[SPIR-V] Add clang builtin for subgroup shuffles #174655

jhuber6 · 2026-01-06T21:42:17Z

Summary:
This is an attempt to begin filling out some missing pieces to allow
more generic compute code to use SPIR-V flavored builtins. This should
provide the basic shuffle operation. The next most important one is the
ballot, but I don't think we have an IR intrinsic for that yet.

I don't know SPIR-V very well so let me know if this is the proper
function with the proper semantic checks.

llvmbot · 2026-01-06T21:42:51Z

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-clang-codegen

Author: Joseph Huber (jhuber6)

Changes

Summary:
This is an attempt to begin filling out some missing pieces to allow
more generic compute code to use SPIR-V flavored builtins. This should
provide the basic shuffle operation. The next most important one is the
ballot, but I don't think we have an IR intrinsic for that yet.

I don't know SPIR-V very well so let me know if this is the proper
function with the proper semantic checks.

Full diff: https://github.com/llvm/llvm-project/pull/174655.diff

5 Files Affected:

(modified) clang/include/clang/Basic/BuiltinsSPIRVCommon.td (+2)
(modified) clang/lib/CodeGen/TargetBuiltins/SPIR.cpp (+9)
(modified) clang/lib/Sema/SemaSPIRV.cpp (+26)
(modified) clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c (+8)
(modified) clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c (+6)

diff --git a/clang/include/clang/Basic/BuiltinsSPIRVCommon.td b/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
index d2ef6f99a0502..e31758b889d3d 100644
--- a/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
+++ b/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
@@ -21,3 +21,5 @@ def subgroup_local_invocation_id : SPIRVBuiltin<"uint32_t()", [NoThrow, Const]>;
 def distance : SPIRVBuiltin<"void(...)", [NoThrow, Const]>;
 def length : SPIRVBuiltin<"void(...)", [NoThrow, Const]>;
 def smoothstep : SPIRVBuiltin<"void(...)", [NoThrow, Const, CustomTypeChecking]>;
+
+def subgroup_shuffle : SPIRVBuiltin<"void(...)", [NoThrow, Const, CustomTypeChecking]>;
diff --git a/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp b/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
index 9b0ca3eb0035a..1ea23e8c2195f 100644
--- a/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
+++ b/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
@@ -109,6 +109,15 @@ Value *CodeGenFunction::EmitSPIRVBuiltinExpr(unsigned BuiltinID,
     Call->addRetAttr(llvm::Attribute::AttrKind::NoUndef);
     return Call;
   }
+  case SPIRV::BI__builtin_spirv_subgroup_shuffle: {
+    Value *X = EmitScalarExpr(E->getArg(0));
+    Value *Y = EmitScalarExpr(E->getArg(1));
+    assert(E->getArg(1)->getType()->hasIntegerRepresentation());
+    return Builder.CreateIntrinsic(
+        /*ReturnType=*/getTypes().ConvertType(E->getArg(0)->getType()),
+        Intrinsic::spv_wave_readlane, ArrayRef<Value *>{X, Y}, nullptr,
+        "spv.shuffle");
+  }
   case SPIRV::BI__builtin_spirv_num_workgroups:
     return Builder.CreateIntrinsic(
         /*ReturnType=*/getTypes().ConvertType(E->getType()),
diff --git a/clang/lib/Sema/SemaSPIRV.cpp b/clang/lib/Sema/SemaSPIRV.cpp
index fa30e149c209a..5f3dd71f28c67 100644
--- a/clang/lib/Sema/SemaSPIRV.cpp
+++ b/clang/lib/Sema/SemaSPIRV.cpp
@@ -380,6 +380,32 @@ bool SemaSPIRV::CheckSPIRVBuiltinFunctionCall(const TargetInfo &TI,
     TheCall->setType(RetTy);
     break;
   }
+  case SPIRV::BI__builtin_spirv_subgroup_shuffle: {
+    if (SemaRef.checkArgCount(TheCall, 2))
+      return true;
+
+    ExprResult A = TheCall->getArg(0);
+    QualType ArgTyA = A.get()->getType();
+    if (!ArgTyA->isIntegerType() && !ArgTyA->isFloatingType()) {
+      SemaRef.Diag(A.get()->getBeginLoc(), diag::err_builtin_invalid_arg_type)
+          << /* ordinal */ 1 << /* scalar */ 1 << /* no int */ 0
+          << /* no fp */ 0 << ArgTyA;
+      return true;
+    }
+
+    ExprResult B = TheCall->getArg(1);
+    QualType ArgTyB = B.get()->getType();
+    if (!ArgTyB->isIntegerType()) {
+      SemaRef.Diag(B.get()->getBeginLoc(), diag::err_builtin_invalid_arg_type)
+          << /* ordinal */ 2 << /* scalar */ 1 << /* int */ 1 << /* no fp */ 0
+          << ArgTyB;
+      return true;
+    }
+
+    QualType RetTy = ArgTyA;
+    TheCall->setType(RetTy);
+    break;
+  }
   }
   return false;
 }
diff --git a/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c b/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
index bff850b3622b9..7142dbbc2be73 100644
--- a/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
+++ b/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
@@ -104,3 +104,11 @@
 [[clang::sycl_external]] unsigned int test_subgroup_local_invocation_id() {
     return __builtin_spirv_subgroup_local_invocation_id();
 }
+
+// CHECK: @{{.*}}test_subgroup_shuffle{{.*}}(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    tail call float @llvm.spv.wave.readlane.f32(float %f, i32 %i)
+//
+[[clang::sycl_external]] float test_subgroup_shuffle(float f, int i) {
+    return __builtin_spirv_subgroup_shuffle(f, i);
+}
diff --git a/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c b/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
index 0d98a552bb1b9..43b2de4e31485 100644
--- a/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
+++ b/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
@@ -75,3 +75,9 @@ void test_subgroup_local_invocation_id() {
   __builtin_spirv_subgroup_local_invocation_id();
   __builtin_spirv_subgroup_local_invocation_id(0); // expected-error{{too many arguments to function call, expected 0, have 1}}
 }
+
+void test_subgroup_shuffle(int i, float f, int *p) {
+  __builtin_spirv_subgroup_shuffle(f, i);
+  __builtin_spirv_subgroup_shuffle(p, i); // expected-error{{1st argument must be a scalar type}}
+  __builtin_spirv_subgroup_shuffle(i, f); // expected-error{{2nd argument must be a scalar integer}}
+}

