AMDGPU: Add v_prng_b32 instruction for gfx950

Pravin Jagtap · arsenm · commit d613fe273083 · 2024-11-14T17:38:43.000-08:00
Rand num instruction for stochastic rounding.
diff --git a/clang/include/clang/Basic/BuiltinsAMDGPU.def b/clang/include/clang/Basic/BuiltinsAMDGPU.def
@@ -522,5 +522,7 @@ TARGET_BUILTIN(__builtin_amdgcn_swmmac_f32_16x16x32_fp8_bf8_w64, "V4fiV2iV4fs",
 TARGET_BUILTIN(__builtin_amdgcn_swmmac_f32_16x16x32_bf8_fp8_w64, "V4fiV2iV4fs", "nc", "gfx12-insts,wavefrontsize64")
 TARGET_BUILTIN(__builtin_amdgcn_swmmac_f32_16x16x32_bf8_bf8_w64, "V4fiV2iV4fs", "nc", "gfx12-insts,wavefrontsize64")
 
+TARGET_BUILTIN(__builtin_amdgcn_prng_b32, "UiUi", "nc", "prng-inst")
+
 #undef BUILTIN
 #undef TARGET_BUILTIN
diff --git a/clang/test/CodeGenOpenCL/amdgpu-features.cl b/clang/test/CodeGenOpenCL/amdgpu-features.cl
@@ -89,7 +89,7 @@
 // GFX941: "target-features"="+16-bit-insts,+atomic-buffer-global-pk-add-f16-insts,+atomic-ds-pk-add-16-insts,+atomic-fadd-rtn-insts,+atomic-flat-pk-add-16-insts,+atomic-global-pk-add-bf16-inst,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+fp8-conversion-insts,+fp8-insts,+gfx8-insts,+gfx9-insts,+gfx90a-insts,+gfx940-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,+xf32-insts"
 // GFX942: "target-features"="+16-bit-insts,+atomic-buffer-global-pk-add-f16-insts,+atomic-ds-pk-add-16-insts,+atomic-fadd-rtn-insts,+atomic-flat-pk-add-16-insts,+atomic-global-pk-add-bf16-inst,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+fp8-conversion-insts,+fp8-insts,+gfx8-insts,+gfx9-insts,+gfx90a-insts,+gfx940-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64,+xf32-insts"
 // GFX9_4_Generic: "target-features"="+16-bit-insts,+atomic-buffer-global-pk-add-f16-insts,+atomic-ds-pk-add-16-insts,+atomic-fadd-rtn-insts,+atomic-flat-pk-add-16-insts,+atomic-global-pk-add-bf16-inst,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+gfx90a-insts,+gfx940-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64"
-// GFX950: "target-features"="+16-bit-insts,+atomic-buffer-global-pk-add-f16-insts,+atomic-ds-pk-add-16-insts,+atomic-fadd-rtn-insts,+atomic-flat-pk-add-16-insts,+atomic-global-pk-add-bf16-inst,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+fp8-conversion-insts,+fp8-insts,+gfx8-insts,+gfx9-insts,+gfx90a-insts,+gfx940-insts,+gfx950-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64"
+// GFX950: "target-features"="+16-bit-insts,+atomic-buffer-global-pk-add-f16-insts,+atomic-ds-pk-add-16-insts,+atomic-fadd-rtn-insts,+atomic-flat-pk-add-16-insts,+atomic-global-pk-add-bf16-inst,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+fp8-conversion-insts,+fp8-insts,+gfx8-insts,+gfx9-insts,+gfx90a-insts,+gfx940-insts,+gfx950-insts,+mai-insts,+prng-inst,+s-memrealtime,+s-memtime-inst,+wavefrontsize64"
 // GFX1010: "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dpp,+gfx10-insts,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize32"
 // GFX1011: "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+gfx10-insts,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize32"
 // GFX1012: "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,+gfx10-insts,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize32"
diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx950-err.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx950-err.cl
@@ -0,0 +1,16 @@
+// RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx906 -emit-llvm \
+// RUN:   -verify -o - %s
+// RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx90a -emit-llvm \
+// RUN:   -verify -o - %s
+// RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx940 -emit-llvm \
+// RUN:   -verify -o - %s
+// RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx1200 -emit-llvm \
+// RUN:   -verify -o - %s
+
+
+// REQUIRES: amdgpu-registered-target
+
+typedef unsigned int uint;
+void test_prng_b32(global uint* out, uint a) {
+  *out = __builtin_amdgcn_prng_b32(a); // expected-error{{'__builtin_amdgcn_prng_b32' needs target feature prng-inst}}
+}
diff --git a/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx950.cl b/clang/test/CodeGenOpenCL/builtins-amdgcn-gfx950.cl
@@ -0,0 +1,21 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// RUN: %clang_cc1 -cl-std=CL1.2 -O0 -triple amdgcn-unknown-unknown -target-cpu gfx950 -emit-llvm -o - %s | FileCheck %s
+// REQUIRES: amdgpu-registered-target
+
+typedef unsigned int uint;
+
+// CHECK-LABEL: @test_prng_b32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[OUT_ADDR:%.*]] = alloca ptr addrspace(1), align 8, addrspace(5)
+// CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT:    store ptr addrspace(1) [[OUT:%.*]], ptr addrspace(5) [[OUT_ADDR]], align 8
+// CHECK-NEXT:    store i32 [[A:%.*]], ptr addrspace(5) [[A_ADDR]], align 4
+// CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr addrspace(5) [[A_ADDR]], align 4
+// CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.amdgcn.prng.b32(i32 [[TMP0]])
+// CHECK-NEXT:    [[TMP2:%.*]] = load ptr addrspace(1), ptr addrspace(5) [[OUT_ADDR]], align 8
+// CHECK-NEXT:    store i32 [[TMP1]], ptr addrspace(1) [[TMP2]], align 4
+// CHECK-NEXT:    ret void
+//
+void test_prng_b32(global uint* out, uint a) {
+  *out = __builtin_amdgcn_prng_b32(a);
+}
diff --git a/llvm/include/llvm/IR/IntrinsicsAMDGPU.td b/llvm/include/llvm/IR/IntrinsicsAMDGPU.td
@@ -594,6 +594,10 @@ def int_amdgcn_ds_ordered_swap : AMDGPUDSOrderedIntrinsic;
 def int_amdgcn_ds_append : AMDGPUDSAppendConsumedIntrinsic;
 def int_amdgcn_ds_consume : AMDGPUDSAppendConsumedIntrinsic;
 
