AMDGPU: Rename intrinsic to better match instruction name

[• tstellarAMD]

LGTM.

[arsenm]

r260780, corresponding clang change r260781

include/llvm/IR/IntrinsicsAMDGPU.td

 def int_amdgcn_rcp : Intrinsic<
   [llvm_anyfloat_ty], [LLVMMatchType<0>], [IntrNoMem]
 >;
 
 def int_amdgcn_rsq :  Intrinsic<
   [llvm_anyfloat_ty], [LLVMMatchType<0>], [IntrNoMem]
 >;
 
-def int_amdgcn_rsq_clamped :  Intrinsic<
+def int_amdgcn_rsq_clamp : Intrinsic<
   [llvm_anyfloat_ty], [LLVMMatchType<0>], [IntrNoMem]>;
 
 def int_amdgcn_ldexp : Intrinsic<
   [llvm_anyfloat_ty], [LLVMMatchType<0>, llvm_i32_ty], [IntrNoMem]
 >;
 
 def int_amdgcn_class : Intrinsic<
   [llvm_i1_ty], [llvm_anyfloat_ty, llvm_i32_ty], [IntrNoMem]

lib/Target/AMDGPU/AMDGPUISelLowering.h

enum NodeType : unsigned {
   DIV_FIXUP,
   TRIG_PREOP, // 1 ULP max error for f64
 
   // RCP, RSQ - For f32, 1 ULP max error, no denormal handling.
   //            For f64, max error 2^29 ULP, handles denormals.
   RCP,
   RSQ,
   RSQ_LEGACY,
-  RSQ_CLAMPED,
+  RSQ_CLAMP,
   LDEXP,
   FP_CLASS,
   DOT4,
   CARRY,
   BORROW,
   BFE_U32, // Extract range of bits with zero extension to 32-bits.
   BFE_I32, // Extract range of bits with sign extension to 32-bits.
   BFI, // (src0 & src1) | (~src0 & src2)

lib/Target/AMDGPU/AMDGPUISelLowering.cpp

const char* AMDGPUTargetLowering::getTargetNodeName(unsigned Opcode) const {
   NODE_NAME_CASE(URECIP)
   NODE_NAME_CASE(DIV_SCALE)
   NODE_NAME_CASE(DIV_FMAS)
   NODE_NAME_CASE(DIV_FIXUP)
   NODE_NAME_CASE(TRIG_PREOP)
   NODE_NAME_CASE(RCP)
   NODE_NAME_CASE(RSQ)
   NODE_NAME_CASE(RSQ_LEGACY)
-  NODE_NAME_CASE(RSQ_CLAMPED)
+  NODE_NAME_CASE(RSQ_CLAMP)
   NODE_NAME_CASE(LDEXP)
   NODE_NAME_CASE(FP_CLASS)
   NODE_NAME_CASE(DOT4)
   NODE_NAME_CASE(CARRY)
   NODE_NAME_CASE(BORROW)
   NODE_NAME_CASE(BFE_U32)
   NODE_NAME_CASE(BFE_I32)
   NODE_NAME_CASE(BFI)

lib/Target/AMDGPU/AMDGPUInstrInfo.td

 
 // out = 1.0 / sqrt(a)
 def AMDGPUrsq : SDNode<"AMDGPUISD::RSQ", SDTFPUnaryOp>;
 
 // out = 1.0 / sqrt(a)
 def AMDGPUrsq_legacy : SDNode<"AMDGPUISD::RSQ_LEGACY", SDTFPUnaryOp>;
 
 // out = 1.0 / sqrt(a) result clamped to +/- max_float.
-def AMDGPUrsq_clamped : SDNode<"AMDGPUISD::RSQ_CLAMPED", SDTFPUnaryOp>;
+def AMDGPUrsq_clamp : SDNode<"AMDGPUISD::RSQ_CLAMP", SDTFPUnaryOp>;
 
 def AMDGPUldexp : SDNode<"AMDGPUISD::LDEXP", AMDGPULdExpOp>;
 
 def AMDGPUfp_class : SDNode<"AMDGPUISD::FP_CLASS", AMDGPUFPClassOp>;
 
 // out = max(a, b) a and b are floats, where a nan comparison fails.
 // This is not commutative because this gives the second operand:
 //   x < nan ? x : nan -> nan

lib/Target/AMDGPU/R600ISelLowering.cpp

case Intrinsic::r600_read_tidig_y:
       return CreateLiveInRegister(DAG, &AMDGPU::R600_TReg32RegClass,
                                   AMDGPU::T0_Y, VT);
     case Intrinsic::r600_read_tidig_z:
       return CreateLiveInRegister(DAG, &AMDGPU::R600_TReg32RegClass,
                                   AMDGPU::T0_Z, VT);
 
