[clang] 24c75a2 - [AMDGPU][Clang] Support for type inferring extended image builtins for AMDGPU (#164358)
via cfe-commits
cfe-commits at lists.llvm.org
Thu Oct 30 09:50:32 PDT 2025
Author: Rana Pratap Reddy
Date: 2025-10-30T22:20:28+05:30
New Revision: 24c75a21b8109908ed10d795d54c837b4621e941
URL: https://github.com/llvm/llvm-project/commit/24c75a21b8109908ed10d795d54c837b4621e941
DIFF: https://github.com/llvm/llvm-project/commit/24c75a21b8109908ed10d795d54c837b4621e941.diff
LOG: [AMDGPU][Clang] Support for type inferring extended image builtins for AMDGPU (#164358)
Introduces the builtins for extended image insts for amdgcn.
Added:
clang/test/CodeGen/builtins-extended-image.c
clang/test/SemaOpenCL/builtins-extended-image-param-gfx1100-err.cl
clang/test/SemaOpenCL/builtins-extended-image-param-gfx942-err.cl
Modified:
clang/include/clang/Basic/Builtins.def
clang/include/clang/Basic/BuiltinsAMDGPU.def
clang/lib/AST/ASTContext.cpp
clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp
clang/lib/Sema/SemaAMDGPU.cpp
Removed:
################################################################################
diff --git a/clang/include/clang/Basic/Builtins.def b/clang/include/clang/Basic/Builtins.def
index b856ad145824d..3a5b72e20afab 100644
--- a/clang/include/clang/Basic/Builtins.def
+++ b/clang/include/clang/Basic/Builtins.def
@@ -43,6 +43,7 @@
// SJ -> sigjmp_buf
// K -> ucontext_t
// p -> pid_t
+// e -> _Float16 for HIP/C++ and __fp16 for OpenCL
// . -> "...". This may only occur at the end of the function list.
//
// Types may be prefixed with the following modifiers:
diff --git a/clang/include/clang/Basic/BuiltinsAMDGPU.def b/clang/include/clang/Basic/BuiltinsAMDGPU.def
index f265d82efee75..36cb527a9c806 100644
--- a/clang/include/clang/Basic/BuiltinsAMDGPU.def
+++ b/clang/include/clang/Basic/BuiltinsAMDGPU.def
@@ -967,6 +967,47 @@ TARGET_BUILTIN(__builtin_amdgcn_image_sample_3d_v4f32_f32, "V4fifffQtV4ibii", "n
TARGET_BUILTIN(__builtin_amdgcn_image_sample_3d_v4f16_f32, "V4hifffQtV4ibii", "nc", "image-insts")
TARGET_BUILTIN(__builtin_amdgcn_image_sample_cube_v4f32_f32, "V4fifffQtV4ibii", "nc", "image-insts")
TARGET_BUILTIN(__builtin_amdgcn_image_sample_cube_v4f16_f32, "V4hifffQtV4ibii", "nc", "image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_1d_v4f32_f32, "V4fifQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_1d_v4f16_f32, "V4eifQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_1darray_v4f32_f32, "V4fiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_1darray_v4f16_f32, "V4eiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2d_f32_f32, "fiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2d_v4f32_f32, "V4fiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2d_v4f16_f32, "V4eiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2darray_f32_f32, "fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2darray_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_2darray_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_3d_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_3d_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_cube_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_lz_cube_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_1d_v4f32_f32, "V4fiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_1d_v4f16_f32, "V4eiffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_1darray_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_1darray_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2d_f32_f32, "fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2d_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2d_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2darray_f32_f32, "fiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2darray_v4f32_f32, "V4fiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_2darray_v4f16_f32, "V4eiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_3d_v4f32_f32, "V4fiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_3d_v4f16_f32, "V4eiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_cube_v4f32_f32, "V4fiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_l_cube_v4f16_f32, "V4eiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_1d_v4f32_f32, "V4fifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_1d_v4f16_f32, "V4eifffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_1darray_v4f32_f32, "V4fiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_1darray_v4f16_f32, "V4eiffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2d_f32_f32, "fiffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2d_v4f32_f32, "V4fiffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2d_v4f16_f32, "V4eiffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2darray_f32_f32, "fifffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2darray_v4f32_f32, "V4fifffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_2darray_v4f16_f32, "V4eifffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_3d_v4f32_f32, "V4fifffffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_sample_d_3d_v4f16_f32, "V4eifffffffffQtV4ibii", "nc", "extended-image-insts")
+TARGET_BUILTIN(__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32, "V4fiffQtV4ibii", "nc", "extended-image-insts")
#undef BUILTIN
#undef TARGET_BUILTIN
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index 687cd46773f43..2669f62456711 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -12403,6 +12403,11 @@ static QualType DecodeTypeFromStr(const char *&Str, const ASTContext &Context,
// Read the base type.
