R600/SI: Add pattern for truncating i32 to i1
Fixes half a dozen piglit tests with radeonsi.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 200283
diff --git a/llvm/lib/Target/R600/SIInstructions.td b/llvm/lib/Target/R600/SIInstructions.td
index a38121b..912b59a 100644
--- a/llvm/lib/Target/R600/SIInstructions.td
+++ b/llvm/lib/Target/R600/SIInstructions.td
@@ -2126,6 +2126,11 @@
(EXTRACT_SUBREG $a, sub0)
>;
+def : Pat <
+ (i1 (trunc i32:$a)),
+ (V_CMP_EQ_I32_e64 (V_AND_B32_e32 (i32 1), $a), 1)
+>;
+
// V_ADD_I32_e32/S_ADD_I32 produces carry in VCC/SCC. For the vector
// case, the sgpr-copies pass will fix this to use the vector version.
def : Pat <
diff --git a/llvm/test/CodeGen/R600/trunc.ll b/llvm/test/CodeGen/R600/trunc.ll
index 0bd320a..6bbd7f7 100644
--- a/llvm/test/CodeGen/R600/trunc.ll
+++ b/llvm/test/CodeGen/R600/trunc.ll
@@ -28,3 +28,13 @@
store i32 %result, i32 addrspace(1)* %out, align 4
ret void
}
+
+; SI-LABEL: @trunc_i32_to_i1:
+; SI: V_AND_B32
+; SI: V_CMP_EQ_I32
+define void @trunc_i32_to_i1(i32 addrspace(1)* %out, i32 %a) {
+ %trunc = trunc i32 %a to i1
+ %result = select i1 %trunc, i32 1, i32 0
+ store i32 %result, i32 addrspace(1)* %out, align 4
+ ret void
+}