Skip to content

[SPIR-V] Fix out-of-range value for NumWorkgroups builtin #127198

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Feb 18, 2025
Merged

[SPIR-V] Fix out-of-range value for NumWorkgroups builtin #127198

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions llvm/lib/Target/SPIRV/SPIRVBuiltins.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1779,6 +1779,7 @@ static bool generateGetQueryInst(const SPIRV::IncomingCall *Call,
SPIRV::BuiltIn::BuiltIn Value =
SPIRV::lookupGetBuiltin(Call->Builtin->Name, Call->Builtin->Set)->Value;
uint64_t IsDefault = (Value == SPIRV::BuiltIn::GlobalSize ||
Value == SPIRV::BuiltIn::NumWorkgroups ||
Value == SPIRV::BuiltIn::WorkgroupSize ||
Value == SPIRV::BuiltIn::EnqueuedWorkgroupSize);
return genWorkgroupQuery(Call, MIRBuilder, GR, Value, IsDefault ? 1 : 0);
Expand Down
55 changes: 55 additions & 0 deletions llvm/test/CodeGen/SPIRV/opencl/get_num_groups.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,55 @@
; RUN: llc -O0 -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s

;; The set of valid inputs for get_num_groups depends on the runtime NDRange,
;; but inputs outside of [0, 2] always return 1.
;; Here we assume Itanium mangling for function name.
declare i64 @_Z14get_num_groupsj(i32)

define i64 @foo(i32 %dim) {
%x = call i64 @_Z14get_num_groupsj(i32 0)
%y = call i64 @_Z14get_num_groupsj(i32 5)
%acc = add i64 %x, %y
%unknown = call i64 @_Z14get_num_groupsj(i32 %dim)
%ret = add i64 %acc, %unknown
ret i64 %ret
}

;; Capabilities:
; CHECK-DAG: OpCapability Kernel
; CHECK-DAG: OpCapability Int64

; CHECK-NOT: DAG-FENCE

;; Decorations:
; CHECK-DAG: OpDecorate %[[#GET_NUM_GROUPS:]] BuiltIn NumWorkgroups
; CHECK-DAG: OpDecorate %[[#GET_NUM_GROUPS]] Constant

; CHECK-NOT: DAG-FENCE

;; Types, Constants and Variables:
; CHECK-DAG: %[[#BOOL:]] = OpTypeBool
; CHECK-DAG: %[[#I32:]] = OpTypeInt 32 0
; CHECK-DAG: %[[#I64:]] = OpTypeInt 64 0
; CHECK-DAG: %[[#VEC:]] = OpTypeVector %[[#I64]] 3
; CHECK-DAG: %[[#PTR:]] = OpTypePointer Input %[[#VEC]]
; CHECK-DAG: %[[#FN:]] = OpTypeFunction %[[#I64]] %[[#I32]]
; CHECK-DAG: %[[#GET_NUM_GROUPS]] = OpVariable %[[#PTR]] Input
; CHECK-DAG: %[[#ONE:]] = OpConstant %[[#I64]] 1
; CHECK-DAG: %[[#THREE:]] = OpConstant %[[#I32]] 3

;; Functions:
; CHECK: OpFunction %[[#I64]] None %[[#FN]]
; CHECK: %[[#DIM:]] = OpFunctionParameter %[[#I32]]

;; get_num_groups(0): OpLoad + OpCompositeExtract.
; CHECK: %[[#TMP1:]] = OpLoad %[[#VEC]] %[[#GET_NUM_GROUPS]]
; CHECK: %[[#X:]] = OpCompositeExtract %[[#I64]] %[[#TMP1]] 0

;; get_num_groups(5): OpConstant of one.
; CHECK: OpIAdd %[[#I64]] %[[#X]] %[[#ONE]]

;; get_num_groups(dim): Implementation using OpSelect.
; CHECK-DAG: %[[#TMP2:]] = OpLoad %[[#VEC]] %[[#GET_NUM_GROUPS]]
; CHECK-DAG: %[[#TMP3:]] = OpVectorExtractDynamic %[[#I64]] %[[#TMP2]] %[[#DIM]]
; CHECK-DAG: %[[#COND:]] = OpULessThan %[[#BOOL]] %[[#DIM]] %[[#THREE]]
; CHECK: %[[#UNKNOWN:]] = OpSelect %[[#I64]] %[[#COND]] %[[#TMP3]] %[[#ONE]]