From: Olivier Sallenave <ohsallen@us.ibm.com>
Date: Wed, 14 Jan 2015 14:47:24 +0000 (+0000)
Subject: Override the TLI callback enableAggressiveFMAFusion and return true. Indeed, fmul... 
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=01cef58e6b27611a6d9d2967ef9f9441dc730c0f;p=oota-llvm.git

Override the TLI callback enableAggressiveFMAFusion and return true. Indeed, fmul, fmadd and fadd nodes cost the same number of cycles, so we can enable more combining heuristics to produce more fmadd nodes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225984 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/NVPTX/NVPTXISelLowering.h b/lib/Target/NVPTX/NVPTXISelLowering.h
index d66d81a5316..b3fea3f4a36 100644
--- a/lib/Target/NVPTX/NVPTXISelLowering.h
+++ b/lib/Target/NVPTX/NVPTXISelLowering.h
@@ -507,6 +507,8 @@ public:
 
   bool isFMAFasterThanFMulAndFAdd(EVT) const override { return true; }
 
+  bool enableAggressiveFMAFusion(EVT VT) const override { return true; }
+
 private:
   const NVPTXSubtarget &nvptxSubtarget; // cache the subtarget here