+def int_amdgcn_prng_b32 : DefaultAttrsIntrinsic<
+  [llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]
+>, ClangBuiltin<"__builtin_amdgcn_prng_b32">;
+
 } // TargetPrefix = "amdgcn"
 
 // New-style image intrinsics
diff --git a/llvm/lib/Target/AMDGPU/AMDGPU.td b/llvm/lib/Target/AMDGPU/AMDGPU.td
@@ -978,6 +978,12 @@ def FeatureVmemWriteVgprInOrder : SubtargetFeature<"vmem-write-vgpr-in-order",
   "VMEM instructions of the same type write VGPR results in order"
 >;
 
+def FeaturePrngInst : SubtargetFeature<"prng-inst",
+  "HasPrngInst",
+  "true",
+  "Has v_prng_b32 instruction"
+>;
+
 //===------------------------------------------------------------===//
 // Subtarget Features (options and debugging)
 //===------------------------------------------------------------===//
@@ -1495,7 +1501,8 @@ def FeatureISAVersion9_5_Common : FeatureSet<
    FeatureFP8ConversionInsts,
    FeatureCvtFP8VOP1Bug,
    FeatureGFX950Insts,
-   FeatureAddressableLocalMemorySize163840
+   FeatureAddressableLocalMemorySize163840,
+   FeaturePrngInst
    ])>;
 
 def FeatureISAVersion9_4_0 : FeatureSet<
@@ -2344,6 +2351,9 @@ def HasSALUFloatInsts : Predicate<"Subtarget->hasSALUFloatInsts()">,
 def HasPseudoScalarTrans : Predicate<"Subtarget->hasPseudoScalarTrans()">,
   AssemblerPredicate<(all_of FeaturePseudoScalarTrans)>;
 
+def HasPrngInst : Predicate<"Subtarget->hasPrngInst()">,
+  AssemblerPredicate<(all_of FeaturePrngInst)>;
+
 def HasGDS : Predicate<"Subtarget->hasGDS()">;
 
 def HasGWS : Predicate<"Subtarget->hasGWS()">;
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp b/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
@@ -1253,6 +1253,12 @@ GCNTTIImpl::instCombineIntrinsic(InstCombiner &IC, IntrinsicInst &II) const {
 
