R600/SI: Expand all v8[if]32 operations

5 files changed, 95 insertions, 17 deletions

diff --git a/lib/Target/R600/AMDGPUISelLowering.cpp b/lib/Target/R600/AMDGPUISelLowering.cpp
index b155267dd2f..53c2a88725c 100644
--- a/lib/Target/R600/AMDGPUISelLowering.cpp
+++ b/lib/Target/R600/AMDGPUISelLowering.cpp
@@ -120,8 +120,14 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(TargetMachine &TM) :
 
   setOperationAction(ISD::CONCAT_VECTORS, MVT::v4i32, Custom);
   setOperationAction(ISD::CONCAT_VECTORS, MVT::v4f32, Custom);
-  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v2i32, Custom);
+  setOperationAction(ISD::CONCAT_VECTORS, MVT::v8i32, Custom);
+  setOperationAction(ISD::CONCAT_VECTORS, MVT::v8f32, Custom);
   setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v2f32, Custom);
+  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v2i32, Custom);
+  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v4f32, Custom);
+  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v4i32, Custom);
+  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v8f32, Custom);
+  setOperationAction(ISD::EXTRACT_SUBVECTOR, MVT::v8i32, Custom);
 
   setLoadExtAction(ISD::EXTLOAD, MVT::v2i8, Expand);
   setLoadExtAction(ISD::SEXTLOAD, MVT::v2i8, Expand);
diff --git a/lib/Target/R600/SIISelLowering.cpp b/lib/Target/R600/SIISelLowering.cpp
index 36dd3cf7f0b..4696fdaf219 100644
--- a/lib/Target/R600/SIISelLowering.cpp
+++ b/lib/Target/R600/SIISelLowering.cpp
@@ -97,6 +97,7 @@ SITargetLowering::SITargetLowering(TargetMachine &TM) :
   setOperationAction(ISD::LOAD, MVT::i64, Custom);
   setOperationAction(ISD::LOAD, MVT::v2i32, Custom);
   setOperationAction(ISD::LOAD, MVT::v4i32, Custom);
+  setOperationAction(ISD::LOAD, MVT::v8i32, Custom);
 
   setOperationAction(ISD::STORE, MVT::i32, Custom);
   setOperationAction(ISD::STORE, MVT::i64, Custom);
@@ -146,6 +147,33 @@ SITargetLowering::SITargetLowering(TargetMachine &TM) :
   setOperationAction(ISD::GlobalAddress, MVT::i64, Custom);
   setOperationAction(ISD::FrameIndex, MVT::i32, Custom);
 
+  // We only support LOAD/STORE and vector manipulation ops for vectors
+  // with > 4 elements.
+  MVT VecTypes[] = {
+    MVT::v8i32, MVT::v8f32
+  };
+
+  const size_t NumVecTypes = array_lengthof(VecTypes);
+  for (unsigned Type = 0; Type < NumVecTypes; ++Type) {
+    for (unsigned Op = 0; Op < ISD::BUILTIN_OP_END; ++Op) {
+      switch(Op) {
+      case ISD::LOAD:
+      case ISD::STORE:
+      case ISD::BUILD_VECTOR:
+      case ISD::BITCAST:
+      case ISD::EXTRACT_VECTOR_ELT:
+      case ISD::INSERT_VECTOR_ELT:
+      case ISD::CONCAT_VECTORS:
+      case ISD::INSERT_SUBVECTOR:
+      case ISD::EXTRACT_SUBVECTOR:
+        break;
+      default:
+        setOperationAction(Op, VecTypes[Type], Expand);
+        break;
+      }
+    }
+  }
+
   setTargetDAGCombine(ISD::SELECT_CC);
 
   setTargetDAGCombine(ISD::SETCC);
diff --git a/lib/Target/R600/SIInstructions.td b/lib/Target/R600/SIInstructions.td
index 912b59a9e7b..cdd34e2bcfa 100644
--- a/lib/Target/R600/SIInstructions.td
+++ b/lib/Target/R600/SIInstructions.td
@@ -1668,6 +1668,8 @@ def : BitConvert <v4i32, v4f32, VReg_128>;
 def : BitConvert <v4i32, i128,  VReg_128>;
 def : BitConvert <i128, v4i32,  VReg_128>;
 
+def : BitConvert <v8f32, v8i32, SReg_256>;
+def : BitConvert <v8i32, v8f32, SReg_256>;
 def : BitConvert <v8i32, v32i8, SReg_256>;
 def : BitConvert <v32i8, v8i32, SReg_256>;
 def : BitConvert <v8i32, v32i8, VReg_256>;
diff --git a/test/CodeGen/R600/add.ll b/test/CodeGen/R600/add.ll
index 3d5506bfa5d..e4e7bc6c7f8 100644
--- a/test/CodeGen/R600/add.ll
+++ b/test/CodeGen/R600/add.ll
@@ -1,10 +1,9 @@
-; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=EG-CHECK %s
-; RUN: llc < %s -march=r600 -mcpu=verde -verify-machineinstrs | FileCheck --check-prefix=SI-CHECK %s
+; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=EG-CHECK --check-prefix=FUNC %s
+; RUN: llc < %s -march=r600 -mcpu=verde -verify-machineinstrs | FileCheck --check-prefix=SI-CHECK --check-prefix=FUNC %s
 
