llvm / llvm/llvm-project

[HLSL][LongVec] Add support for the ddx_fine intrinsic

Closed
#220,629 0 comments 0 reactions 0 assignees View on GitHub
HLSL
Dominant language
LLVM
Stars
40.5k
Forks
18.7k
PR merge metrics
PR metrics pending

Description

The offload test is failing for clang.

> XFAIL: OffloadTest-clang-warp-d3d12 :: Feature/HLSLLib/ddx-fine.long-vector.32.test (148 of 710)

Even though we have enabled long vectors in the frontend and checked the scalarization is happening in the backend.

- https://github.com/llvm/llvm-project/pull/216905
- https://github.com/llvm/llvm-project/pull/216893

Contributor guide

Open the contributing guide

Research direction

Start with the failing OffloadTest-clang-warp-d3d12 case, Feature/HLSLLib/ddx-fine.long-vector.32.test, and read the two linked pull requests. Trace how ddx_fine is handled with long vectors across the frontend and backend, including the stated scalarization check. Done means the test no longer needs its XFAIL and passes for the clang D3D12 offload configuration.

Written by the indexing model from the issue text.

Assessment

Domain
compilers, testing-qa
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.