This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Allow int types for MUBUF vdata
ClosedPublic

Authored by tpr on Aug 5 2018, 11:40 AM.

Details

Summary

Previously the new llvm.amdgcn.raw/struct.buffer.load/store intrinsics
only allowed float types for the data to be loaded or stored, which
sometimes meant the frontend needed to generate a bitcast. In this, the
new intrinsics copied the old buffer intrinsics.

This commit extends the new intrinsics to allow int types as well.

Change-Id: I8202af2d036455553681dcbb3d7d32ae273f8f85

Diff Detail

Repository
rL LLVM

Event Timeline

tpr created this revision.Aug 5 2018, 11:40 AM
arsenm accepted this revision.Aug 5 2018, 11:08 PM

LGTM

test/CodeGen/AMDGPU/llvm.amdgcn.raw.buffer.load.ll
1–2 ↗(On Diff #159229)

Should add -enable-var-scope

This revision is now accepted and ready to land.Aug 5 2018, 11:08 PM
tpr added a comment.Aug 17 2018, 2:57 AM

This change depends on D50306, so I can't land it until that one is approved and landed.

This revision was automatically updated to reflect the committed changes.