ProjectPhysX / ProjectPhysX/OpenCL-Benchmark

Incorrect checking for the float type support

Open
#5 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
331
Forks
37
PR merge metrics
No merged PRs in 30d

Description

This is duplication of https://github.com/ProjectPhysX/OpenCL-Benchmark/issues/4
I'm opening a new one because it doesn't allow reopening the issue.

And, no, I don't agree to your answer.
cl_khr_fp16/64 are the correct flags to reflect data type support in vendor's OpenCL implementation by the specification.
CL_DEVICE_NATIVE_VECTOR_WIDTH_* are just hints given from compiler/HW capability.
Intel's implementation seems to have a bug in it and you're currently propagating the bug into your code.

``For Intel Arc GPUs I had to add an additional failsafe:`
If it's a WA for the Intel's bug, you'd better check either one of them not both of them.

Regards,
Jungwook

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reading the duplicated issue at ProjectPhysX/OpenCL-Benchmark#4 and compare the reported float-support checks with the OpenCL specification. Done means resolving whether the extension flags or native vector width checks should determine support, including the Intel Arc workaround described here.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
hpc, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.