     break;
   }
+  case Intrinsic::amdgcn_prng_b32: {
+    auto *Src = II.getArgOperand(0);
+    if (isa<UndefValue>(Src)) {
+      return IC.replaceInstUsesWith(II, Src);
+    }
+  }
   }
   if (const AMDGPU::ImageDimIntrinsicInfo *ImageDimIntr =
             AMDGPU::getImageDimIntrinsicInfo(II.getIntrinsicID())) {
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
@@ -4515,6 +4515,7 @@ AMDGPURegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
     case Intrinsic::amdgcn_cvt_pk_u8_f32:
     case Intrinsic::amdgcn_alignbyte:
     case Intrinsic::amdgcn_perm:
+    case Intrinsic::amdgcn_prng_b32:
     case Intrinsic::amdgcn_fdot2:
     case Intrinsic::amdgcn_sdot2:
     case Intrinsic::amdgcn_udot2:
diff --git a/llvm/lib/Target/AMDGPU/GCNSubtarget.h b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
@@ -220,7 +220,7 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
   bool HasSALUFloatInsts = false;
   bool HasPseudoScalarTrans = false;
   bool HasRestrictedSOffset = false;
-
+  bool HasPrngInst = false;
   bool HasVcmpxPermlaneHazard = false;
   bool HasVMEMtoScalarWriteHazard = false;
   bool HasSMEMtoVectorWriteHazard = false;
@@ -1321,6 +1321,8 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
   /// instruction.
   unsigned maxHardClauseLength() const { return MaxHardClauseLength; }
 
+  bool hasPrngInst() const { return HasPrngInst; }
+
   /// Return the maximum number of waves per SIMD for kernels using \p SGPRs
   /// SGPRs
   unsigned getOccupancyWithNumSGPRs(unsigned SGPRs) const;
diff --git a/llvm/lib/Target/AMDGPU/VOP1Instructions.td b/llvm/lib/Target/AMDGPU/VOP1Instructions.td
@@ -761,6 +761,9 @@ let SubtargetPredicate = isGFX11Plus in {
   defm V_CVT_U32_U16    : VOP1Inst_t16<"v_cvt_u32_u16", VOP_I32_I16>;
 } // End SubtargetPredicate = isGFX11Plus
 
+let SubtargetPredicate = HasPrngInst in
+defm V_PRNG_B32 : VOP1Inst <"v_prng_b32", VOP_I32_I32, int_amdgcn_prng_b32>;
+
 foreach vt = Reg32Types.types in {
   def : GCNPat<(int_amdgcn_permlane64 (vt VRegSrc_32:$src0)),
         (vt (V_PERMLANE64_B32 (vt VRegSrc_32:$src0)))
@@ -1516,6 +1519,8 @@ defm V_CVT_F32_BF8       : VOP1_Real_NoDstSel_SDWA_gfx9<0x55>;
 defm V_CVT_PK_F32_FP8    : VOP1_Real_NoDstSel_SDWA_gfx9<0x56>;
 defm V_CVT_PK_F32_BF8    : VOP1_Real_NoDstSel_SDWA_gfx9<0x57>;
 
