shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/llvm/test/CodeGen/SPIRV/transcoding/sub_group_clustered_reduce.ll
Vyacheslav Levytskyy 54cc4141e4
[SPIR-V] Rework duplicate tracker and tracking of IR entities and types to improve compile-time performance (#130605) 
This PR is to thoroughly rework duplicate tracker implementation and
tracking of IR entities and types. These are legacy parts of the project
resulting in an extremely bloated intermediate representation and
computational delays due to inefficient data flow and structure choices.

Main results of the rework:

1) Improved compile-time performance. The reference binary LLVM IR used
to measure speed gains in
https://github.com/llvm/llvm-project/pull/120415 shows ~x5 speed up also
after this PR. The timing before this PR is ~42s and after this PR it's
~7.5s. In total this PR and the previous overhaul of the module analysis
in https://github.com/llvm/llvm-project/pull/120415 results in ~x25
speed improvement.
```
$ time llc -O0 -mtriple=spirv64v1.6-unknown-unknown _group_barrier_phi.bc -o 1 --filetype=obj

real    0m7.545s
user    0m6.685s
sys     0m0.859s
```

2) Less bloated intermediate representation of internal translation
steps. Elimination of `spv_track_constant` intrinsic usage for scalar
constants, rework of `spv_assign_name`, removal of the gMIR `GET_XXX`
pseudo code and a smaller number of generated `ASSIGN_TYPE` pseudo codes
substantially decrease volume of data generated during translation.

3) Simpler code and easier maintenance. The duplicate tracker
implementation is simplified, as well as other features.

4) Numerous fixes of issues and logical flaws in different passes. The
main achievement is rework of the duplicate tracker itself that had
never guaranteed a correct caching of LLVM IR entities, rarely and
randomly returning stale/incorrect records (like, remove an instruction
from gMIR but still refer to it). Other fixes comprise consistent
generation of OpConstantNull, assigning types to newly created
registers, creation of integer/bool types, and other minor fixes.

5) Numerous fixes of LIT tests: mainly CHECK-DAG to properly reflect
SPIR-V spec guarantees, `{{$}}` at the end of constants to avoid
matching of substrings, and XFAILS for `SPV_INTEL_long_composites` test
cases, because the feature is not completed in full yet and doesn't
generate a requested by the extension sequence of instructions.

6) New test cases are added.
2025-03-26 17:58:10 +01:00

745 lines
39 KiB
LLVM
Raw Blame History

;; #pragma OPENCL EXTENSION cl_khr_subgroup_clustered_reduce : enable
;; #pragma OPENCL EXTENSION cl_khr_fp16 : enable
;; #pragma OPENCL EXTENSION cl_khr_fp64 : enable
;;
;; kernel void testClusteredArithmeticChar(global char* dst)
;; {
;; char v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticUChar(global uchar* dst)
;; {
;; uchar v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticShort(global short* dst)
;; {
;; short v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticUShort(global ushort* dst)
;; {
;; ushort v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticInt(global int* dst)
;; {
;; int v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticUInt(global uint* dst)
;; {
;; uint v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticLong(global long* dst)
;; {
;; long v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticULong(global ulong* dst)
;; {
;; ulong v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticFloat(global float* dst)
;; {
;; float v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticHalf(global half* dst)
;; {
;; half v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredArithmeticDouble(global double* dst)
;; {
;; double v = 0;
;; dst[0] = sub_group_clustered_reduce_add(v, 2);
;; dst[1] = sub_group_clustered_reduce_mul(v, 2);
;; dst[2] = sub_group_clustered_reduce_min(v, 2);
;; dst[3] = sub_group_clustered_reduce_max(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseChar(global char* dst)
;; {
;; char v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseUChar(global uchar* dst)
;; {
;; uchar v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseShort(global short* dst)
;; {
;; short v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseUShort(global ushort* dst)
;; {
;; ushort v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseInt(global int* dst)
;; {
;; int v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseUInt(global uint* dst)
;; {
;; uint v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseLong(global long* dst)
;; {
;; long v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredBitwiseULong(global ulong* dst)
;; {
;; ulong v = 0;
;; dst[0] = sub_group_clustered_reduce_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_xor(v, 2);
;; }
;;
;; kernel void testClusteredLogical(global int* dst)
