[PATCH] D14906: [X86][FMA] Optimize FNEG(FMA) Patterns

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Mon Nov 23 06:15:18 PST 2015

RKSimon added inline comments.

Comment at: lib/Target/X86/X86ISelLowering.cpp:13400
@@ -13399,1 +13399,3 @@
+  // If we're negating a FMA node, then we can adjust the
+  // instruction to include the extra negation.
delena wrote:
> What we are doing in PerformFMACombine() ? May be we can extend there?
In PerformFMACombine we convert ISD::FMA nodes to the equivalent X86ISD FMADD/FMSUB/FNMADD/FNMSUB nodes based on which inputs are FNEG.

This patch deal with FNEG which has a X86ISD FMA type as its input - we don't have a PerformFNEGCombine but I could move this patch there if you think it necessary.



More information about the llvm-commits mailing list