Use builtins instead of inline asm for readlane

ROCm / rocALUTION

Next generation library for iterative sparse solvers for ROCm platform

https://rocm.docs.amd.com/projects/rocALUTION/en/latest/

MIT License

74 stars 38 forks source link

Use builtins instead of inline asm for readlane #220

Closed vikramRH closed 2 months ago

vikramRH commented 2 months ago

I currently have a compiler patch up for review (https://github.com/llvm/llvm-project/pull/89217) which will break with inline asm. This patch is a prerequisite

vikramRH commented 2 months ago

ping..

jsandham commented 2 months ago

@vikramRH Can this PR be made in rocALUTION-internal instead?

vikramRH commented 2 months ago

@jsandham , I dont seem to have permissions to rocALUTION-internal and no merge permissions here. could you help me land these changes ? (Im okay with either creating new PR directly to internal repo or cherry picking this, whichever seems proper to you)