switch (*Str++) {
default: llvm_unreachable("Unknown builtin type letter!");
+ case 'e':
+ assert(HowLong == 0 && !Signed && !Unsigned &&
+ "Bad modifiers used with 'e'!");
+ Type = Context.getLangOpts().OpenCL ? Context.HalfTy : Context.Float16Ty;
+ break;
case 'x':
assert(HowLong == 0 && !Signed && !Unsigned &&
"Bad modifiers used with 'x'!");
diff --git a/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp b/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp
index f49a5af2c9587..9eab70955b6b9 100644
--- a/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp
+++ b/clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp
@@ -647,8 +647,8 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
case AMDGPU::BI__builtin_amdgcn_ballot_w64: {
llvm::Type *ResultType = ConvertType(E->getType());
llvm::Value *Src = EmitScalarExpr(E->getArg(0));
- Function *F = CGM.getIntrinsic(Intrinsic::amdgcn_ballot, { ResultType });
- return Builder.CreateCall(F, { Src });
+ Function *F = CGM.getIntrinsic(Intrinsic::amdgcn_ballot, {ResultType});
+ return Builder.CreateCall(F, {Src});
}
case AMDGPU::BI__builtin_amdgcn_inverse_ballot_w32:
case AMDGPU::BI__builtin_amdgcn_inverse_ballot_w64: {
@@ -1139,6 +1139,83 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
case AMDGPU::BI__builtin_amdgcn_image_sample_cube_v4f16_f32:
return emitAMDGCNImageOverloadedReturnType(
*this, E, Intrinsic::amdgcn_image_sample_cube, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_1d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_1d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_1d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_1d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_1d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_1d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_1d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_1d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_d_1d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_2d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_2d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_d_2d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_3d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_3d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_3d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_3d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_3d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_3d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_3d_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_3d_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_d_3d, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_cube_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_cube_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_cube, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_cube_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_cube_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_cube, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_1darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_1darray_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_1darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_1darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_1darray_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_1darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_1darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_1darray_v4f16_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_d_1darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_lz_2darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_l_2darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_v4f32_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_v4f16_f32:
+ case clang::AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_sample_d_2darray, false);
+ case clang::AMDGPU::BI__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32:
+ return emitAMDGCNImageOverloadedReturnType(
+ *this, E, Intrinsic::amdgcn_image_gather4_lz_2d, false);
case AMDGPU::BI__builtin_amdgcn_mfma_scale_f32_16x16x128_f8f6f4:
case AMDGPU::BI__builtin_amdgcn_mfma_scale_f32_32x32x64_f8f6f4: {
llvm::FixedVectorType *VT = FixedVectorType::get(Builder.getInt32Ty(), 8);
diff --git a/clang/lib/Sema/SemaAMDGPU.cpp b/clang/lib/Sema/SemaAMDGPU.cpp
index e32f4376a5ebf..139c4abc040df 100644
--- a/clang/lib/Sema/SemaAMDGPU.cpp
+++ b/clang/lib/Sema/SemaAMDGPU.cpp
@@ -153,7 +153,48 @@ bool SemaAMDGPU::CheckAMDGCNBuiltinFunctionCall(unsigned BuiltinID,
case AMDGPU::BI__builtin_amdgcn_image_sample_3d_v4f32_f32:
case AMDGPU::BI__builtin_amdgcn_image_sample_3d_v4f16_f32:
case AMDGPU::BI__builtin_amdgcn_image_sample_cube_v4f32_f32:
- case AMDGPU::BI__builtin_amdgcn_image_sample_cube_v4f16_f32: {
+ case AMDGPU::BI__builtin_amdgcn_image_sample_cube_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_1d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_1d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_1darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_1darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_2darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_3d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_3d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_cube_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_lz_cube_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_1d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_1d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_1darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_1darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_2darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_3d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_3d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_cube_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_l_cube_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_1d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_1d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_1darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_1darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_2darray_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_3d_v4f32_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_sample_d_3d_v4f16_f32:
+ case AMDGPU::BI__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32: {
StringRef FeatureList(
getASTContext().BuiltinInfo.getRequiredFeatures(BuiltinID));
if (!Builtin::evaluateRequiredTargetFeatures(FeatureList,
diff --git a/clang/test/CodeGen/builtins-extended-image.c b/clang/test/CodeGen/builtins-extended-image.c
new file mode 100644
index 0000000000000..0dbf81dabd77b
--- /dev/null
+++ b/clang/test/CodeGen/builtins-extended-image.c
@@ -0,0 +1,1528 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// RUN: %clang_cc1 -triple amdgcn-- -target-cpu gfx1100 -target-feature +extended-image-insts %s -emit-llvm -o - | FileCheck %s
+
+typedef int int4 __attribute__((ext_vector_type(4)));
+typedef float float4 __attribute__((ext_vector_type(4)));
