[PATCH] D120279: [AMDGPU] Extend SILoadStoreOptimizer to handle global loads
Stanislav Mekhanoshin via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Feb 22 08:29:14 PST 2022
rampitec added a comment.
In D120279#3337833 <https://reviews.llvm.org/D120279#3337833>, @arsenm wrote:
> Overall I think this pass should do less than it currently does. All of these cases should have been handled in the IR already, it's not the disjoint case of the DS operations.
>
> This is another example where I think having constant address space is overcomplicated and would be better served by global loads marked invariant
Invariant load can be reordered with stores, these constant loads cannot.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D120279/new/
https://reviews.llvm.org/D120279
More information about the llvm-commits
mailing list