+defm V_PRNG_B32            : VOP1_Real_gfx9 <0x58>;
+
 class MovDPP8Pattern<Predicate Pred, Instruction Inst, ValueType vt> : GCNPat <
   (vt (int_amdgcn_mov_dpp8 vt:$src, timm:$dpp8)),
   (Inst VGPR_32:$src, VGPR_32:$src, (as_i32timm $dpp8), (i32 DPP8Mode.FI_0))> {
diff --git a/llvm/lib/TargetParser/TargetParser.cpp b/llvm/lib/TargetParser/TargetParser.cpp
@@ -470,6 +470,7 @@ void AMDGPU::fillAMDGPUFeatureMap(StringRef GPU, const Triple &T,
       Features["gws"] = true;
       break;
     case GK_GFX950:
+      Features["prng-inst"] = true;
       Features["gfx950-insts"] = true;
       [[fallthrough]];
     case GK_GFX942:
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.prng.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.prng.ll
@@ -0,0 +1,32 @@
+; RUN: llc -global-isel=0 -march=amdgcn -mcpu=gfx950 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN %s
+; RUN: llc -global-isel=1 -march=amdgcn -mcpu=gfx950 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+declare i32 @llvm.amdgcn.prng.b32(i32) #0
+
+; GCN-LABEL: {{^}}prng_b32:
+; GCN: v_prng_b32_e32 {{v[0-9]+}}, {{s[0-9]+}}
+define amdgpu_kernel void @prng_b32(ptr addrspace(1) %out, i32 %src) #1 {
+  %prng = call i32 @llvm.amdgcn.prng.b32(i32 %src) #0
+  store i32 %prng, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}prng_b32_constant_4
+; GCN: v_prng_b32_e32 {{v[0-9]+}}, 4
+define amdgpu_kernel void @prng_b32_constant_4(ptr addrspace(1) %out) #1 {
+  %prng = call i32 @llvm.amdgcn.prng.b32(i32 4) #0
+  store i32 %prng, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}prng_b32_constant_100
+; GCN: v_prng_b32_e32 {{v[0-9]+}}, 0x64
+define amdgpu_kernel void @prng_b32_constant_100(ptr addrspace(1) %out) #1 {
+  %prng = call i32 @llvm.amdgcn.prng.b32(i32 100) #0
+  store i32 %prng, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+
+attributes #0 = { nounwind readnone }
+attributes #1 = { nounwind }
diff --git a/llvm/test/MC/AMDGPU/gfx950_asm_vop1.s b/llvm/test/MC/AMDGPU/gfx950_asm_vop1.s
@@ -0,0 +1,57 @@
+// RUN: llvm-mc -arch=amdgcn -mcpu=gfx950 -show-encoding %s | FileCheck --check-prefix=GFX950 %s
+
+v_prng_b32 v5, v1
+// GFX950: v_prng_b32_e32 v5, v1                   ; encoding: [0x01,0xb1,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, v255
+// GFX950: v_prng_b32_e32 v5, v255                 ; encoding: [0xff,0xb1,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, s1
+// GFX950: v_prng_b32_e32 v5, s1                   ; encoding: [0x01,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, s101
+// GFX950: v_prng_b32_e32 v5, s101                 ; encoding: [0x65,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, vcc_lo
+// GFX950: v_prng_b32_e32 v5, vcc_lo               ; encoding: [0x6a,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, vcc_hi
+// GFX950: v_prng_b32_e32 v5, vcc_hi               ; encoding: [0x6b,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, ttmp15
+// GFX950: v_prng_b32_e32 v5, ttmp15               ; encoding: [0x7b,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, m0
+// GFX950: v_prng_b32_e32 v5, m0                   ; encoding: [0x7c,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, exec_lo
+// GFX950: v_prng_b32_e32 v5, exec_lo              ; encoding: [0x7e,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, exec_hi
+// GFX950: v_prng_b32_e32 v5, exec_hi              ; encoding: [0x7f,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, -1
+// GFX950: v_prng_b32_e32 v5, -1                   ; encoding: [0xc1,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, 0.5
+// GFX950: v_prng_b32_e32 v5, 0.5                  ; encoding: [0xf0,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v5, src_scc
+// GFX950: v_prng_b32_e32 v5, src_scc              ; encoding: [0xfd,0xb0,0x0a,0x7e]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+v_prng_b32 v255, 0xaf123456
+// GFX950: v_prng_b32_e32 v255, 0xaf123456         ; encoding: [0xff,0xb0,0xfe,0x7f,0x56,0x34,0x12,0xaf]
+// GFX940-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
diff --git a/llvm/test/MC/AMDGPU/gfx950_asm_vop1_dpp16.s b/llvm/test/MC/AMDGPU/gfx950_asm_vop1_dpp16.s
@@ -0,0 +1,31 @@
+// RUN: llvm-mc -arch=amdgcn -mcpu=gfx950 -show-encoding %s | FileCheck --check-prefixes=GFX950 %s
+
+v_prng_b32 v5, v1 quad_perm:[3,2,1,0]
+// GFX950: v_prng_b32_dpp v5, v1 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ;   encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1b,0x00,0xff]
+
+v_prng_b32 v5, v1 quad_perm:[0,1,2,3]
+// GFX950: v_prng_b32_dpp v5, v1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ;   encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0xe4,0x00,0xff]
+
+v_prng_b32 v5, v1 row_mirror
