projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b6572ad) | patch
[AMDGPU] Allow for MFMA Inst Clustering
authorjeff <jrbyrnes1989@gmail.com>
Tue, 26 Apr 2022 18:23:13 +0000 (11:23 -0700)
committerjeff <jrbyrnes1989@gmail.com>
Tue, 10 May 2022 19:57:40 +0000 (12:57 -0700)
commitf822db7670d4399bcc90830f23fdb5cec6878c73
tree357457ed00264ce3734f09cd29c256df90ccafebtree | snapshot
parentb6572ad504753c6c27ba49c2ed595a091c977dcdcommit | diff
[AMDGPU] Allow for MFMA Inst Clustering

This patch adds cluster edges between independent MFMA instructions. Additionally, it propogates all predecessors of cluster insts to the root of the cluster(s), and all successors to the leaf(ves) of the cluster(s) -- this is done to remove the possibility that those insts will be interspersed within the cluster.

Reviewed By: kerbowa

Differential Revision: https://reviews.llvm.org/D124678
llvm/lib/Target/AMDGPU/AMDGPUMFMAClustering.cpp [new file with mode: 0644] blob
llvm/lib/Target/AMDGPU/AMDGPUMFMAClustering.h [new file with mode: 0644] blob
llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp diff | blob | history
llvm/lib/Target/AMDGPU/CMakeLists.txt diff | blob | history
llvm/test/CodeGen/AMDGPU/mfma-cluster-edges.mir [new file with mode: 0644] blob
llvm/test/CodeGen/AMDGPU/mfma-cluster.mir [new file with mode: 0644] blob
Domain: System / Toolchain;