[llvm] [X86] Fast AVX-512-VNNI vpdpwssd tuning (PR #85033)

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Wed Mar 13 03:21:59 PDT 2024


================
@@ -683,6 +683,12 @@ def TuningFastGather
     : SubtargetFeature<"fast-gather", "HasFastGather", "true",
                        "Indicates if gather is reasonably fast (this is true for Skylake client and all AVX-512 CPUs)">;
 
+// Generate vpdpwssd instead of vpmaddwd+vpaddd sequence.
+def TuningFastPWSSD
+    : SubtargetFeature<
+          "fast-pwssd", "HasFastPWSSD", "true",
----------------
RKSimon wrote:

Shouldn't it be fast-dpwssd?

https://github.com/llvm/llvm-project/pull/85033


More information about the llvm-commits mailing list