+typedef _Float16 half4 __attribute__((ext_vector_type(4)));
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_gather4_lz_2d_v4f32_f32_r(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.gather4.lz.2d.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_r(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(1, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_gather4_lz_2d_v4f32_f32_g(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.gather4.lz.2d.v4f32.f32.v8i32.v4i32(i32 2, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_g(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(2, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_gather4_lz_2d_v4f32_f32_b(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.gather4.lz.2d.v4f32.f32.v8i32.v4i32(i32 4, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_b(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(4, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_gather4_lz_2d_v4f32_f32_a(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.gather4.lz.2d.v4f32.f32.v8i32.v4i32(i32 8, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_a(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(8, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_1d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32
+// CHECK-NEXT: [[TMP2:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP3:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP2]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP3]]
+//
+float4 test_amdgcn_image_sample_lz_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f32_f32(100, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_1d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_sample_l_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_d_1d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_d_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f32_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_2d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_sample_lz_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_2d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f32.v8i32.v4i32(i32 10, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_l_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f32_f32(10, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_d_2d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP6]], align 32
+// CHECK-NEXT: [[TMP7:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP8:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP7]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP8]]
+//
+float4 test_amdgcn_image_sample_d_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f32_f32(100, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_3d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.3d.v4f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_lz_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f32_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_3d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.3d.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP6]]
+//
+float4 test_amdgcn_image_sample_l_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_d_3d_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP8:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP9]], align 32
+// CHECK-NEXT: [[TMP10:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP11:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], float [[TMP6]], float [[TMP7]], float [[TMP8]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP10]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP11]]
+//
+float4 test_amdgcn_image_sample_d_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_cube_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.cube.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_lz_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_cube_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.cube.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP6]]
+//
+float4 test_amdgcn_image_sample_l_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_1darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.1darray.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP4]]
+//
+float4 test_amdgcn_image_sample_lz_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f32_f32(1, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_1darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.1darray.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_l_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_d_1darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.d.1darray.v4f32.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP6]]
+//
+float4 test_amdgcn_image_sample_d_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_lz_2darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.lz.2darray.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP5]]
+//
+float4 test_amdgcn_image_sample_lz_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_l_2darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.l.2darray.v4f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP6]]
+//
+float4 test_amdgcn_image_sample_l_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x float> @test_amdgcn_image_sample_d_2darray_v4f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP7]], align 32
+// CHECK-NEXT: [[TMP8:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP9:%.*]] = call <4 x float> @llvm.amdgcn.image.sample.d.2darray.v4f32.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], float [[TMP6]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP8]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x float> [[TMP9]]
+//
+float4 test_amdgcn_image_sample_d_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_1d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP1]], align 32
+// CHECK-NEXT: [[TMP2:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP3:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.1d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP2]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP3]]
+//
+half4 test_amdgcn_image_sample_lz_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f16_f32(100, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_1d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.1d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP4]]
+//
+half4 test_amdgcn_image_sample_l_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f16_f32(100, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_d_1d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.d.1d.v4f16.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_d_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_2d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.2d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP4]]
+//
+half4 test_amdgcn_image_sample_lz_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f16_f32(100, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_2d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.2d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_l_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_d_2d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP6]], align 32
