[flang-commits] [flang] [mlir] [Flang][OpenMP] Add lowering support for DO SIMD (PR #97718)
Sergio Afonso via flang-commits
flang-commits at lists.llvm.org
Tue Jul 9 02:34:20 PDT 2024
https://github.com/skatrak updated https://github.com/llvm/llvm-project/pull/97718
>From 6fb2f2bfa68b49801fa3a23076915db721458cf2 Mon Sep 17 00:00:00 2001
From: Sergio Afonso <safonsof at amd.com>
Date: Thu, 4 Jul 2024 12:56:43 +0100
Subject: [PATCH] [Flang][OpenMP] Add lowering support for DO SIMD
This patch adds support for lowering 'DO SIMD' constructs to MLIR. SIMD
information is now stored in an `omp.simd` loop wrapper, which is currently
ignored by the OpenMP dialect to LLVM IR translation stage.
The end result is that runtime behavior of compiled 'DO SIMD' constructs does
not change after this patch, so 'DO SIMD' still runs like 'DO' (i.e. SIMD width
= 1). However, all of the required information is now present in the resulting
MLIR representation.
To avoid confusion, the previous wsloop-simd.f90 lit test is renamed to
wsloop-schedule.f90 and a new wsloop-simd.f90 test is created to check the
addition of SIMD clauses to the `omp.simd` operation produced when a 'DO SIMD'
construct is lowered to MLIR.
---
flang/lib/Lower/OpenMP/OpenMP.cpp | 51 +++++++++----
.../Lower/OpenMP/Todo/omp-do-simd-aligned.f90 | 16 ----
.../Lower/OpenMP/Todo/omp-do-simd-linear.f90 | 2 +-
.../Lower/OpenMP/Todo/omp-do-simd-safelen.f90 | 14 ----
.../Lower/OpenMP/Todo/omp-do-simd-simdlen.f90 | 14 ----
flang/test/Lower/OpenMP/if-clause.f90 | 31 ++++++++
flang/test/Lower/OpenMP/loop-compound.f90 | 3 +
flang/test/Lower/OpenMP/wsloop-schedule.f90 | 37 ++++++++++
flang/test/Lower/OpenMP/wsloop-simd.f90 | 74 +++++++++++--------
.../OpenMP/OpenMPToLLVMIRTranslation.cpp | 3 +
10 files changed, 155 insertions(+), 90 deletions(-)
delete mode 100644 flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90
delete mode 100644 flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90
delete mode 100644 flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90
create mode 100644 flang/test/Lower/OpenMP/wsloop-schedule.f90
diff --git a/flang/lib/Lower/OpenMP/OpenMP.cpp b/flang/lib/Lower/OpenMP/OpenMP.cpp
index 3dac44ac63691..b1bb4c11f86dd 100644
--- a/flang/lib/Lower/OpenMP/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP/OpenMP.cpp
@@ -1986,19 +1986,44 @@ static void genCompositeDoSimd(lower::AbstractConverter &converter,
const ConstructQueue &queue,
ConstructQueue::iterator item,
DataSharingProcessor &dsp) {
- ClauseProcessor cp(converter, semaCtx, item->clauses);
- cp.processTODO<clause::Aligned, clause::Allocate, clause::Linear,
- clause::Safelen, clause::Simdlen>(loc,
- llvm::omp::OMPD_do_simd);
- // TODO: Add support for vectorization - add vectorization hints inside loop
- // body.
- // OpenMP standard does not specify the length of vector instructions.
- // Currently we safely assume that for !$omp do simd pragma the SIMD length
- // is equal to 1 (i.e. we generate standard workshare loop).
- // When support for vectorization is enabled, then we need to add handling of
- // if clause. Currently if clause can be skipped because we always assume
- // SIMD length = 1.
- genStandaloneDo(converter, symTable, semaCtx, eval, loc, queue, item, dsp);
+ lower::StatementContext stmtCtx;
+
+ // Clause processing.
+ mlir::omp::WsloopClauseOps wsloopClauseOps;
+ llvm::SmallVector<const semantics::Symbol *> wsloopReductionSyms;
+ llvm::SmallVector<mlir::Type> wsloopReductionTypes;
+ genWsloopClauses(converter, semaCtx, stmtCtx, item->clauses, loc,
+ wsloopClauseOps, wsloopReductionTypes, wsloopReductionSyms);
+
+ mlir::omp::SimdClauseOps simdClauseOps;
+ genSimdClauses(converter, semaCtx, item->clauses, loc, simdClauseOps);
+
+ mlir::omp::LoopNestClauseOps loopNestClauseOps;
+ llvm::SmallVector<const semantics::Symbol *> iv;
+ genLoopNestClauses(converter, semaCtx, eval, item->clauses, loc,
+ loopNestClauseOps, iv);
+
+ // Operation creation.
+ // TODO: Add private variables to entry block arguments.
+ auto wsloopOp = genWrapperOp<mlir::omp::WsloopOp>(
+ converter, loc, wsloopClauseOps, wsloopReductionTypes);
+
+ // TODO: Populate entry block arguments with reduction and private variables.