;; {
;; int v = 0;
;; dst[0] = sub_group_clustered_reduce_logical_and(v, 2);
;; dst[1] = sub_group_clustered_reduce_logical_or(v, 2);
;; dst[2] = sub_group_clustered_reduce_logical_xor(v, 2);
;; }
; RUN: llc -O0 -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s --check-prefix=CHECK-SPIRV
; CHECK-SPIRV-DAG: OpCapability GroupNonUniformClustered
; CHECK-SPIRV-DAG: %[[#bool:]] = OpTypeBool
; CHECK-SPIRV-DAG: %[[#char:]] = OpTypeInt 8 0
; CHECK-SPIRV-DAG: %[[#short:]] = OpTypeInt 16 0
; CHECK-SPIRV-DAG: %[[#int:]] = OpTypeInt 32 0
; CHECK-SPIRV-DAG: %[[#long:]] = OpTypeInt 64 0
; CHECK-SPIRV-DAG: %[[#half:]] = OpTypeFloat 16
; CHECK-SPIRV-DAG: %[[#float:]] = OpTypeFloat 32
; CHECK-SPIRV-DAG: %[[#double:]] = OpTypeFloat 64
; CHECK-SPIRV-DAG: %[[#false:]] = OpConstantFalse %[[#bool]]
; CHECK-SPIRV-DAG: %[[#ScopeSubgroup:]] = OpConstant %[[#int]] 3{{$}}
; CHECK-SPIRV-DAG: %[[#char_0:]] = OpConstantNull %[[#char]]
; CHECK-SPIRV-DAG: %[[#short_0:]] = OpConstantNull %[[#short]]
; CHECK-SPIRV-DAG: %[[#int_0:]] = OpConstantNull %[[#int]]
; CHECK-SPIRV-DAG: %[[#int_2:]] = OpConstant %[[#int]] 2{{$}}
; CHECK-SPIRV-DAG: %[[#long_0:]] = OpConstantNull %[[#long]]
; CHECK-SPIRV-DAG: %[[#half_0:]] = OpConstantNull %[[#half]]
; CHECK-SPIRV-DAG: %[[#float_0:]] = OpConstantNull %[[#float]]
; CHECK-SPIRV-DAG: %[[#double_0:]] = OpConstantNull %[[#double]]
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMin %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMax %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticChar(i8 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_addcj(i8 signext 0, i32 2)
store i8 %2, i8 addrspace(1)* %0, align 1
%3 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_mulcj(i8 signext 0, i32 2)
%4 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 1
store i8 %3, i8 addrspace(1)* %4, align 1
%5 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_mincj(i8 signext 0, i32 2)
%6 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 2
store i8 %5, i8 addrspace(1)* %6, align 1
%7 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_maxcj(i8 signext 0, i32 2)
%8 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 3
store i8 %7, i8 addrspace(1)* %8, align 1
ret void
}
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_addcj(i8 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_mulcj(i8 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_mincj(i8 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_maxcj(i8 signext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMin %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMax %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticUChar(i8 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_addhj(i8 zeroext 0, i32 2)
store i8 %2, i8 addrspace(1)* %0, align 1
%3 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_mulhj(i8 zeroext 0, i32 2)
%4 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 1
store i8 %3, i8 addrspace(1)* %4, align 1
%5 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_minhj(i8 zeroext 0, i32 2)
%6 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 2
store i8 %5, i8 addrspace(1)* %6, align 1
%7 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_maxhj(i8 zeroext 0, i32 2)
%8 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 3
store i8 %7, i8 addrspace(1)* %8, align 1
ret void
}
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_addhj(i8 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_mulhj(i8 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_minhj(i8 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_maxhj(i8 zeroext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMin %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMax %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticShort(i16 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_addsj(i16 signext 0, i32 2)
store i16 %2, i16 addrspace(1)* %0, align 2
%3 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_mulsj(i16 signext 0, i32 2)
%4 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 1
store i16 %3, i16 addrspace(1)* %4, align 2
%5 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_minsj(i16 signext 0, i32 2)
%6 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 2
store i16 %5, i16 addrspace(1)* %6, align 2
%7 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_maxsj(i16 signext 0, i32 2)
%8 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 3
store i16 %7, i16 addrspace(1)* %8, align 2
ret void
}
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_addsj(i16 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_mulsj(i16 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_minsj(i16 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_maxsj(i16 signext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMin %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMax %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticUShort(i16 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_addtj(i16 zeroext 0, i32 2)
