[llvm-branch-commits] [SPARC] Use umulxhi to do extending 64x64->128 multiply when we have VIS3 (PR #135714)

[Sergei Barannikov via llvm-branch-commits]

================
@@ -294,4 +294,13 @@ def : Pat<(f32 fpnegimm0), (FNEGS (FZEROS))>;
 // VIS3 instruction patterns.
 let Predicates = [HasVIS3] in {
 def : Pat<(i64 (adde i64:$lhs, i64:$rhs)), (ADDXCCC $lhs, $rhs)>;
+
+def : Pat<(i64 (mulhu i64:$lhs, i64:$rhs)), (UMULXHI $lhs, $rhs)>;
+// Signed "MULXHI".
+// Based on the formula presented in OSA2011 §7.140, but with bitops to select
+// the values to be added.
+def : Pat<(i64 (mulhs i64:$lhs, i64:$rhs)),
+      (SUBrr (UMULXHI $lhs, $rhs),
+             (ADDrr (ANDrr (SRAXri $lhs, 63), $rhs),
+                    (ANDrr (SRAXri $rhs, 63), $lhs)))>;
----------------
s-barannikov wrote:

Thanks for trying.
It probably needs more work at SparcTargetLowering for the expansion to be optimal. Can you add a TODO near the pattern that reads something like "Consider moving this expansion to DAG legalization phase"?

(For the record: I don't really understand the expansion and assume it is taken from a proved source, like Sparc architecture manual.)


https://github.com/llvm/llvm-project/pull/135714