[PATCH] D44210: [AMDGPU] Supported ds_read_b128 generation; Widened vector length for local address-space

[Nicolai Hähnle via Phabricator via llvm-commits]

nhaehnle added a comment.

Have you tested this on real hardware? I remember reading that there is a hardware bug on gfx7 with this instruction. The bug may apply only to early gfx7 chips.



================
Comment at: test/CodeGen/AMDGPU/reorder-stores.ll:2
+; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=amdgcn < %s | FileCheck -check-prefixes=SI,GCN %s
+; RUN:  llc -amdgpu-scalarize-global-loads=false  -march=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck -check-prefixes=SI,GCN-SEA %s
 
----------------
What does SEA mean? We usually use CI for Sea Islands.


https://reviews.llvm.org/D44210