+// CHECK-NEXT: [[TMP7:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP8:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.d.2d.v4f16.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP7]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP8]]
+//
+half4 test_amdgcn_image_sample_d_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_3d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.3d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_lz_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_3d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.3d.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP6]]
+//
+half4 test_amdgcn_image_sample_l_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_d_3d_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP8:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP9]], align 32
+// CHECK-NEXT: [[TMP10:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP11:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.d.3d.v4f16.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], float [[TMP6]], float [[TMP7]], float [[TMP8]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP10]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP11]]
+//
+half4 test_amdgcn_image_sample_d_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_cube_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.cube.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_lz_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_cube_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.cube.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP6]]
+//
+half4 test_amdgcn_image_sample_l_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_1darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.1darray.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP4]]
+//
+half4 test_amdgcn_image_sample_lz_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f16_f32(100, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_1darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.1darray.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_l_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_d_1darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.d.1darray.v4f16.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP6]]
+//
+half4 test_amdgcn_image_sample_d_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_lz_2darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.lz.2darray.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP5]]
+//
+half4 test_amdgcn_image_sample_lz_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_l_2darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.l.2darray.v4f16.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP6]]
+//
+half4 test_amdgcn_image_sample_l_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local <4 x half> @test_amdgcn_image_sample_d_2darray_v4f16_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca <4 x half>, align 8, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP7]], align 32
+// CHECK-NEXT: [[TMP8:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP9:%.*]] = call <4 x half> @llvm.amdgcn.image.sample.d.2darray.v4f16.f32.f32.v8i32.v4i32(i32 100, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], float [[TMP6]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP8]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret <4 x half> [[TMP9]]
+//
+half4 test_amdgcn_image_sample_d_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_lz_2d_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP2]], align 32
+// CHECK-NEXT: [[TMP3:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP4:%.*]] = call float @llvm.amdgcn.image.sample.lz.2d.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP3]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP4]]
+//
+float test_amdgcn_image_sample_lz_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_f32_f32(1, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_l_2d_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call float @llvm.amdgcn.image.sample.l.2d.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP5]]
+//
+float test_amdgcn_image_sample_l_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_d_2d_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP6]], align 32
+// CHECK-NEXT: [[TMP7:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP8:%.*]] = call float @llvm.amdgcn.image.sample.d.2d.f32.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP7]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP8]]
+//
+float test_amdgcn_image_sample_d_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_f32_f32(1, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_lz_2darray_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP3]], align 32
+// CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP5:%.*]] = call float @llvm.amdgcn.image.sample.lz.2darray.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP4]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP5]]
+//
+float test_amdgcn_image_sample_lz_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_l_2darray_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP4]], align 32
+// CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP6:%.*]] = call float @llvm.amdgcn.image.sample.l.2darray.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP5]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP6]]
+//
+float test_amdgcn_image_sample_l_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
+
+// CHECK-LABEL: define dso_local float @test_amdgcn_image_sample_d_2darray_f32_f32(
+// CHECK-SAME: <4 x float> noundef [[V4F32:%.*]], float noundef [[F32:%.*]], i32 noundef [[I32:%.*]], ptr [[TEX:%.*]], <4 x i32> noundef [[VEC4I32:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[RETVAL:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[V4F32_ADDR:%.*]] = alloca <4 x float>, align 16, addrspace(5)
+// CHECK-NEXT: [[F32_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[I32_ADDR:%.*]] = alloca i32, align 4, addrspace(5)
+// CHECK-NEXT: [[TEX_ADDR:%.*]] = alloca ptr, align 32, addrspace(5)
+// CHECK-NEXT: [[VEC4I32_ADDR:%.*]] = alloca <4 x i32>, align 16, addrspace(5)
+// CHECK-NEXT: [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr
+// CHECK-NEXT: [[V4F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[V4F32_ADDR]] to ptr
+// CHECK-NEXT: [[F32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[F32_ADDR]] to ptr
+// CHECK-NEXT: [[I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[I32_ADDR]] to ptr
+// CHECK-NEXT: [[TEX_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[TEX_ADDR]] to ptr
+// CHECK-NEXT: [[VEC4I32_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VEC4I32_ADDR]] to ptr