     // FIXME: Should be renamed to r600 prefix
     case AMDGPUIntrinsic::AMDGPU_rsq_clamped:
-      return DAG.getNode(AMDGPUISD::RSQ_CLAMPED, DL, VT, Op.getOperand(1));
+      return DAG.getNode(AMDGPUISD::RSQ_CLAMP, DL, VT, Op.getOperand(1));
 
     case Intrinsic::r600_rsq:
     case AMDGPUIntrinsic::AMDGPU_rsq: // Legacy name
       // XXX - I'm assuming SI's RSQ_LEGACY matches R600's behavior.
       return DAG.getNode(AMDGPUISD::RSQ_LEGACY, DL, VT, Op.getOperand(1));
     }
     // break out of case ISD::INTRINSIC_WO_CHAIN in switch(Op.getOpcode())
     break;

lib/Target/AMDGPU/R600Instructions.td

 class RECIP_UINT_Common <bits<11> inst> : R600_1OP_Helper <
   inst, "RECIP_UINT", AMDGPUurecip
 > {
   let Itinerary = TransALU;
 }
 
 // Clamped to maximum.
 class RECIPSQRT_CLAMPED_Common <bits<11> inst> : R600_1OP_Helper <
-  inst, "RECIPSQRT_CLAMPED", AMDGPUrsq_clamped
+  inst, "RECIPSQRT_CLAMPED", AMDGPUrsq_clamp
 > {
   let Itinerary = TransALU;
 }
 
 class RECIPSQRT_IEEE_Common <bits<11> inst> : R600_1OP_Helper <
   inst, "RECIPSQRT_IEEE", AMDGPUrsq_legacy
 > {
   let Itinerary = TransALU;

lib/Target/AMDGPU/SIISelLowering.cpp

case Intrinsic::amdgcn_dispatch_ptr:
 
     return CreateLiveInRegister(DAG, &AMDGPU::SReg_64RegClass,
       TRI->getPreloadedValue(MF, SIRegisterInfo::DISPATCH_PTR), VT);
   case Intrinsic::amdgcn_rcp:
     return DAG.getNode(AMDGPUISD::RCP, DL, VT, Op.getOperand(1));
   case Intrinsic::amdgcn_rsq:
   case AMDGPUIntrinsic::AMDGPU_rsq: // Legacy name
     return DAG.getNode(AMDGPUISD::RSQ, DL, VT, Op.getOperand(1));
-  case Intrinsic::amdgcn_rsq_clamped:
+  case Intrinsic::amdgcn_rsq_clamp:
   case AMDGPUIntrinsic::AMDGPU_rsq_clamped: { // Legacy name
     if (Subtarget->getGeneration() < AMDGPUSubtarget::VOLCANIC_ISLANDS)
-      return DAG.getNode(AMDGPUISD::RSQ_CLAMPED, DL, VT, Op.getOperand(1));
+      return DAG.getNode(AMDGPUISD::RSQ_CLAMP, DL, VT, Op.getOperand(1));
 
     Type *Type = VT.getTypeForEVT(*DAG.getContext());
     APFloat Max = APFloat::getLargest(Type->getFltSemantics());
     APFloat Min = APFloat::getLargest(Type->getFltSemantics(), true);
 
     SDValue Rsq = DAG.getNode(AMDGPUISD::RSQ, DL, VT, Op.getOperand(1));
     SDValue Tmp = DAG.getNode(ISD::FMINNUM, DL, VT, Rsq,
                               DAG.getConstantFP(Max, DL, VT));

lib/Target/AMDGPU/SIInstructions.td

 
 let SchedRW = [WriteQuarterRate32] in {
 
 defm V_MOV_FED_B32 : VOP1InstSI <vop1<0x9>, "v_mov_fed_b32", VOP_I32_I32>;
 defm V_LOG_CLAMP_F32 : VOP1InstSI <vop1<0x26>, "v_log_clamp_f32", VOP_F32_F32>;
 defm V_RCP_CLAMP_F32 : VOP1InstSI <vop1<0x28>, "v_rcp_clamp_f32", VOP_F32_F32>;
 defm V_RCP_LEGACY_F32 : VOP1InstSI <vop1<0x29>, "v_rcp_legacy_f32", VOP_F32_F32>;
 defm V_RSQ_CLAMP_F32 : VOP1InstSI <vop1<0x2c>, "v_rsq_clamp_f32",
-  VOP_F32_F32, AMDGPUrsq_clamped
+  VOP_F32_F32, AMDGPUrsq_clamp
 >;
 defm V_RSQ_LEGACY_F32 : VOP1InstSI <vop1<0x2d>, "v_rsq_legacy_f32",
   VOP_F32_F32, AMDGPUrsq_legacy
 >;
 