llvmbot · 2026-01-06T21:42:52Z

@llvm/pr-subscribers-backend-spir-v

Author: Joseph Huber (jhuber6)

Changes

Summary:
This is an attempt to begin filling out some missing pieces to allow
more generic compute code to use SPIR-V flavored builtins. This should
provide the basic shuffle operation. The next most important one is the
ballot, but I don't think we have an IR intrinsic for that yet.

I don't know SPIR-V very well so let me know if this is the proper
function with the proper semantic checks.

Full diff: https://github.com/llvm/llvm-project/pull/174655.diff

5 Files Affected:

(modified) clang/include/clang/Basic/BuiltinsSPIRVCommon.td (+2)
(modified) clang/lib/CodeGen/TargetBuiltins/SPIR.cpp (+9)
(modified) clang/lib/Sema/SemaSPIRV.cpp (+26)
(modified) clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c (+8)
(modified) clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c (+6)

diff --git a/clang/include/clang/Basic/BuiltinsSPIRVCommon.td b/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
index d2ef6f99a0502..e31758b889d3d 100644
--- a/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
+++ b/clang/include/clang/Basic/BuiltinsSPIRVCommon.td
@@ -21,3 +21,5 @@ def subgroup_local_invocation_id : SPIRVBuiltin<"uint32_t()", [NoThrow, Const]>;
 def distance : SPIRVBuiltin<"void(...)", [NoThrow, Const]>;
 def length : SPIRVBuiltin<"void(...)", [NoThrow, Const]>;
 def smoothstep : SPIRVBuiltin<"void(...)", [NoThrow, Const, CustomTypeChecking]>;
+
+def subgroup_shuffle : SPIRVBuiltin<"void(...)", [NoThrow, Const, CustomTypeChecking]>;
diff --git a/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp b/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
index 9b0ca3eb0035a..1ea23e8c2195f 100644
--- a/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
+++ b/clang/lib/CodeGen/TargetBuiltins/SPIR.cpp
@@ -109,6 +109,15 @@ Value *CodeGenFunction::EmitSPIRVBuiltinExpr(unsigned BuiltinID,
     Call->addRetAttr(llvm::Attribute::AttrKind::NoUndef);
     return Call;
   }
+  case SPIRV::BI__builtin_spirv_subgroup_shuffle: {
+    Value *X = EmitScalarExpr(E->getArg(0));
+    Value *Y = EmitScalarExpr(E->getArg(1));
+    assert(E->getArg(1)->getType()->hasIntegerRepresentation());
+    return Builder.CreateIntrinsic(
+        /*ReturnType=*/getTypes().ConvertType(E->getArg(0)->getType()),
+        Intrinsic::spv_wave_readlane, ArrayRef<Value *>{X, Y}, nullptr,
+        "spv.shuffle");
+  }
   case SPIRV::BI__builtin_spirv_num_workgroups:
     return Builder.CreateIntrinsic(
         /*ReturnType=*/getTypes().ConvertType(E->getType()),
diff --git a/clang/lib/Sema/SemaSPIRV.cpp b/clang/lib/Sema/SemaSPIRV.cpp
index fa30e149c209a..5f3dd71f28c67 100644
--- a/clang/lib/Sema/SemaSPIRV.cpp
+++ b/clang/lib/Sema/SemaSPIRV.cpp
@@ -380,6 +380,32 @@ bool SemaSPIRV::CheckSPIRVBuiltinFunctionCall(const TargetInfo &TI,
     TheCall->setType(RetTy);
     break;
   }