+// CHECK-NEXT: store <4 x float> [[V4F32]], ptr [[V4F32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: store float [[F32]], ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store i32 [[I32]], ptr [[I32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: store ptr [[TEX]], ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: store <4 x i32> [[VEC4I32]], ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[F32_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TEX_ADDR_ASCAST]], align 32
+// CHECK-NEXT: [[TEX_RSRC_VAL:%.*]] = load <8 x i32>, ptr [[TMP7]], align 32
+// CHECK-NEXT: [[TMP8:%.*]] = load <4 x i32>, ptr [[VEC4I32_ADDR_ASCAST]], align 16
+// CHECK-NEXT: [[TMP9:%.*]] = call float @llvm.amdgcn.image.sample.d.2darray.f32.f32.f32.v8i32.v4i32(i32 1, float [[TMP0]], float [[TMP1]], float [[TMP2]], float [[TMP3]], float [[TMP4]], float [[TMP5]], float [[TMP6]], <8 x i32> [[TEX_RSRC_VAL]], <4 x i32> [[TMP8]], i1 false, i32 120, i32 110)
+// CHECK-NEXT: ret float [[TMP9]]
+//
+float test_amdgcn_image_sample_d_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 120, 110);
+}
diff --git a/clang/test/SemaOpenCL/builtins-extended-image-param-gfx1100-err.cl b/clang/test/SemaOpenCL/builtins-extended-image-param-gfx1100-err.cl
new file mode 100644
index 0000000000000..47dbdd4e51782
--- /dev/null
+++ b/clang/test/SemaOpenCL/builtins-extended-image-param-gfx1100-err.cl
@@ -0,0 +1,227 @@
+// RUN: %clang_cc1 -triple amdgcn-- -target-cpu gfx1100 -target-feature +extended-image-insts -S -verify=expected -o - %s
+// REQUIRES: amdgpu-registered-target
+
+#pragma OPENCL EXTENSION cl_khr_fp16 : enable
+
+typedef int int4 __attribute__((ext_vector_type(4)));
+typedef float float4 __attribute__((ext_vector_type(4)));
+typedef half half4 __attribute__((ext_vector_type(4)));
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_r(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(1, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_g(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(2, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_b(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(4, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_a(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(8, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_gather4_lz_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_lz_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f32_f32(i32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_1d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_1d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_d_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f32_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_1d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_lz_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f32_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, 103); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_d_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f32_f32(i32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2d_v4f32_f32' must be a constant integer}}
+}
+float4 test_amdgcn_image_sample_lz_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f32_f32(i32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_3d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_3d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_d_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_3d_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_lz_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_cube_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_cube_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_lz_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f32_f32(1, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_1darray_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_1darray_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_d_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_1darray_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_lz_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2darray_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_l_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2darray_v4f32_f32' must be a constant integer}}
+}
+
+float4 test_amdgcn_image_sample_d_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2darray_v4f32_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f16_f32(23, f32, tex, vec4i32, 0, i32, 11); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_1d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f16_f32(i32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_1d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_d_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f16_f32(i32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_1d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f16_f32(100, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_d_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_3d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_3d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_d_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_3d_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_cube_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f16_f32(i32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_cube_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f16_f32(i32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_1darray_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f16_f32(i32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_1darray_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_d_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_1darray_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_lz_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2darray_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_l_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2darray_v4f16_f32' must be a constant integer}}
+}
+
+half4 test_amdgcn_image_sample_d_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2darray_v4f16_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_lz_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_f32_f32(1, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2d_f32_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_l_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2d_f32_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_d_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_f32_f32(1, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2d_f32_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_lz_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_lz_2darray_f32_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_l_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_l_2darray_f32_f32' must be a constant integer}}
+}
+
+float test_amdgcn_image_sample_d_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, f32, i32); //expected-error{{argument to '__builtin_amdgcn_image_sample_d_2darray_f32_f32' must be a constant integer}}
+}
