[PATCH] D107046: [NVPTX] Add NVPTX intrinsics for CUDA PTX 6.5 ldmatrix instructions

[Artem Belevich via Phabricator via llvm-commits]

tra accepted this revision.
tra added a comment.
This revision is now accepted and ready to land.

Nice!

BTW, I think it may be a good time for you to ask for LLVM commit access. llvm.org/docs/DeveloperPolicy.html#obtaining-commit-access



================
Comment at: llvm/lib/Target/NVPTX/NVPTXIntrinsics.td:7919
+//
+// ldmatrix.sync.aligned.m8n8[|.trans][|.shared].b16
+//
----------------
Nit: `|` is redundant here, IMO.
I think `[something]` already reads as `something is optional.`


================
Comment at: llvm/lib/Target/NVPTX/NVPTXIntrinsics.td:7954-7955
+      } // space
+    } // transposed
+} // defset
 
----------------
Nit: something is off with the indentation here.


================
Comment at: llvm/test/CodeGen/NVPTX/wmma.py:636
+"""
+  intrinsic_template = "llvm.nvvm.ldmatrix.sync.aligned.${geom}.${x}${trans}.${itype}.${pspace}"
+  instruction_template = "ldmatrix.sync.aligned.${geom}.${x}${trans}${space}.${itype}"
----------------
`x` -> `frag` ?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107046/new/

https://reviews.llvm.org/D107046