store i16 %2, i16 addrspace(1)* %0, align 2
%3 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_multj(i16 zeroext 0, i32 2)
%4 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 1
store i16 %3, i16 addrspace(1)* %4, align 2
%5 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_mintj(i16 zeroext 0, i32 2)
%6 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 2
store i16 %5, i16 addrspace(1)* %6, align 2
%7 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_maxtj(i16 zeroext 0, i32 2)
%8 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 3
store i16 %7, i16 addrspace(1)* %8, align 2
ret void
}
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_addtj(i16 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_multj(i16 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_mintj(i16 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_maxtj(i16 zeroext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMin %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMax %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticInt(i32 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_addij(i32 0, i32 2)
store i32 %2, i32 addrspace(1)* %0, align 4
%3 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_mulij(i32 0, i32 2)
%4 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 1
store i32 %3, i32 addrspace(1)* %4, align 4
%5 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_minij(i32 0, i32 2)
%6 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 2
store i32 %5, i32 addrspace(1)* %6, align 4
%7 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_maxij(i32 0, i32 2)
%8 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 3
store i32 %7, i32 addrspace(1)* %8, align 4
ret void
}
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_addij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_mulij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_minij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_maxij(i32, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMin %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMax %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticUInt(i32 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_addjj(i32 0, i32 2)
store i32 %2, i32 addrspace(1)* %0, align 4
%3 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_muljj(i32 0, i32 2)
%4 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 1
store i32 %3, i32 addrspace(1)* %4, align 4
%5 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_minjj(i32 0, i32 2)
%6 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 2
store i32 %5, i32 addrspace(1)* %6, align 4
%7 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_maxjj(i32 0, i32 2)
%8 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 3
store i32 %7, i32 addrspace(1)* %8, align 4
ret void
}
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_addjj(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_muljj(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_minjj(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_maxjj(i32, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMin %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformSMax %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticLong(i64 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_addlj(i64 0, i32 2)
store i64 %2, i64 addrspace(1)* %0, align 8
%3 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_mullj(i64 0, i32 2)
%4 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 1
store i64 %3, i64 addrspace(1)* %4, align 8
%5 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_minlj(i64 0, i32 2)
%6 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 2
store i64 %5, i64 addrspace(1)* %6, align 8
%7 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_maxlj(i64 0, i32 2)
%8 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 3
store i64 %7, i64 addrspace(1)* %8, align 8
ret void
}
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_addlj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_mullj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_minlj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_maxlj(i64, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIAdd %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformIMul %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMin %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformUMax %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticULong(i64 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_addmj(i64 0, i32 2)
store i64 %2, i64 addrspace(1)* %0, align 8
%3 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_mulmj(i64 0, i32 2)