-;EG-CHECK-LABEL: @test1:
+;FUNC-LABEL: @test1:
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI-CHECK-LABEL: @test1:
 ;SI-CHECK: V_ADD_I32_e32 [[REG:v[0-9]+]], {{v[0-9]+, v[0-9]+}}
 ;SI-CHECK-NOT: [[REG]]
 ;SI-CHECK: BUFFER_STORE_DWORD [[REG]],
@@ -17,11 +16,10 @@ define void @test1(i32 addrspace(1)* %out, i32 addrspace(1)* %in) {
   ret void
 }
 
-;EG-CHECK-LABEL: @test2:
+;FUNC-LABEL: @test2:
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI-CHECK-LABEL: @test2:
 ;SI-CHECK: V_ADD_I32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
 ;SI-CHECK: V_ADD_I32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
 
@@ -34,13 +32,12 @@ define void @test2(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) {
   ret void
 }
 
-;EG-CHECK-LABEL: @test4:
+;FUNC-LABEL: @test4:
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ;EG-CHECK: ADD_INT {{[* ]*}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
-;SI-CHECK-LABEL: @test4:
 ;SI-CHECK: V_ADD_I32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
 ;SI-CHECK: V_ADD_I32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
 ;SI-CHECK: V_ADD_I32_e32 v{{[0-9]+, v[0-9]+, v[0-9]+}}
@@ -54,3 +51,27 @@ define void @test4(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
   store <4 x i32> %result, <4 x i32> addrspace(1)* %out
   ret void
 }
+
+; FUNC-LABEL: @test8
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; EG-CHECK: ADD_INT
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+; SI-CHECK: S_ADD_I32
+define void @test8(<8 x i32> addrspace(1)* %out, <8 x i32> %a, <8 x i32> %b) {
+entry:
+  %0 = add <8 x i32> %a, %b
+  store <8 x i32> %0, <8 x i32> addrspace(1)* %out
+  ret void
+}
diff --git a/test/CodeGen/R600/fadd.ll b/test/CodeGen/R600/fadd.ll
index f467bb78577..5d2b806039a 100644
--- a/test/CodeGen/R600/fadd.ll
+++ b/test/CodeGen/R600/fadd.ll
@@ -1,9 +1,8 @@
-; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s --check-prefix=R600-CHECK
-; RUN: llc < %s -march=r600 -mcpu=SI -verify-machineinstrs | FileCheck %s --check-prefix=SI-CHECK
+; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s --check-prefix=R600-CHECK --check-prefix=FUNC
+; RUN: llc < %s -march=r600 -mcpu=SI -verify-machineinstrs | FileCheck %s --check-prefix=SI-CHECK --check-prefix=FUNC
 
-; R600-CHECK: @fadd_f32
+; FUNC-LABEL: @fadd_f32
 ; R600-CHECK: ADD {{\** *}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z, KC0[2].W
-; SI-CHECK: @fadd_f32
 ; SI-CHECK: V_ADD_F32
 define void @fadd_f32(float addrspace(1)* %out, float %a, float %b) {
 entry:
@@ -12,10 +11,9 @@ entry:
    ret void
 }
 
-; R600-CHECK: @fadd_v2f32
+; FUNC-LABEL: @fadd_v2f32
 ; R600-CHECK-DAG: ADD {{\** *}}T{{[0-9]\.[XYZW]}}, KC0[3].X, KC0[3].Z
 ; R600-CHECK-DAG: ADD {{\** *}}T{{[0-9]\.[XYZW]}}, KC0[2].W, KC0[3].Y
-; SI-CHECK: @fadd_v2f32
 ; SI-CHECK: V_ADD_F32
 ; SI-CHECK: V_ADD_F32
 define void @fadd_v2f32(<2 x float> addrspace(1)* %out, <2 x float> %a, <2 x float> %b) {
@@ -25,12 +23,11 @@ entry:
   ret void
 }
 
-; R600-CHECK: @fadd_v4f32
+; FUNC-LABEL: @fadd_v4f32
 ; R600-CHECK: ADD {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-CHECK: ADD {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-CHECK: ADD {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-CHECK: ADD {{\** *}}T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
-; SI-CHECK: @fadd_v4f32
 ; SI-CHECK: V_ADD_F32
 ; SI-CHECK: V_ADD_F32
 ; SI-CHECK: V_ADD_F32
@@ -43,3 +40,27 @@ define void @fadd_v4f32(<4 x float> addrspace(1)* %out, <4 x float> addrspace(1)
   store <4 x float> %result, <4 x float> addrspace(1)* %out
   ret void
 }
+
+; FUNC-LABEL: @fadd_v8f32
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; R600-CHECK: ADD
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+; SI-CHECK: V_ADD_F32
+define void @fadd_v8f32(<8 x float> addrspace(1)* %out, <8 x float> %a, <8 x float> %b) {
+entry:
+  %0 = fadd <8 x float> %a, %b
+  store <8 x float> %0, <8 x float> addrspace(1)* %out
+  ret void
+}