[MLIR][LLVMDialect] Added bitreverse and ctpop intrinsics

[georgemitenkov]

Summary

Introduced llvm.intr.bitreverse and llvm.intr.ctpop LLVM bit intrinsics to LLVM dialect. These intrinsics help with SPIR-V to LLVM conversion, allowing a direct mapping from spv.BitReverse and spv.BitCount respectively. Tests are added to roundtrip.mlir and llvm-intrinsics.mlir.

[ftynse]

mlir/test/Dialect/LLVMIR/roundtrip.mlir

106

Let's not pattern-match against SSA value name - llvm.intr.bitreverse"(%{{.*}}) - I know this file needs to be updated in that direction, but at least let's have new code clean.

[georgemitenkov]

• [MLIR][LLVMDialect] Removed pattern match on args

Removed pattern-match against SSA value name when testing bitreverse and ctpop intrinsivs in roundtrip.mlir.

mlir/include/mlir/Dialect/LLVMIR/LLVMOps.td

 def LLVM_Log2Op : LLVM_UnaryIntrinsicOp<"log2">;
 def LLVM_LogOp : LLVM_UnaryIntrinsicOp<"log">;
 def LLVM_Prefetch : LLVM_ZeroResultIntrOp<"prefetch", [0]>,
                     Arguments<(ins LLVM_Type:$addr, LLVM_Type:$rw,
                                LLVM_Type:$hint, LLVM_Type:$cache)>;
 def LLVM_SinOp : LLVM_UnaryIntrinsicOp<"sin">;
 def LLVM_SqrtOp : LLVM_UnaryIntrinsicOp<"sqrt">;
 def LLVM_PowOp : LLVM_BinarySameArgsIntrinsicOp<"pow">;
+def LLVM_BitReverseOp : LLVM_UnaryIntrinsicOp<"bitreverse">;
+def LLVM_CtPopOp : LLVM_UnaryIntrinsicOp<"ctpop">;
 
 //
 // Vector Reductions.
 //
 
 def LLVM_experimental_vector_reduce_add : LLVM_VectorReduction<"add">;
 def LLVM_experimental_vector_reduce_and : LLVM_VectorReduction<"and">;
 def LLVM_experimental_vector_reduce_mul : LLVM_VectorReduction<"mul">;

mlir/test/Dialect/LLVMIR/roundtrip.mlir

// CHECK:  %29 = llvm.fneg %arg1 : !llvm.float
   %29 = llvm.fneg %arg1 : !llvm.float
 
 // CHECK: "llvm.intr.sin"(%arg1) : (!llvm.float) -> !llvm.float
   %30 = "llvm.intr.sin"(%arg1) : (!llvm.float) -> !llvm.float
 
 // CHECK: "llvm.intr.pow"(%arg1, %arg1) : (!llvm.float, !llvm.float) -> !llvm.float
   %31 = "llvm.intr.pow"(%arg1, %arg1) : (!llvm.float, !llvm.float) -> !llvm.float
 
+// CHECK: "llvm.intr.bitreverse"(%{{.*}}) : (!llvm.i32) -> !llvm.i32
+  %32 = "llvm.intr.bitreverse"(%arg0) : (!llvm.i32) -> !llvm.i32
+
+// CHECK: "llvm.intr.ctpop"(%{{.*}}) : (!llvm.i32) -> !llvm.i32
+  %33 = "llvm.intr.ctpop"(%arg0) : (!llvm.i32) -> !llvm.i32
+
 // CHECK:  llvm.return
   llvm.return
 }
 
 // An larger self-contained function.
 // CHECK-LABEL:func @foo(%arg0: !llvm.i32) -> !llvm<"{ i32, double, i32 }"> {
 func @foo(%arg0: !llvm.i32) -> !llvm<"{ i32, double, i32 }"> {
 // CHECK-NEXT:  %0 = llvm.mlir.constant(3 : i64) : !llvm.i32
func @useFenceInst() {
   // CHECK:  syncscope("agent") seq_cst 
   llvm.fence syncscope("agent") seq_cst
   // CHECK:  seq_cst 
   llvm.fence syncscope("") seq_cst
   // CHECK:  release
   llvm.fence release
   return
 }
 No newline at end of file

mlir/test/Target/llvmir-intrinsics.mlir

 llvm.func @pow_test(%arg0: !llvm.float, %arg1: !llvm.float, %arg2: !llvm<"<8 x float>">, %arg3: !llvm<"<8 x float>">) {
   // CHECK: call float @llvm.pow.f32
   "llvm.intr.pow"(%arg0, %arg1) : (!llvm.float, !llvm.float) -> !llvm.float
   // CHECK: call <8 x float> @llvm.pow.v8f32
   "llvm.intr.pow"(%arg2, %arg3) : (!llvm<"<8 x float>">, !llvm<"<8 x float>">) -> !llvm<"<8 x float>">
   llvm.return
 }
 
+// CHECK-LABEL: @bitreverse_test
+llvm.func @bitreverse_test(%arg0: !llvm.i32, %arg1: !llvm<"<8 x i32>">) {
+  // CHECK: call i32 @llvm.bitreverse.i32
+  "llvm.intr.bitreverse"(%arg0) : (!llvm.i32) -> !llvm.i32
+  // CHECK: call <8 x i32> @llvm.bitreverse.v8i32
+  "llvm.intr.bitreverse"(%arg1) : (!llvm<"<8 x i32>">) -> !llvm<"<8 x i32>">
+  llvm.return
+}
+
+// CHECK-LABEL: @ctpop_test
+llvm.func @ctpop_test(%arg0: !llvm.i32, %arg1: !llvm<"<8 x i32>">) {
+  // CHECK: call i32 @llvm.ctpop.i32
+  "llvm.intr.ctpop"(%arg0) : (!llvm.i32) -> !llvm.i32
+  // CHECK: call <8 x i32> @llvm.ctpop.v8i32
+  "llvm.intr.ctpop"(%arg1) : (!llvm<"<8 x i32>">) -> !llvm<"<8 x i32>">
+  llvm.return
+}
+
 // CHECK-LABEL: @vector_reductions
 llvm.func @vector_reductions(%arg0: !llvm.float, %arg1: !llvm<"<8 x float>">, %arg2: !llvm<"<8 x i32>">) {
   // CHECK: call i32 @llvm.experimental.vector.reduce.add.v8i32
   "llvm.intr.experimental.vector.reduce.add"(%arg2) : (!llvm<"<8 x i32>">) -> !llvm.i32
   // CHECK: call i32 @llvm.experimental.vector.reduce.and.v8i32
   "llvm.intr.experimental.vector.reduce.and"(%arg2) : (!llvm<"<8 x i32>">) -> !llvm.i32
   // CHECK: call float @llvm.experimental.vector.reduce.fmax.v8f32
   "llvm.intr.experimental.vector.reduce.fmax"(%arg1) : (!llvm<"<8 x float>">) -> !llvm.float