IR: Give 'DI' prefix to debug info metadata

[oota-llvm.git] / test / CodeGen / R600 / setcc.ll

diff --git a/test/CodeGen/R600/setcc.ll b/test/CodeGen/R600/setcc.ll
index ad72732cab0aa476c81e28aafb46ac353254a308..f33a82df5ffb5557c9d8f33e2274854eee8677ec 100644 (file)
--- a/test/CodeGen/R600/setcc.ll
+++ b/test/CodeGen/R600/setcc.ll
@@ -1,7 +1,9 @@
-;RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=R600 --check-prefix=FUNC %s
-;RUN: llc < %s -march=r600 -mcpu=SI -verify-machineinstrs| FileCheck --check-prefix=SI --check-prefix=FUNC %s
+; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=R600 --check-prefix=FUNC %s
+; RUN: llc < %s -march=amdgcn -mcpu=SI -verify-machineinstrs | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 
-; FUNC-LABEL: @setcc_v2i32
+declare i32 @llvm.r600.read.tidig.x() nounwind readnone
+
+; FUNC-LABEL: {{^}}setcc_v2i32:
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW]}}, KC0[3].X, KC0[3].Z
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW]}}, KC0[2].W, KC0[3].Y
 
@@ -12,16 +14,16 @@ define void @setcc_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> %a, <2 x i32> %
   ret void
 }
 
-; FUNC-LABEL: @setcc_v4i32
+; FUNC-LABEL: {{^}}setcc_v4i32:
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 ; R600-DAG: SETE_INT * T{{[0-9]+\.[XYZW], T[0-9]+\.[XYZW], T[0-9]+\.[XYZW]}}
 
 define void @setcc_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) {
-  %b_ptr = getelementptr <4 x i32> addrspace(1)* %in, i32 1
-  %a = load <4 x i32> addrspace(1) * %in
-  %b = load <4 x i32> addrspace(1) * %b_ptr
+  %b_ptr = getelementptr <4 x i32>, <4 x i32> addrspace(1)* %in, i32 1
+  %a = load <4 x i32>, <4 x i32> addrspace(1) * %in
+  %b = load <4 x i32>, <4 x i32> addrspace(1) * %b_ptr
   %result = icmp eq <4 x i32> %a, %b
   %sext = sext <4 x i1> %result to <4 x i32>
   store <4 x i32> %sext, <4 x i32> addrspace(1)* %out
@@ -32,9 +34,9 @@ define void @setcc_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %
 ;; Float comparisons
 ;;;==========================================================================;;;
 