%4 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 1
store i64 %3, i64 addrspace(1)* %4, align 8
%5 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_minmj(i64 0, i32 2)
%6 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 2
store i64 %5, i64 addrspace(1)* %6, align 8
%7 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_maxmj(i64 0, i32 2)
%8 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 3
store i64 %7, i64 addrspace(1)* %8, align 8
ret void
}
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_addmj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_mulmj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_minmj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_maxmj(i64, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFAdd %[[#float]] %[[#ScopeSubgroup]] ClusteredReduce %[[#float_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMul %[[#float]] %[[#ScopeSubgroup]] ClusteredReduce %[[#float_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMin %[[#float]] %[[#ScopeSubgroup]] ClusteredReduce %[[#float_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMax %[[#float]] %[[#ScopeSubgroup]] ClusteredReduce %[[#float_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticFloat(float addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func float @_Z30sub_group_clustered_reduce_addfj(float 0.000000e+00, i32 2)
store float %2, float addrspace(1)* %0, align 4
%3 = tail call spir_func float @_Z30sub_group_clustered_reduce_mulfj(float 0.000000e+00, i32 2)
%4 = getelementptr inbounds float, float addrspace(1)* %0, i64 1
store float %3, float addrspace(1)* %4, align 4
%5 = tail call spir_func float @_Z30sub_group_clustered_reduce_minfj(float 0.000000e+00, i32 2)
%6 = getelementptr inbounds float, float addrspace(1)* %0, i64 2
store float %5, float addrspace(1)* %6, align 4
%7 = tail call spir_func float @_Z30sub_group_clustered_reduce_maxfj(float 0.000000e+00, i32 2)
%8 = getelementptr inbounds float, float addrspace(1)* %0, i64 3
store float %7, float addrspace(1)* %8, align 4
ret void
}
declare dso_local spir_func float @_Z30sub_group_clustered_reduce_addfj(float, i32) local_unnamed_addr
declare dso_local spir_func float @_Z30sub_group_clustered_reduce_mulfj(float, i32) local_unnamed_addr
declare dso_local spir_func float @_Z30sub_group_clustered_reduce_minfj(float, i32) local_unnamed_addr
declare dso_local spir_func float @_Z30sub_group_clustered_reduce_maxfj(float, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFAdd %[[#half]] %[[#ScopeSubgroup]] ClusteredReduce %[[#half_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMul %[[#half]] %[[#ScopeSubgroup]] ClusteredReduce %[[#half_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMin %[[#half]] %[[#ScopeSubgroup]] ClusteredReduce %[[#half_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMax %[[#half]] %[[#ScopeSubgroup]] ClusteredReduce %[[#half_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticHalf(half addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func half @_Z30sub_group_clustered_reduce_addDhj(half 0xH0000, i32 2)
store half %2, half addrspace(1)* %0, align 2
%3 = tail call spir_func half @_Z30sub_group_clustered_reduce_mulDhj(half 0xH0000, i32 2)
%4 = getelementptr inbounds half, half addrspace(1)* %0, i64 1
store half %3, half addrspace(1)* %4, align 2
%5 = tail call spir_func half @_Z30sub_group_clustered_reduce_minDhj(half 0xH0000, i32 2)
%6 = getelementptr inbounds half, half addrspace(1)* %0, i64 2
store half %5, half addrspace(1)* %6, align 2
%7 = tail call spir_func half @_Z30sub_group_clustered_reduce_maxDhj(half 0xH0000, i32 2)
%8 = getelementptr inbounds half, half addrspace(1)* %0, i64 3
store half %7, half addrspace(1)* %8, align 2
ret void
}
declare dso_local spir_func half @_Z30sub_group_clustered_reduce_addDhj(half, i32) local_unnamed_addr
declare dso_local spir_func half @_Z30sub_group_clustered_reduce_mulDhj(half, i32) local_unnamed_addr
declare dso_local spir_func half @_Z30sub_group_clustered_reduce_minDhj(half, i32) local_unnamed_addr
declare dso_local spir_func half @_Z30sub_group_clustered_reduce_maxDhj(half, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFAdd %[[#double]] %[[#ScopeSubgroup]] ClusteredReduce %[[#double_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMul %[[#double]] %[[#ScopeSubgroup]] ClusteredReduce %[[#double_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMin %[[#double]] %[[#ScopeSubgroup]] ClusteredReduce %[[#double_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformFMax %[[#double]] %[[#ScopeSubgroup]] ClusteredReduce %[[#double_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredArithmeticDouble(double addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func double @_Z30sub_group_clustered_reduce_adddj(double 0.000000e+00, i32 2)
store double %2, double addrspace(1)* %0, align 8