+  case SPIRV::BI__builtin_spirv_subgroup_shuffle: {
+    if (SemaRef.checkArgCount(TheCall, 2))
+      return true;
+
+    ExprResult A = TheCall->getArg(0);
+    QualType ArgTyA = A.get()->getType();
+    if (!ArgTyA->isIntegerType() && !ArgTyA->isFloatingType()) {
+      SemaRef.Diag(A.get()->getBeginLoc(), diag::err_builtin_invalid_arg_type)
+          << /* ordinal */ 1 << /* scalar */ 1 << /* no int */ 0
+          << /* no fp */ 0 << ArgTyA;
+      return true;
+    }
+
+    ExprResult B = TheCall->getArg(1);
+    QualType ArgTyB = B.get()->getType();
+    if (!ArgTyB->isIntegerType()) {
+      SemaRef.Diag(B.get()->getBeginLoc(), diag::err_builtin_invalid_arg_type)
+          << /* ordinal */ 2 << /* scalar */ 1 << /* int */ 1 << /* no fp */ 0
+          << ArgTyB;
+      return true;
+    }
+
+    QualType RetTy = ArgTyA;
+    TheCall->setType(RetTy);
+    break;
+  }
   }
   return false;
 }
diff --git a/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c b/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
index bff850b3622b9..7142dbbc2be73 100644
--- a/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
+++ b/clang/test/CodeGenSPIRV/Builtins/ids_and_ranges.c
@@ -104,3 +104,11 @@
 [[clang::sycl_external]] unsigned int test_subgroup_local_invocation_id() {
     return __builtin_spirv_subgroup_local_invocation_id();
 }
+
+// CHECK: @{{.*}}test_subgroup_shuffle{{.*}}(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    tail call float @llvm.spv.wave.readlane.f32(float %f, i32 %i)
+//
+[[clang::sycl_external]] float test_subgroup_shuffle(float f, int i) {
+    return __builtin_spirv_subgroup_shuffle(f, i);
+}
diff --git a/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c b/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
index 0d98a552bb1b9..43b2de4e31485 100644
--- a/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
+++ b/clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c
@@ -75,3 +75,9 @@ void test_subgroup_local_invocation_id() {
   __builtin_spirv_subgroup_local_invocation_id();
   __builtin_spirv_subgroup_local_invocation_id(0); // expected-error{{too many arguments to function call, expected 0, have 1}}
 }
+
+void test_subgroup_shuffle(int i, float f, int *p) {
+  __builtin_spirv_subgroup_shuffle(f, i);
+  __builtin_spirv_subgroup_shuffle(p, i); // expected-error{{1st argument must be a scalar type}}
+  __builtin_spirv_subgroup_shuffle(i, f); // expected-error{{2nd argument must be a scalar integer}}
+}

jhuber6 · 2026-01-06T23:49:56Z

Apparently __builtin_hlsl_wave_read_lane_at exists, but we probably want the generic version. I'm not familiar with HLSL, is there a reason we define these as HLSL builtins first?

farzonl · 2026-01-06T23:53:42Z

Apparently __builtin_hlsl_wave_read_lane_at exists, but we probably want the generic version. I'm not familiar with HLSL, is there a reason we define these as HLSL builtins first?