-; FUNC-LABEL: @f32_oeq
+; FUNC-LABEL: {{^}}f32_oeq:
 ; R600: SETE_DX10
-; SI: V_CMP_EQ_F32
+; SI: v_cmp_eq_f32
 define void @f32_oeq(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp oeq float %a, %b
@@ -43,9 +45,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ogt
+; FUNC-LABEL: {{^}}f32_ogt:
 ; R600: SETGT_DX10
-; SI: V_CMP_GT_F32
+; SI: v_cmp_gt_f32
 define void @f32_ogt(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ogt float %a, %b
@@ -54,9 +56,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_oge
+; FUNC-LABEL: {{^}}f32_oge:
 ; R600: SETGE_DX10
-; SI: V_CMP_GE_F32
+; SI: v_cmp_ge_f32
 define void @f32_oge(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp oge float %a, %b
@@ -65,9 +67,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_olt
+; FUNC-LABEL: {{^}}f32_olt:
 ; R600: SETGT_DX10
-; SI: V_CMP_LT_F32
+; SI: v_cmp_lt_f32
 define void @f32_olt(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp olt float %a, %b
@@ -76,9 +78,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ole
+; FUNC-LABEL: {{^}}f32_ole:
 ; R600: SETGE_DX10
-; SI: V_CMP_LE_F32
+; SI: v_cmp_le_f32
 define void @f32_ole(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ole float %a, %b
@@ -87,16 +89,16 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_one
+; FUNC-LABEL: {{^}}f32_one:
 ; R600-DAG: SETE_DX10
 ; R600-DAG: SETE_DX10
 ; R600-DAG: AND_INT
 ; R600-DAG: SETNE_DX10
 ; R600-DAG: AND_INT
 ; R600-DAG: SETNE_INT
-; SI: V_CMP_O_F32
-; SI: V_CMP_NEQ_F32
-; SI: S_AND_B64
+
+; SI: v_cmp_lg_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_one(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp one float %a, %b
@@ -105,12 +107,12 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ord
+; FUNC-LABEL: {{^}}f32_ord:
 ; R600-DAG: SETE_DX10
 ; R600-DAG: SETE_DX10
 ; R600-DAG: AND_INT
 ; R600-DAG: SETNE_INT
-; SI: V_CMP_O_F32
+; SI: v_cmp_o_f32
 define void @f32_ord(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ord float %a, %b
@@ -119,16 +121,16 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ueq
+; FUNC-LABEL: {{^}}f32_ueq:
 ; R600-DAG: SETNE_DX10
 ; R600-DAG: SETNE_DX10
 ; R600-DAG: OR_INT
 ; R600-DAG: SETE_DX10
 ; R600-DAG: OR_INT
 ; R600-DAG: SETNE_INT
-; SI: V_CMP_U_F32
-; SI: V_CMP_EQ_F32
-; SI: S_OR_B64
+
+; SI: v_cmp_nlg_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_ueq(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ueq float %a, %b
@@ -137,12 +139,11 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ugt
+; FUNC-LABEL: {{^}}f32_ugt:
 ; R600: SETGE
 ; R600: SETE_DX10
-; SI: V_CMP_U_F32
-; SI: V_CMP_GT_F32
-; SI: S_OR_B64
+; SI: v_cmp_nle_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_ugt(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ugt float %a, %b
@@ -151,12 +152,12 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_uge
+; FUNC-LABEL: {{^}}f32_uge:
 ; R600: SETGT
 ; R600: SETE_DX10
-; SI: V_CMP_U_F32
-; SI: V_CMP_GE_F32
-; SI: S_OR_B64
+
+; SI: v_cmp_nlt_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_uge(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp uge float %a, %b
@@ -165,12 +166,12 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ult
+; FUNC-LABEL: {{^}}f32_ult:
 ; R600: SETGE
 ; R600: SETE_DX10
-; SI: V_CMP_U_F32
-; SI: V_CMP_LT_F32
-; SI: S_OR_B64
+
+; SI: v_cmp_nge_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_ult(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ult float %a, %b
@@ -179,12 +180,12 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_ule
+; FUNC-LABEL: {{^}}f32_ule:
 ; R600: SETGT
 ; R600: SETE_DX10
-; SI: V_CMP_U_F32
-; SI: V_CMP_LE_F32
-; SI: S_OR_B64
+
+; SI: v_cmp_ngt_f32_e32 vcc
+; SI-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1, vcc
 define void @f32_ule(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp ule float %a, %b
@@ -193,9 +194,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_une
+; FUNC-LABEL: {{^}}f32_une:
 ; R600: SETNE_DX10
-; SI: V_CMP_NEQ_F32
+; SI: v_cmp_neq_f32
 define void @f32_une(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp une float %a, %b
@@ -204,12 +205,12 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @f32_uno
+; FUNC-LABEL: {{^}}f32_uno:
 ; R600: SETNE_DX10
 ; R600: SETNE_DX10
 ; R600: OR_INT
 ; R600: SETNE_INT
-; SI: V_CMP_U_F32
+; SI: v_cmp_u_f32
 define void @f32_uno(i32 addrspace(1)* %out, float %a, float %b) {
 entry:
   %0 = fcmp uno float %a, %b
@@ -222,9 +223,9 @@ entry:
 ;; 32-bit integer comparisons
 ;;;==========================================================================;;;
 