 } // End SchedRW = [WriteQuarterRate32]
 
 let SchedRW = [WriteDouble] in {
 
 defm V_RCP_CLAMP_F64 : VOP1InstSI <vop1<0x30>, "v_rcp_clamp_f64", VOP_F64_F64>;
 defm V_RSQ_CLAMP_F64 : VOP1InstSI <vop1<0x32>, "v_rsq_clamp_f64",
-  VOP_F64_F64, AMDGPUrsq_clamped
+  VOP_F64_F64, AMDGPUrsq_clamp
 >;
 
 } // End SchedRW = [WriteDouble]
 
 } // End SubtargetPredicate = isSICI
 
 //===----------------------------------------------------------------------===//
 // VINTRP Instructions

test/CodeGen/AMDGPU/llvm.amdgcn.rsq.clamp.ll

+; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
+; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=FUNC %s
+
+declare float @llvm.amdgcn.rsq.clamp.f32(float) #1
+declare double @llvm.amdgcn.rsq.clamp.f64(double) #1
+
+; FUNC-LABEL: {{^}}rsq_clamp_f32:
+; SI: v_rsq_clamp_f32_e32
+
+; VI: s_load_dword [[SRC:s[0-9]+]]
+; VI: v_rsq_f32_e32 [[RSQ:v[0-9]+]], [[SRC]]
+; VI: v_min_f32_e32 [[MIN:v[0-9]+]], 0x7f7fffff, [[RSQ]]
+; TODO: this constant should be folded:
+; VI: v_mov_b32_e32 [[K:v[0-9]+]], 0xff7fffff
+; VI: v_max_f32_e32 [[RESULT:v[0-9]+]], [[MIN]], [[K]]
+; VI: buffer_store_dword [[RESULT]]
+define void @rsq_clamp_f32(float addrspace(1)* %out, float %src) #0 {
+  %rsq_clamp = call float @llvm.amdgcn.rsq.clamp.f32(float %src)
+  store float %rsq_clamp, float addrspace(1)* %out
+  ret void
+}
+
+
+; FUNC-LABEL: {{^}}rsq_clamp_f64:
+; SI: v_rsq_clamp_f64_e32
+
+; VI: v_rsq_f64_e32 [[RSQ:v\[[0-9]+:[0-9]+\]]], s[2:3]
+; TODO: this constant should be folded:
+; VI: s_mov_b32 s[[ALLBITS:[0-9+]]], -1
+; VI: s_mov_b32 s[[HIGH1:[0-9+]]], 0x7fefffff
+; VI: s_mov_b32 s[[LOW1:[0-9+]]], s[[ALLBITS]]
+; VI: v_min_f64 v[0:1], [[RSQ]], s{{\[}}[[LOW1]]:[[HIGH1]]]
+; VI: s_mov_b32 s[[HIGH2:[0-9+]]], 0xffefffff
+; VI: s_mov_b32 s[[LOW2:[0-9+]]], s[[ALLBITS]]
+; VI: v_max_f64 v[0:1], v[0:1], s{{\[}}[[LOW2]]:[[HIGH2]]]
+define void @rsq_clamp_f64(double addrspace(1)* %out, double %src) #0 {
+  %rsq_clamp = call double @llvm.amdgcn.rsq.clamp.f64(double %src)
+  store double %rsq_clamp, double addrspace(1)* %out
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }

test/CodeGen/AMDGPU/llvm.amdgcn.rsq.clamped.ll

-; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
-; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=VI -check-prefix=FUNC %s
-
-declare double @llvm.amdgcn.rsq.clamped.f64(double) nounwind readnone
-
-; FUNC-LABEL: {{^}}rsq_clamped_f64:
-; SI: v_rsq_clamp_f64_e32
-
-; VI: v_rsq_f64_e32 [[RSQ:v\[[0-9]+:[0-9]+\]]], s[2:3]
-; TODO: this constant should be folded:
-; VI: s_mov_b32 s[[ALLBITS:[0-9+]]], -1
-; VI: s_mov_b32 s[[HIGH1:[0-9+]]], 0x7fefffff
-; VI: s_mov_b32 s[[LOW1:[0-9+]]], s[[ALLBITS]]
-; VI: v_min_f64 v[0:1], [[RSQ]], s{{\[}}[[LOW1]]:[[HIGH1]]]
-; VI: s_mov_b32 s[[HIGH2:[0-9+]]], 0xffefffff
-; VI: s_mov_b32 s[[LOW2:[0-9+]]], s[[ALLBITS]]
-; VI: v_max_f64 v[0:1], v[0:1], s{{\[}}[[LOW2]]:[[HIGH2]]]
-
-define void @rsq_clamped_f64(double addrspace(1)* %out, double %src) nounwind {
-  %rsq_clamped = call double @llvm.amdgcn.rsq.clamped.f64(double %src) nounwind readnone
-  store double %rsq_clamped, double addrspace(1)* %out, align 8
-  ret void
-}