The team could look into this case, but the answer is usually because we use builtins as alias for codegen that needs to support multiple targets. A secondary more minor reason is it gives us tighter control of sema rules.

jhuber6 · 2026-01-07T00:02:22Z

Apparently __builtin_hlsl_wave_read_lane_at exists, but we probably want the generic version. I'm not familiar with HLSL, is there a reason we define these as HLSL builtins first?

The team could look into this case, but the answer is usually because we use builtins as alias for codegen that needs to support multiple targets. A secondary more minor reason is it gives us tighter control of sema rules.

Alright, I'm mostly just interested in a basis set of SPIR-V builtins that will let me port some existing code. I think the only missing one if this one works is a ballot function. Everything else complicated can be made from that. I see an intrinsic for int_spv_wave_active_countbits which returns the popcount of the ballot's result. Is there a reason we don't have one for a regular ballot? Seeing as the subgroup size is unknown to SPIR-V (as far as I know) I can see this being difficult to define. The result is a vector, but what size would that be? Sorry I don't know who contributes to the SPIR-V stuff, mostly just trying to unify compute handling between the three "real" GPU targets.

clang/lib/CodeGen/TargetBuiltins/SPIR.cpp

clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

github-actions · 2026-01-08T02:52:36Z

✅ With the latest revision this PR passed the C/C++ code formatter.

Summary: llvm#174862 and llvm#174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these.

jhuber6 · 2026-01-08T16:24:59Z

Random question, why are the intrinsics called spv.wave when the spec seems to use subgroup? I wasn't sure if I should keep that convention or change it in #174862

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and #174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

farzonl · 2026-01-08T18:03:59Z

Random question, why are the intrinsics called spv.wave when the spec seems to use subgroup? I wasn't sure if I should keep that convention or change it in #174862

subgroup (SPIRV), wave (DX), warp (Nvidia), wavefront (AMD). Are all the same thing. We picked wave because its the only one that seems to be used as an api-level name and not a harware unit. It was also the only one that shares a name with a harware unit.

More importantly the intrinsic toggling we are doing is very basic between DirectX and SPIRV so the intrinsic names need to match.

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm/llvm-project#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

farzonl · 2026-01-08T18:18:17Z

Random question, why are the intrinsics called spv.wave when the spec seems to use subgroup? I wasn't sure if I should keep that convention or change it in #174862

More importantly the intrinsic toggling we are doing is very basic between DirectX and SPIRV so the intrinsic names need to match.

It mostly became this way because when we started spirv in the frontend wasn't a thing yet.

It doesn't have to stay this way in clang/lib/CodeGen/CGHLSLRuntime.h there is a macro GENERATE_HLSL_INTRINSIC_FUNCTION we could make a GENERATE_HLSL_INTRINSIC_FUNCTION_DEFAULT that uses the same IntrinsicPostfix name for both spirv and directx but then for these cases where we want some intrinsics to be subgroup vs wave we have GENERATE_HLSL_INTRINSIC_FUNCTION a spvPostfix adn dxpostfix name.

github-actions · 2026-01-08T18:30:39Z

🐧 Linux x64 Test Results

112563 tests passed
4604 tests skipped

✅ The build succeeded and all tests passed.

github-actions · 2026-01-08T18:30:39Z

🪟 Windows x64 Test Results

53365 tests passed
2220 tests skipped

✅ The build succeeded and all tests passed.

Summary: This is an attempt to begin filling out some missing pieces to allow more generic compute code to use SPIR-V flavored builtins. This should provide the basic shuffle operation. The next most important one is the ballot, but I don't think we have an IR intrinsic for that yet. I don't know SPIR-V very well so let me know if this is the proper function with the proper semantic checks. Sema test nit Update sema checking for int argument

Summary: llvm#174862 and llvm#174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these.

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

Summary: This is an attempt to begin filling out some missing pieces to allow more generic compute code to use SPIR-V flavored builtins. This should provide the basic shuffle operation. The next most important one is the ballot, but I don't think we have an IR intrinsic for that yet. I don't know SPIR-V very well so let me know if this is the proper function with the proper semantic checks.

Summary: #174862 and #174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these.

…910) Summary: llvm/llvm-project#174862 and llvm/llvm-project#174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these.