-; FUNC-LABEL: @i32_eq
+; FUNC-LABEL: {{^}}i32_eq:
 ; R600: SETE_INT
-; SI: V_CMP_EQ_I32
+; SI: v_cmp_eq_i32
 define void @i32_eq(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp eq i32 %a, %b
@@ -233,9 +234,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_ne
+; FUNC-LABEL: {{^}}i32_ne:
 ; R600: SETNE_INT
-; SI: V_CMP_NE_I32
+; SI: v_cmp_ne_i32
 define void @i32_ne(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp ne i32 %a, %b
@@ -244,9 +245,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_ugt
+; FUNC-LABEL: {{^}}i32_ugt:
 ; R600: SETGT_UINT
-; SI: V_CMP_GT_U32
+; SI: v_cmp_gt_u32
 define void @i32_ugt(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp ugt i32 %a, %b
@@ -255,9 +256,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_uge
+; FUNC-LABEL: {{^}}i32_uge:
 ; R600: SETGE_UINT
-; SI: V_CMP_GE_U32
+; SI: v_cmp_ge_u32
 define void @i32_uge(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp uge i32 %a, %b
@@ -266,9 +267,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_ult
+; FUNC-LABEL: {{^}}i32_ult:
 ; R600: SETGT_UINT
-; SI: V_CMP_LT_U32
+; SI: v_cmp_lt_u32
 define void @i32_ult(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp ult i32 %a, %b
@@ -277,9 +278,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_ule
+; FUNC-LABEL: {{^}}i32_ule:
 ; R600: SETGE_UINT
-; SI: V_CMP_LE_U32
+; SI: v_cmp_le_u32
 define void @i32_ule(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp ule i32 %a, %b
@@ -288,9 +289,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_sgt
+; FUNC-LABEL: {{^}}i32_sgt:
 ; R600: SETGT_INT
-; SI: V_CMP_GT_I32
+; SI: v_cmp_gt_i32
 define void @i32_sgt(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp sgt i32 %a, %b
@@ -299,9 +300,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_sge
+; FUNC-LABEL: {{^}}i32_sge:
 ; R600: SETGE_INT
-; SI: V_CMP_GE_I32
+; SI: v_cmp_ge_i32
 define void @i32_sge(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp sge i32 %a, %b
@@ -310,9 +311,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_slt
+; FUNC-LABEL: {{^}}i32_slt:
 ; R600: SETGT_INT
-; SI: V_CMP_LT_I32
+; SI: v_cmp_lt_i32
 define void @i32_slt(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp slt i32 %a, %b
@@ -321,9 +322,9 @@ entry:
   ret void
 }
 
-; FUNC-LABEL: @i32_sle
+; FUNC-LABEL: {{^}}i32_sle:
 ; R600: SETGE_INT
-; SI: V_CMP_LE_I32
+; SI: v_cmp_le_i32
 define void @i32_sle(i32 addrspace(1)* %out, i32 %a, i32 %b) {
 entry:
   %0 = icmp sle i32 %a, %b
@@ -331,3 +332,46 @@ entry:
   store i32 %1, i32 addrspace(1)* %out
   ret void
 }
+
+; FIXME: This does 4 compares
+; FUNC-LABEL: {{^}}v3i32_eq:
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI: s_endpgm
+define void @v3i32_eq(<3 x i32> addrspace(1)* %out, <3 x i32> addrspace(1)* %ptra, <3 x i32> addrspace(1)* %ptrb) {
+  %tid = call i32 @llvm.r600.read.tidig.x() nounwind readnone
+  %gep.a = getelementptr <3 x i32>, <3 x i32> addrspace(1)* %ptra, i32 %tid
+  %gep.b = getelementptr <3 x i32>, <3 x i32> addrspace(1)* %ptrb, i32 %tid
+  %gep.out = getelementptr <3 x i32>, <3 x i32> addrspace(1)* %out, i32 %tid
+  %a = load <3 x i32>, <3 x i32> addrspace(1)* %gep.a
+  %b = load <3 x i32>, <3 x i32> addrspace(1)* %gep.b
+  %cmp = icmp eq <3 x i32> %a, %b
+  %ext = sext <3 x i1> %cmp to <3 x i32>
+  store <3 x i32> %ext, <3 x i32> addrspace(1)* %gep.out
+  ret void
+}
+
+; FUNC-LABEL: {{^}}v3i8_eq:
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI-DAG: v_cmp_eq_i32
+; SI-DAG: v_cndmask_b32_e64 {{v[0-9]+}}, 0, -1,
+; SI: s_endpgm
+define void @v3i8_eq(<3 x i8> addrspace(1)* %out, <3 x i8> addrspace(1)* %ptra, <3 x i8> addrspace(1)* %ptrb) {
+  %tid = call i32 @llvm.r600.read.tidig.x() nounwind readnone
+  %gep.a = getelementptr <3 x i8>, <3 x i8> addrspace(1)* %ptra, i32 %tid
+  %gep.b = getelementptr <3 x i8>, <3 x i8> addrspace(1)* %ptrb, i32 %tid
+  %gep.out = getelementptr <3 x i8>, <3 x i8> addrspace(1)* %out, i32 %tid
+  %a = load <3 x i8>, <3 x i8> addrspace(1)* %gep.a
+  %b = load <3 x i8>, <3 x i8> addrspace(1)* %gep.b
+  %cmp = icmp eq <3 x i8> %a, %b
+  %ext = sext <3 x i1> %cmp to <3 x i8>
+  store <3 x i8> %ext, <3 x i8> addrspace(1)* %gep.out
+  ret void
+}