commit c370d7b33d0aba8b80f3a0b633f4f8d92c448833
author Farhana Aleen <farhana.aleen@gmail.com> Mon Jul 16 18:19:59 2018 +0000
committer Farhana Aleen <farhana.aleen@gmail.com> Mon Jul 16 18:19:59 2018 +0000
tree b4e3f2030a2771b0aba9849c07db42611c322fb4
parent 7f01d209939c3a08c7d15ea145d7056dc138c23a

[AMDGPU] [AMDGPU] Support a fdot2 pattern.

Summary: Optimize fma((float)S0.x, (float)S1.x fma((float)S0.y, (float)S1.y, z))
                   -> fdot2((v2f16)S0, (v2f16)S1, (float)z)

Author: FarhanaAleen

Reviewed By: rampitec, b-sumner

Subscribers: AMDGPU

Differential Revision: https://reviews.llvm.org/D49146

llvm-svn: 337198

llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
index acdedab..485927c 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
@@ -3993,6 +3993,7 @@
   NODE_NAME_CASE(FMED3)
   NODE_NAME_CASE(SMED3)
   NODE_NAME_CASE(UMED3)
+  NODE_NAME_CASE(FDOT2)
   NODE_NAME_CASE(URECIP)
   NODE_NAME_CASE(DIV_SCALE)
   NODE_NAME_CASE(DIV_FMAS)