diff --git a/onnxruntime/core/mlas/lib/x86_64/QgemvU8S8KernelAvxVnni.S b/onnxruntime/core/mlas/lib/x86_64/QgemvU8S8KernelAvxVnni.S index fb544ce34d..2d8f7d656e 100644 --- a/onnxruntime/core/mlas/lib/x86_64/QgemvU8S8KernelAvxVnni.S +++ b/onnxruntime/core/mlas/lib/x86_64/QgemvU8S8KernelAvxVnni.S @@ -130,7 +130,7 @@ C_UNDERSCORE(MlasGemvU8S8KernelAvxVnni): vperm2i128 ymm7,ymm7,ymm8,0x20 vperm2i128 ymm8,ymm9,ymm10,0x20 vperm2i128 ymm10,ymm9,ymm10,0x31 - vmovaps ymm3,ymm4 + vmovaps ymm9,ymm4 .LStoreOutput4By32: vmovdqu YMMWORD PTR [rbx],ymm7