[clang] [llvm] [AArch64] Add intrinsics for bflaot16 min/max/minnm/maxnm (PR #90105)

[via llvm-commits]

================
@@ -109,6 +109,19 @@ define { <vscale x 2 x i64>, <vscale x 2 x i64> } @multi_vec_max_single_x2_u64(<
   ret { <vscale x 2 x i64>, <vscale x 2 x i64> } %res
 }
 
+; BFMAX (Single, x2)
+
+define { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @multi_vec_max_single_x2_bf16(<vscale x 8 x bfloat> %zdn1, <vscale x 8 x bfloat> %zdn2, <vscale x 8 x bfloat> %zm) {
+; CHECK-LABEL: multi_vec_max_single_x2_bf16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    bfmax { z0.h, z1.h }, { z0.h, z1.h }, z2.h
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.bfmax.single.x2.nxv8bf16(<vscale x 8 x bfloat> %zdn1, <vscale x 8 x bfloat> %zdn2, <vscale x 8 x bfloat> %zm)
----------------
CarolineConcatto wrote:

I am not sure we should keep this. I understand that in InstriniscAArch64.td you are declaring together with previous implementation, but maybe we should split it and have one just for BF types

https://github.com/llvm/llvm-project/pull/90105