When a Swedish techsite had a 20yr anniversary Joe Macri was a speaker and i later read in the forum that one had asked him after the speech and the recording "doesn't AMD still invests in OpenCL?".
The answer was no, better to use CUDA over HIP.
I hope nothing got lost in translation.
Here's the speech if anyone is interested.
https://www.youtube.com/watch?v=D8m6CSlNmQA
Comment
-
Originally posted by bridgman View PostOne option worth trying for OpenCL would be to install the packaged version from the AMDGPU-PRO download. That version runs over PAL rather than ROCm but should work OK with an otherwise upstream-based driver stack. Not sure about GL/CL interop with Mesa though...
I have now finally managed to get also the ROCm-opencl-runtime building but clinfo is failing almost on first steps when calling the clGetPlatformIDs from the /lib64/libOpenCL.so.1 that is provided by my distribution from the opensource ocl-icd package ocl-icd-2.2.12-2.mga7.src.rpm. I still lack the understanding whether that should be ok and I just need to reconfigure something or whether I need to rebuild that by myself from khronos git version available at https://github.com/KhronosGroup/OpenCL-ICD-Loader.
I also just bought parts for newer ryzen 3700x / navi 5700 machine and plan to install that today. Not sure whether any of the hip/hcc/opencl is supporting that out of the box but at least the llvm should build much faster :-)
Comment
-
Originally posted by Nille_kungen View PostAnd what's PAL in this case? Platform Abstraction Library?
https://github.com/GPUOpen-Drivers/pal
Originally posted by Nille_kungen View PostIs ROCm being deprecated?
Originally posted by Nille_kungen View PostIs OpenCL also deprecated and one should look at CUDA over HIP?Last edited by bridgman; 22 August 2020, 03:23 PM.Test signature
- Likes 2
Comment
-
Originally posted by bridgman View PostOpenCL is not deprecated (or being deprecated), however if you are looking at porting an application from CUDA then using HIP rather than OpenCL makes the porting effort much easier.
Comment
Comment