%3 = tail call spir_func double @_Z30sub_group_clustered_reduce_muldj(double 0.000000e+00, i32 2)
%4 = getelementptr inbounds double, double addrspace(1)* %0, i64 1
store double %3, double addrspace(1)* %4, align 8
%5 = tail call spir_func double @_Z30sub_group_clustered_reduce_mindj(double 0.000000e+00, i32 2)
%6 = getelementptr inbounds double, double addrspace(1)* %0, i64 2
store double %5, double addrspace(1)* %6, align 8
%7 = tail call spir_func double @_Z30sub_group_clustered_reduce_maxdj(double 0.000000e+00, i32 2)
%8 = getelementptr inbounds double, double addrspace(1)* %0, i64 3
store double %7, double addrspace(1)* %8, align 8
ret void
}
declare dso_local spir_func double @_Z30sub_group_clustered_reduce_adddj(double, i32) local_unnamed_addr
declare dso_local spir_func double @_Z30sub_group_clustered_reduce_muldj(double, i32) local_unnamed_addr
declare dso_local spir_func double @_Z30sub_group_clustered_reduce_mindj(double, i32) local_unnamed_addr
declare dso_local spir_func double @_Z30sub_group_clustered_reduce_maxdj(double, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseChar(i8 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_andcj(i8 signext 0, i32 2)
store i8 %2, i8 addrspace(1)* %0, align 1
%3 = tail call spir_func signext i8 @_Z29sub_group_clustered_reduce_orcj(i8 signext 0, i32 2)
%4 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 1
store i8 %3, i8 addrspace(1)* %4, align 1
%5 = tail call spir_func signext i8 @_Z30sub_group_clustered_reduce_xorcj(i8 signext 0, i32 2)
%6 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 2
store i8 %5, i8 addrspace(1)* %6, align 1
ret void
}
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_andcj(i8 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i8 @_Z29sub_group_clustered_reduce_orcj(i8 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i8 @_Z30sub_group_clustered_reduce_xorcj(i8 signext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#char]] %[[#ScopeSubgroup]] ClusteredReduce %[[#char_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseUChar(i8 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_andhj(i8 zeroext 0, i32 2)
store i8 %2, i8 addrspace(1)* %0, align 1
%3 = tail call spir_func zeroext i8 @_Z29sub_group_clustered_reduce_orhj(i8 zeroext 0, i32 2)
%4 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 1
store i8 %3, i8 addrspace(1)* %4, align 1
%5 = tail call spir_func zeroext i8 @_Z30sub_group_clustered_reduce_xorhj(i8 zeroext 0, i32 2)
%6 = getelementptr inbounds i8, i8 addrspace(1)* %0, i64 2
store i8 %5, i8 addrspace(1)* %6, align 1
ret void
}
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_andhj(i8 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i8 @_Z29sub_group_clustered_reduce_orhj(i8 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i8 @_Z30sub_group_clustered_reduce_xorhj(i8 zeroext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseShort(i16 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_andsj(i16 signext 0, i32 2)
store i16 %2, i16 addrspace(1)* %0, align 2
%3 = tail call spir_func signext i16 @_Z29sub_group_clustered_reduce_orsj(i16 signext 0, i32 2)
%4 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 1
store i16 %3, i16 addrspace(1)* %4, align 2
%5 = tail call spir_func signext i16 @_Z30sub_group_clustered_reduce_xorsj(i16 signext 0, i32 2)
%6 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 2
store i16 %5, i16 addrspace(1)* %6, align 2
ret void
}
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_andsj(i16 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i16 @_Z29sub_group_clustered_reduce_orsj(i16 signext, i32) local_unnamed_addr
declare dso_local spir_func signext i16 @_Z30sub_group_clustered_reduce_xorsj(i16 signext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#short]] %[[#ScopeSubgroup]] ClusteredReduce %[[#short_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseUShort(i16 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_andtj(i16 zeroext 0, i32 2)
store i16 %2, i16 addrspace(1)* %0, align 2
%3 = tail call spir_func zeroext i16 @_Z29sub_group_clustered_reduce_ortj(i16 zeroext 0, i32 2)
%4 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 1
store i16 %3, i16 addrspace(1)* %4, align 2
%5 = tail call spir_func zeroext i16 @_Z30sub_group_clustered_reduce_xortj(i16 zeroext 0, i32 2)
%6 = getelementptr inbounds i16, i16 addrspace(1)* %0, i64 2
store i16 %5, i16 addrspace(1)* %6, align 2
ret void
}
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_andtj(i16 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i16 @_Z29sub_group_clustered_reduce_ortj(i16 zeroext, i32) local_unnamed_addr
declare dso_local spir_func zeroext i16 @_Z30sub_group_clustered_reduce_xortj(i16 zeroext, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseInt(i32 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_andij(i32 0, i32 2)