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm/llvm-project#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need. (cherry picked from commit f8b68c7)

Summary: llvm/llvm-project#174862 and llvm/llvm-project#174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these. (cherry picked from commit 5c43243)

Summary: This patch adds an LLVM intrinsic and lowering for a subgroup ballot and a corresponding clang builtin. This uses the already present support but provides in a way accessible to other targets. With this and llvm#174655 we should be able to most of the basic functions, like shuffling, active masks, and reductions. More work will be needed for canonicalizing / exposing the SPIR-V functions, but these are the fundamental builtins I need.

Summary: This is an attempt to begin filling out some missing pieces to allow more generic compute code to use SPIR-V flavored builtins. This should provide the basic shuffle operation. The next most important one is the ballot, but I don't think we have an IR intrinsic for that yet. I don't know SPIR-V very well so let me know if this is the proper function with the proper semantic checks.

Summary: llvm#174862 and llvm#174655 provided the intrinsics required to get the fundamental operations working for these. This patch sets up the basic support (as far as I know). This should be the first step towards allowing SPIR-V to build things like the LLVM libc and the OpenMP Device Runtime Library. The implementations here are intentionally inefficient, such as not using the dedicated SPIR-V opcode for read firstlane. This is just to start and hopefully start testing things later. Would appreciate someone more familiar with the backend double-checking these.

jhuber6 requested review from AlexVlx, Keenuts, MrSidims, Naghasan, arsenm, farzonl and michalpaszkowski January 6, 2026 21:42

llvmbot added clang Clang issues not falling into any other category clang:frontend Language frontend issues, e.g. anything involving "Sema" clang:codegen IR generation bugs: mangling, exceptions, etc. backend:SPIR-V labels Jan 6, 2026

arsenm reviewed Jan 7, 2026

View reviewed changes

clang/lib/CodeGen/TargetBuiltins/SPIR.cpp Outdated Show resolved Hide resolved

clang/test/SemaSPIRV/BuiltIns/ids_and_ranges.c Outdated Show resolved Hide resolved

jhuber6 force-pushed the SPIRVShuffle branch from 3351f9c to fc35ecc Compare January 7, 2026 14:50

jhuber6 mentioned this pull request Jan 7, 2026

[SPIR-V] Add builtin/intrinsic for subgroup ballot #174862

Merged

jhuber6 force-pushed the SPIRVShuffle branch from 391b43b to 7feb315 Compare January 8, 2026 02:53

jhuber6 requested a review from sarnex January 8, 2026 04:11

jhuber6 mentioned this pull request Jan 8, 2026

[SPIR-V] Initial support for SPIR-V in gpuintrin.h #174910

Merged

jhuber6 force-pushed the SPIRVShuffle branch from 7feb315 to f590d17 Compare January 8, 2026 18:02

farzonl approved these changes Jan 8, 2026

View reviewed changes

jhuber6 force-pushed the SPIRVShuffle branch from f590d17 to 559c65d Compare January 8, 2026 18:43

jhuber6 merged commit c02da3d into llvm:main Jan 8, 2026
10 checks passed

farzonl mentioned this pull request Jan 23, 2026

[HLSL][DXIL][SPIRV] Added WaveActiveBitOr HLSL intrinsic #165156

Open

12 tasks

[SPIR-V] Add clang builtin for subgroup shuffles #174655

[SPIR-V] Add clang builtin for subgroup shuffles #174655

Uh oh!

Conversation

jhuber6 commented Jan 6, 2026

Uh oh!

llvmbot commented Jan 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

llvmbot commented Jan 6, 2026

Uh oh!

jhuber6 commented Jan 6, 2026

Uh oh!

farzonl commented Jan 6, 2026

Uh oh!

jhuber6 commented Jan 7, 2026

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Jan 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jhuber6 commented Jan 8, 2026

Uh oh!

farzonl commented Jan 8, 2026

Uh oh!

farzonl commented Jan 8, 2026

Uh oh!

github-actions bot commented Jan 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🐧 Linux x64 Test Results

Uh oh!

github-actions bot commented Jan 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🪟 Windows x64 Test Results

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

llvmbot commented Jan 6, 2026 •

edited

Loading

github-actions bot commented Jan 8, 2026 •

edited

Loading

github-actions bot commented Jan 8, 2026 •

edited

Loading

github-actions bot commented Jan 8, 2026 •

edited

Loading