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

[Benjamin Maxwell]

================
@@ -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
----------------
MacDue wrote:

This is i32 to i4. So it goes trunci i32 to i8 then i8 to i4 (via deinterleave).

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