+// GFX950: v_prng_b32_dpp v5, v1 row_mirror row_mask:0xf bank_mask:0xf ;            encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x40,0x01,0xff]
+
+v_prng_b32 v5, v1 row_half_mirror
+// GFX950: v_prng_b32_dpp v5, v1 row_half_mirror row_mask:0xf bank_mask:0xf ;       encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x41,0x01,0xff]
+
+v_prng_b32 v5, v1 row_shl:1
+// GFX950: v_prng_b32_dpp v5, v1 row_shl:1 row_mask:0xf bank_mask:0xf ;             encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x01,0x01,0xff]
+
+v_prng_b32 v5, v1 row_shl:15
+// GFX950: v_prng_b32_dpp v5, v1 row_shl:15 row_mask:0xf bank_mask:0xf ;            encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x0f,0x01,0xff]
+
+v_prng_b32 v5, v1 row_shr:1
+// GFX950: v_prng_b32_dpp v5, v1 row_shr:1 row_mask:0xf bank_mask:0xf ;             encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x11,0x01,0xff]
+
+v_prng_b32 v5, v1 row_shr:15
+// GFX950: v_prng_b32_dpp v5, v1 row_shr:15 row_mask:0xf bank_mask:0xf ;            encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x1f,0x01,0xff]
+
+v_prng_b32 v5, v1 row_ror:1
+// GFX950: v_prng_b32_dpp v5, v1 row_ror:1 row_mask:0xf bank_mask:0xf ;             encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x21,0x01,0xff]
+
+v_prng_b32 v5, v1 row_ror:15
+// GFX950: v_prng_b32_dpp v5, v1 row_ror:15 row_mask:0xf bank_mask:0xf ;            encoding: [0xfa,0xb0,0x0a,0x7e,0x01,0x2f,0x01,0xff]
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx950_dasm_vop1.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx950_dasm_vop1.txt
@@ -0,0 +1,43 @@
+# RUN: llvm-mc -arch=amdgcn -mcpu=gfx950 -disassemble -show-encoding < %s | FileCheck -check-prefix=GFX950 %s
+
+# GFX950: v_prng_b32_e32 v5, v1                   ; encoding: [0x01,0xb1,0x0a,0x7e]
+0x01,0xb1,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, v255                 ; encoding: [0xff,0xb1,0x0a,0x7e]
+0xff,0xb1,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, s1                   ; encoding: [0x01,0xb0,0x0a,0x7e]
+0x01,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, s101                 ; encoding: [0x65,0xb0,0x0a,0x7e]
+0x65,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, vcc_lo               ; encoding: [0x6a,0xb0,0x0a,0x7e]
+0x6a,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, vcc_hi               ; encoding: [0x6b,0xb0,0x0a,0x7e]
+0x6b,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, ttmp15               ; encoding: [0x7b,0xb0,0x0a,0x7e]
+0x7b,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, m0                   ; encoding: [0x7c,0xb0,0x0a,0x7e]
+0x7c,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, exec_lo              ; encoding: [0x7e,0xb0,0x0a,0x7e]
+0x7e,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, exec_hi              ; encoding: [0x7f,0xb0,0x0a,0x7e]
+0x7f,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, -1                   ; encoding: [0xc1,0xb0,0x0a,0x7e]
+0xc1,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, 0.5                  ; encoding: [0xf0,0xb0,0x0a,0x7e]
+0xf0,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v5, src_scc              ; encoding: [0xfd,0xb0,0x0a,0x7e]
+0xfd,0xb0,0x0a,0x7e
+
+# GFX950: v_prng_b32_e32 v255, 0xaf123456         ; encoding: [0xff,0xb0,0xfe,0x7f,0x56,0x34,0x12,0xaf]
+0xff,0xb0,0xfe,0x7f,0x56,0x34,0x12,0xaf
diff --git a/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-intrinsics.ll b/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-intrinsics.ll
@@ -6547,3 +6547,21 @@ define half @test_constant_fold_exp2_f16_neg_denorm() {
   %val = call half @llvm.amdgcn.exp2.f16(half 0xH83ff)
   ret half %val
 }
+
+; --------------------------------------------------------------------
+; llvm.amdgcn.prng
+; --------------------------------------------------------------------
+declare i32 @llvm.amdgcn.prng.b32(i32)
+define i32 @prng_undef_i32() {
+; CHECK-LABEL: @prng_undef_i32(
+; CHECK-NEXT:    ret i32 undef
+  %prng = call i32 @llvm.amdgcn.prng.b32(i32 undef)
+  ret i32 %prng
+}
+
+define i32 @prng_poison_i32() {
+; CHECK-LABEL: @prng_poison_i32(
+; CHECK-NEXT:    ret i32 poison
+  %prng = call i32 @llvm.amdgcn.prng.b32(i32 poison)
+  ret i32 %prng
+}

Original file line number	Diff line number	Diff line change
`@@ -1253,6 +1253,12 @@ GCNTTIImpl::instCombineIntrinsic(InstCombiner &IC, IntrinsicInst &II) const {`
`1253`	`1253`
`1254`	`1254`	`break;`
`1255`	`1255`	`}`
	`1256`	`+ case Intrinsic::amdgcn_prng_b32: {`
	`1257`	`+ auto *Src = II.getArgOperand(0);`
	`1258`	`+ if (isa<UndefValue>(Src)) {`
	`1259`	`+ return IC.replaceInstUsesWith(II, Src);`
	`1260`	`+ }`
	`1261`	`+ }`
`1256`	`1262`	`}`
`1257`	`1263`	`if (const AMDGPU::ImageDimIntrinsicInfo *ImageDimIntr =`
`1258`	`1264`	`AMDGPU::getImageDimIntrinsicInfo(II.getIntrinsicID())) {`