store i32 %2, i32 addrspace(1)* %0, align 4
%3 = tail call spir_func i32 @_Z29sub_group_clustered_reduce_orij(i32 0, i32 2)
%4 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 1
store i32 %3, i32 addrspace(1)* %4, align 4
%5 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_xorij(i32 0, i32 2)
%6 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 2
store i32 %5, i32 addrspace(1)* %6, align 4
ret void
}
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_andij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z29sub_group_clustered_reduce_orij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_xorij(i32, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#int]] %[[#ScopeSubgroup]] ClusteredReduce %[[#int_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseUInt(i32 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_andjj(i32 0, i32 2)
store i32 %2, i32 addrspace(1)* %0, align 4
%3 = tail call spir_func i32 @_Z29sub_group_clustered_reduce_orjj(i32 0, i32 2)
%4 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 1
store i32 %3, i32 addrspace(1)* %4, align 4
%5 = tail call spir_func i32 @_Z30sub_group_clustered_reduce_xorjj(i32 0, i32 2)
%6 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 2
store i32 %5, i32 addrspace(1)* %6, align 4
ret void
}
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_andjj(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z29sub_group_clustered_reduce_orjj(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z30sub_group_clustered_reduce_xorjj(i32, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseLong(i64 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_andlj(i64 0, i32 2)
store i64 %2, i64 addrspace(1)* %0, align 8
%3 = tail call spir_func i64 @_Z29sub_group_clustered_reduce_orlj(i64 0, i32 2)
%4 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 1
store i64 %3, i64 addrspace(1)* %4, align 8
%5 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_xorlj(i64 0, i32 2)
%6 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 2
store i64 %5, i64 addrspace(1)* %6, align 8
ret void
}
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_andlj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z29sub_group_clustered_reduce_orlj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_xorlj(i64, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseAnd %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseOr %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformBitwiseXor %[[#long]] %[[#ScopeSubgroup]] ClusteredReduce %[[#long_0]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredBitwiseULong(i64 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_andmj(i64 0, i32 2)
store i64 %2, i64 addrspace(1)* %0, align 8
%3 = tail call spir_func i64 @_Z29sub_group_clustered_reduce_ormj(i64 0, i32 2)
%4 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 1
store i64 %3, i64 addrspace(1)* %4, align 8
%5 = tail call spir_func i64 @_Z30sub_group_clustered_reduce_xormj(i64 0, i32 2)
%6 = getelementptr inbounds i64, i64 addrspace(1)* %0, i64 2
store i64 %5, i64 addrspace(1)* %6, align 8
ret void
}
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_andmj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z29sub_group_clustered_reduce_ormj(i64, i32) local_unnamed_addr
declare dso_local spir_func i64 @_Z30sub_group_clustered_reduce_xormj(i64, i32) local_unnamed_addr
; CHECK-SPIRV: OpFunction
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformLogicalAnd %[[#bool]] %[[#ScopeSubgroup]] ClusteredReduce %[[#false]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformLogicalOr %[[#bool]] %[[#ScopeSubgroup]] ClusteredReduce %[[#false]] %[[#int_2]]
; CHECK-SPIRV: %[[#]] = OpGroupNonUniformLogicalXor %[[#bool]] %[[#ScopeSubgroup]] ClusteredReduce %[[#false]] %[[#int_2]]
; CHECK-SPIRV: OpFunctionEnd
define dso_local spir_kernel void @testClusteredLogical(i32 addrspace(1)* nocapture) local_unnamed_addr {
%2 = tail call spir_func i32 @_Z38sub_group_clustered_reduce_logical_andij(i32 0, i32 2)
store i32 %2, i32 addrspace(1)* %0, align 4
%3 = tail call spir_func i32 @_Z37sub_group_clustered_reduce_logical_orij(i32 0, i32 2)
%4 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 1
store i32 %3, i32 addrspace(1)* %4, align 4
%5 = tail call spir_func i32 @_Z38sub_group_clustered_reduce_logical_xorij(i32 0, i32 2)
%6 = getelementptr inbounds i32, i32 addrspace(1)* %0, i64 2
store i32 %5, i32 addrspace(1)* %6, align 4
ret void
}
declare dso_local spir_func i32 @_Z38sub_group_clustered_reduce_logical_andij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z37sub_group_clustered_reduce_logical_orij(i32, i32) local_unnamed_addr
declare dso_local spir_func i32 @_Z38sub_group_clustered_reduce_logical_xorij(i32, i32) local_unnamed_addr
Reference in New Issue View Git Blame Copy Permalink