diff --git a/clang/test/SemaOpenCL/builtins-extended-image-param-gfx942-err.cl b/clang/test/SemaOpenCL/builtins-extended-image-param-gfx942-err.cl
new file mode 100644
index 0000000000000..e60f8c70dc7c4
--- /dev/null
+++ b/clang/test/SemaOpenCL/builtins-extended-image-param-gfx942-err.cl
@@ -0,0 +1,227 @@
+// RUN: %clang_cc1 -triple amdgcn-- -target-cpu gfx942 -verify=GFX94 -S -o - %s
+// REQUIRES: amdgpu-registered-target
+
+#pragma OPENCL EXTENSION cl_khr_fp16 : enable
+
+typedef int int4 __attribute__((ext_vector_type(4)));
+typedef float float4 __attribute__((ext_vector_type(4)));
+typedef half half4 __attribute__((ext_vector_type(4)));
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_r(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(1, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_gather4_lz_2d_v4f32_f32_r' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_g(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(2, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_gather4_lz_2d_v4f32_f32_g' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_b(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(4, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_gather4_lz_2d_v4f32_f32_b' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_gather4_lz_2d_v4f32_f32_a(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_gather4_lz_2d_v4f32_f32(8, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_gather4_lz_2d_v4f32_f32_a' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_lz_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f32_f32(105, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_1d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_1d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_d_1d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f32_f32(100, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_1d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_lz_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f32_f32(100, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f32_f32(10, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_d_2d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f32_f32(105, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2d_v4f32_f32' needs target feature extended-image-insts}}
+}
+float4 test_amdgcn_image_sample_lz_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f32_f32(105, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_3d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_3d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_d_3d_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_3d_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_lz_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_cube_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_cube_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_cube_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_lz_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f32_f32(1, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_1darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_1darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_d_1darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_1darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_lz_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_l_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+float4 test_amdgcn_image_sample_d_2darray_v4f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2darray_v4f32_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1d_v4f16_f32(105, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_1d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1d_v4f16_f32(105, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_1d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_d_1d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1d_v4f16_f32(105, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_1d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_v4f16_f32(100, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_d_2d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_3d_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_3d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_3d_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_3d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_d_3d_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_3d_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_3d_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_cube_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_cube_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_cube_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_cube_v4f16_f32(105, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_cube_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_1darray_v4f16_f32(105, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_1darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_1darray_v4f16_f32(105, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_1darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_d_1darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_1darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_1darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_lz_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_v4f16_f32(100, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_l_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_v4f16_f32(100, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+half4 test_amdgcn_image_sample_d_2darray_v4f16_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_v4f16_f32(100, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2darray_v4f16_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_lz_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2d_f32_f32(1, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2d_f32_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_l_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2d_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2d_f32_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_d_2d_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2d_f32_f32(1, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2d_f32_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_lz_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_lz_2darray_f32_f32(1, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_lz_2darray_f32_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_l_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_l_2darray_f32_f32(1, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_l_2darray_f32_f32' needs target feature extended-image-insts}}
+}
+
+float test_amdgcn_image_sample_d_2darray_f32_f32(float4 v4f32, float f32, int i32, __amdgpu_texture_t tex, int4 vec4i32) {
+
+ return __builtin_amdgcn_image_sample_d_2darray_f32_f32(1, f32, f32, f32, f32, f32, f32, f32, tex, vec4i32, 0, 101, 121); //GFX94-error{{'test_amdgcn_image_sample_d_2darray_f32_f32' needs target feature extended-image-insts}}
+}
More information about the cfe-commits
mailing list