[llvm-branch-commits] [llvm] [AMDGPU] Handle direct loads to LDS in memory model (PR #142018)

https://github.com/kerbowa created https://github.com/llvm/llvm-project/pull/142018 Add additional waitcnt insertion to ensure proper ordering between LDS operations and direct loads from global memory to LDS on pre-GFX10 hardware. Direct LDS loads perform both a global memory load and an LDS s

[llvm-branch-commits] [llvm] [AMDGPU] Handle direct loads to LDS in memory model (PR #142018)

kerbowa wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/142018?utm_source=stack-comment-downstack-mergeability-warning";

[llvm-branch-commits] [llvm] [AMDGPU] Update target helpers & GCNSchedStrategy for dynamic VGPRs (PR #130047)

https://github.com/kerbowa approved this pull request. LGTM We had some offline discussion about my concerns https://github.com/llvm/llvm-project/pull/130047 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org

[llvm-branch-commits] [llvm] [AMDGPU] Update target helpers & GCNSchedStrategy for dynamic VGPRs (PR #130047)

https://github.com/kerbowa commented: Is there any test for the revert scheduling portion of the change? https://github.com/llvm/llvm-project/pull/130047 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi

[llvm-branch-commits] [llvm] [AMDGPU] Update target helpers & GCNSchedStrategy for dynamic VGPRs (PR #130047)

@@ -1452,6 +1452,16 @@ bool GCNSchedStage::shouldRevertScheduling(unsigned WavesAfter) { if (WavesAfter < DAG.MinOccupancy) return true; + // For dynamic VGPR mode, we don't want to waste any VGPR blocks. + if (ST.isDynamicVGPREnabled()) { kerbowa wro

[llvm-branch-commits] [llvm] [AMDGPU] Update target helpers & GCNSchedStrategy for dynamic VGPRs (PR #130047)

https://github.com/kerbowa edited https://github.com/llvm/llvm-project/pull/130047 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits

[llvm-branch-commits] [llvm] [AMDGPU] Push amdgpu-preload-kern-arg-prolog after livedebugvalues (PR #126148)

https://github.com/kerbowa approved this pull request. This LGTM but I think previous PR in stack should be obsolete after #123547 https://github.com/llvm/llvm-project/pull/126148 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org

[llvm-branch-commits] [llvm] [AMDGPU] Remove dead function metadata after amdgpu-lower-kernel-arguments (PR #126147)

https://github.com/kerbowa commented: This shouldn't be needed after #123547, there are other bugs besides this caused by the leftover declarations. https://github.com/llvm/llvm-project/pull/126147 ___ llvm-branch-commits mailing list llvm-branch-com

[llvm-branch-commits] [llvm] 4aa842a - [AMDGPU] Add new pseudos for indirect addressing with VGPR Indexing

Author: Austin Kerbow Date: 2020-12-08T12:24:12-08:00 New Revision: 4aa842a800b53806a9e25f03b4e21f6879801a38 URL: https://github.com/llvm/llvm-project/commit/4aa842a800b53806a9e25f03b4e21f6879801a38 DIFF: https://github.com/llvm/llvm-project/commit/4aa842a800b53806a9e25f03b4e21f6879801a38.diff