[Mlir-commits] [mlir] [mlir][vector] Support n-D vectors in i8 to i4 trunci emulation (PR #94946)

Andrzej WarzyƄski llvmlistbot at llvm.org
Wed Jun 19 11:44:23 PDT 2024


================
@@ -300,16 +298,32 @@ func.func @aligned_trunci_2d(%a: vector<8x32xi32>) -> vector<8x32xi4> {
 // CHECK-NOT:       vector.shli
 // CHECK-NOT:       vector.ori
 // CHECK:           arith.trunci
----------------
banach-space wrote:

OK, thanks! Looks like this this is missing types. You can also add `CHECK-NOT: trunci` after the first `CHECK: trunci`

https://github.com/llvm/llvm-project/pull/94946


More information about the Mlir-commits mailing list