From: Tom Stellard <thomas.stellard@amd.com>
Date: Thu, 18 Jul 2013 21:43:48 +0000 (+0000)
Subject: R600/SI: Add support for v2f32 loads
X-Git-Url: http://plrg.eecs.uci.edu/git/?a=commitdiff_plain;h=ac85f3f65ce67f71bb8e4626e0a50d818500e426;p=oota-llvm.git

R600/SI: Add support for v2f32 loads

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186615 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/R600/AMDGPUISelLowering.cpp b/lib/Target/R600/AMDGPUISelLowering.cpp
index 666e1580918..2a4e44f8624 100644
--- a/lib/Target/R600/AMDGPUISelLowering.cpp
+++ b/lib/Target/R600/AMDGPUISelLowering.cpp
@@ -69,6 +69,9 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(TargetMachine &TM) :
   setOperationAction(ISD::LOAD, MVT::f32, Promote);
   AddPromotedToType(ISD::LOAD, MVT::f32, MVT::i32);
 
+  setOperationAction(ISD::LOAD, MVT::v2f32, Promote);
+  AddPromotedToType(ISD::LOAD, MVT::v2f32, MVT::v2i32);
+
   setOperationAction(ISD::LOAD, MVT::v4f32, Promote);
   AddPromotedToType(ISD::LOAD, MVT::v4f32, MVT::v4i32);
 
diff --git a/lib/Target/R600/SIInstructions.td b/lib/Target/R600/SIInstructions.td
index 76b73de9fb4..c7d97c9d323 100644
--- a/lib/Target/R600/SIInstructions.td
+++ b/lib/Target/R600/SIInstructions.td
@@ -1707,6 +1707,7 @@ multiclass SMRD_Pattern <SMRD Instr_IMM, SMRD Instr_SGPR, ValueType vt> {
 defm : SMRD_Pattern <S_LOAD_DWORD_IMM, S_LOAD_DWORD_SGPR, f32>;
 defm : SMRD_Pattern <S_LOAD_DWORD_IMM, S_LOAD_DWORD_SGPR, i32>;
 defm : SMRD_Pattern <S_LOAD_DWORDX2_IMM, S_LOAD_DWORDX2_SGPR, i64>;
+defm : SMRD_Pattern <S_LOAD_DWORDX2_IMM, S_LOAD_DWORDX2_SGPR, v2i32>;
 defm : SMRD_Pattern <S_LOAD_DWORDX4_IMM, S_LOAD_DWORDX4_SGPR, v16i8>;
 defm : SMRD_Pattern <S_LOAD_DWORDX8_IMM, S_LOAD_DWORDX8_SGPR, v32i8>;
 
diff --git a/lib/Target/R600/SIRegisterInfo.td b/lib/Target/R600/SIRegisterInfo.td
index 244d4c00348..292b9d23db5 100644
--- a/lib/Target/R600/SIRegisterInfo.td
+++ b/lib/Target/R600/SIRegisterInfo.td
@@ -153,7 +153,7 @@ def SReg_32 : RegisterClass<"AMDGPU", [f32, i32], 32,
   (add SGPR_32, M0Reg)
 >;
 
-def SReg_64 : RegisterClass<"AMDGPU", [i64, i1], 64,
+def SReg_64 : RegisterClass<"AMDGPU", [v2i32, i64, i1], 64,
   (add SGPR_64, VCCReg, EXECReg)
 >;
 
diff --git a/test/CodeGen/R600/load.ll b/test/CodeGen/R600/load.ll
index 44c089ba692..a1b15bd0b05 100644
--- a/test/CodeGen/R600/load.ll
+++ b/test/CodeGen/R600/load.ll
@@ -41,6 +41,20 @@ entry:
   ret void
 }
 
+; load a v2f32 value from the global address space
+; R600-CHECK: @load_v2f32
+; R600-CHECK: VTX_READ_32
+; R600-CHECK: VTX_READ_32
+
+; SI-CHECK: @load_v2f32
+; SI-CHECK: BUFFER_LOAD_DWORDX2
+define void @load_v2f32(<2 x float> addrspace(1)* %out, <2 x float> addrspace(1)* %in) {
+entry:
+  %0 = load <2 x float> addrspace(1)* %in
+  store <2 x float> %0, <2 x float> addrspace(1)* %out
+  ret void
+}
+
 ; Load an i32 value from the constant address space.
 ; R600-CHECK: @load_const_addrspace_i32
 ; R600-CHECK: VTX_READ_32 T{{[0-9]+}}.X, T{{[0-9]+}}.X, 0