R600/SI: Add intrinsic for retrieving the current thread ID

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186010 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/R600/SIInstructions.td b/lib/Target/R600/SIInstructions.td
index 09460d83a3fa8200bbe110026d1b272ddb8ef0ab..61755b47a0308adc0f327ca9c025931f957c35ca 100644 (file)
--- a/lib/Target/R600/SIInstructions.td
+++ b/lib/Target/R600/SIInstructions.td
@@ -903,8 +903,8 @@ defm V_MAC_F32 : VOP2_32 <0x0000001f, "V_MAC_F32", []>;
 defm V_MADMK_F32 : VOP2_32 <0x00000020, "V_MADMK_F32", []>;
 defm V_MADAK_F32 : VOP2_32 <0x00000021, "V_MADAK_F32", []>;
 //defm V_BCNT_U32_B32 : VOP2_32 <0x00000022, "V_BCNT_U32_B32", []>;
-//defm V_MBCNT_LO_U32_B32 : VOP2_32 <0x00000023, "V_MBCNT_LO_U32_B32", []>;
-//defm V_MBCNT_HI_U32_B32 : VOP2_32 <0x00000024, "V_MBCNT_HI_U32_B32", []>;
+defm V_MBCNT_LO_U32_B32 : VOP2_32 <0x00000023, "V_MBCNT_LO_U32_B32", []>;
+defm V_MBCNT_HI_U32_B32 : VOP2_32 <0x00000024, "V_MBCNT_HI_U32_B32", []>;
 
 let isCommutable = 1, Defs = [VCC] in { // Carry-out goes to VCC
 defm V_ADD_I32 : VOP2b_32 <0x00000025, "V_ADD_I32",
@@ -1575,6 +1575,12 @@ def : Pat <
                    (V_RCP_IFLAG_F32_e32 (V_CVT_F32_U32_e32 $src0))))
 >;
 
+def : Pat <
+  (int_SI_tid),
+  (V_MBCNT_HI_U32_B32_e32 0xffffffff,
+                          (V_MBCNT_LO_U32_B32_e64 0xffffffff, 0, 0, 0, 0, 0))
+>;
+
 /********** ================== **********/
 /**********   VOP3 Patterns    **********/
 /********** ================== **********/

diff --git a/lib/Target/R600/SIIntrinsics.td b/lib/Target/R600/SIIntrinsics.td
index d2643e086b4489fd6fcc1dcd2f9eed19e209da61..2fa073ea09958995d13fcdb5db27a4ad3d2f5ee2 100644 (file)
--- a/lib/Target/R600/SIIntrinsics.td
+++ b/lib/Target/R600/SIIntrinsics.td
@@ -14,6 +14,7 @@
 
 let TargetPrefix = "SI", isTarget = 1 in {
 
+  def int_SI_tid : Intrinsic <[llvm_i32_ty], [], [IntrNoMem]>;
   def int_SI_packf16 : Intrinsic <[llvm_i32_ty], [llvm_float_ty, llvm_float_ty], [IntrNoMem]>;
   def int_SI_export : Intrinsic <[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty, llvm_i32_ty, llvm_i32_ty, llvm_float_ty, llvm_float_ty, llvm_float_ty, llvm_float_ty], []>;
   def int_SI_load_const : Intrinsic <[llvm_float_ty], [llvm_v16i8_ty, llvm_i32_ty], [IntrNoMem]>;

diff --git a/test/CodeGen/R600/llvm.SI.tid.ll b/test/CodeGen/R600/llvm.SI.tid.ll
new file mode 100644 (file)
index 0000000..238d9f2
--- /dev/null
+++ b/test/CodeGen/R600/llvm.SI.tid.ll
@@ -0,0 +1,16 @@
+;RUN: llc < %s -march=r600 -mcpu=verde | FileCheck %s
+
+;CHECK: V_MBCNT_LO_U32_B32_e64
+;CHECK: V_MBCNT_HI_U32_B32_e32
+
+define void @main(<16 x i8> addrspace(2)* inreg, <16 x i8> addrspace(2)* inreg, <32 x i8> addrspace(2)* inreg, i32 inreg) "ShaderType"="0" {
+main_body:
+  %4 = call i32 @llvm.SI.tid()
+  %5 = bitcast i32 %4 to float
+  call void @llvm.SI.export(i32 15, i32 1, i32 1, i32 0, i32 1, float %5, float %5, float %5, float %5)
+  ret void
+}
+
+declare i32 @llvm.SI.tid() readnone
+
+declare void @llvm.SI.export(i32, i32, i32, i32, i32, float, float, float, float)