================ @@ -2805,7 +2805,16 @@ def int_aarch64_sve_sudot_lane : AdvSIMD_SVE_DOT_Indexed_Intrinsic; // // SVE ACLE: 7.4/5. FP64/FP32 matrix multiply extensions // -def int_aarch64_sve_fmmla : AdvSIMD_3VectorArg_Intrinsic; + +def int_aarch64_sve_fmmla + : DefaultAttrsIntrinsic<[llvm_anyvector_ty], + [LLVMMatchType<0>, llvm_anyvector_ty, LLVMMatchType<1>], + [IntrNoMem]>; + +def int_aarch64_sve_fp8_fmmla + : DefaultAttrsIntrinsic<[llvm_anyvector_ty], + [LLVMMatchType<0>, llvm_anyvector_ty, LLVMMatchType<1>], ---------------- Lukacma wrote:
I don't think fp8 intrinsic needs to be overloaded on the source type. It is always going to be fp8 https://github.com/llvm/llvm-project/pull/165282 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