+ auto simdOp = genWrapperOp<mlir::omp::SimdOp>(converter, loc, simdClauseOps,
+ /*blockArgTypes=*/{});
+
+ // Construct wrapper entry block list and associated symbols. It is important
+ // that the symbol and block argument order match, so that the symbol-value
+ // bindings created are correct.
+ // TODO: Add omp.wsloop private and omp.simd private and reduction args.
+ auto wrapperArgs = llvm::to_vector(llvm::concat<mlir::BlockArgument>(
+ wsloopOp.getRegion().getArguments(), simdOp.getRegion().getArguments()));
+
+ assert(wsloopReductionSyms.size() == wrapperArgs.size() &&
+ "Number of symbols and wrapper block arguments must match");
+ genLoopNestOp(converter, symTable, semaCtx, eval, loc, queue, item,
+ loopNestClauseOps, iv, wsloopReductionSyms, wrapperArgs,
+ llvm::omp::Directive::OMPD_do_simd, dsp);
}
static void genCompositeTaskloopSimd(
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90
deleted file mode 100644
index b62c54182442a..0000000000000
--- a/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90
+++ /dev/null
@@ -1,16 +0,0 @@
-! This test checks lowering of OpenMP do simd aligned() pragma
-
-! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-subroutine testDoSimdAligned(int_array)
- use iso_c_binding
- type(c_ptr) :: int_array
-!CHECK: not yet implemented: Unhandled clause ALIGNED in DO SIMD construct
-!$omp do simd aligned(int_array)
- do index_ = 1, 10
- call c_test_call(int_array)
- end do
-!$omp end do simd
-
-end subroutine testDoSimdAligned
-
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90
index a9e0446ec8c34..2f5366c2a5b36 100644
--- a/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90
+++ b/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90
@@ -4,7 +4,7 @@
! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
subroutine testDoSimdLinear(int_array)
integer :: int_array(*)
-!CHECK: not yet implemented: Unhandled clause LINEAR in DO SIMD construct
+!CHECK: not yet implemented: Unhandled clause LINEAR in DO construct
!$omp do simd linear(int_array)
do index_ = 1, 10
end do
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90
deleted file mode 100644
index 054eb52ea170a..0000000000000
--- a/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90
+++ /dev/null
@@ -1,14 +0,0 @@
-! This test checks lowering of OpenMP do simd safelen() pragma
-
-! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-subroutine testDoSimdSafelen(int_array)
- integer :: int_array(*)
-!CHECK: not yet implemented: Unhandled clause SAFELEN in DO SIMD construct
-!$omp do simd safelen(4)
- do index_ = 1, 10
- end do
-!$omp end do simd
-
-end subroutine testDoSimdSafelen
-
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90
deleted file mode 100644
index bd00b6f336c93..0000000000000
--- a/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90
+++ /dev/null
@@ -1,14 +0,0 @@
-! This test checks lowering of OpenMP do simd simdlen() pragma
-
-! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
-subroutine testDoSimdSimdlen(int_array)
- integer :: int_array(*)
-!CHECK: not yet implemented: Unhandled clause SIMDLEN in DO SIMD construct
-!$omp do simd simdlen(4)
- do index_ = 1, 10
- end do
-!$omp end do simd
-
-end subroutine testDoSimdSimdlen
-
diff --git a/flang/test/Lower/OpenMP/if-clause.f90 b/flang/test/Lower/OpenMP/if-clause.f90
index 2c9a66e7bc11e..ea730b5f1d9db 100644
--- a/flang/test/Lower/OpenMP/if-clause.f90
+++ b/flang/test/Lower/OpenMP/if-clause.f90
@@ -30,6 +30,9 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-NOT: if({{.*}})
+ ! CHECK-SAME: {
! CHECK-NEXT: omp.loop_nest
!$omp do simd
do i = 1, 10
@@ -39,6 +42,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp do simd if(.true.)
do i = 1, 10
@@ -48,6 +53,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp do simd if(simd: .true.)
do i = 1, 10
@@ -122,6 +129,9 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-NOT: if({{.*}})
+ ! CHECK-SAME: {
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd
do i = 1, 10
@@ -133,6 +143,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd if(.true.)
do i = 1, 10
@@ -144,6 +156,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd if(parallel: .true.) if(simd: .false.)
do i = 1, 10
@@ -155,6 +169,9 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-NOT: if({{.*}})
+ ! CHECK-SAME: {
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd if(parallel: .true.)
do i = 1, 10
@@ -167,6 +184,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd if(simd: .true.)
do i = 1, 10
@@ -355,6 +374,9 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-NOT: if({{.*}})
+ ! CHECK-SAME: {
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd
do i = 1, 10
@@ -368,6 +390,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd if(.true.)
do i = 1, 10
@@ -381,6 +405,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd if(target: .true.) if(parallel: .false.) &
!$omp& if(simd: .true.)
@@ -396,6 +422,9 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-NOT: if({{.*}})
+ ! CHECK-SAME: {
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd if(target: .true.)
do i = 1, 10
@@ -410,6 +439,8 @@ program main
! CHECK: omp.wsloop
! CHECK-NOT: if({{.*}})
! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: if({{.*}})
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd if(parallel: .true.) if(simd: .false.)
do i = 1, 10
diff --git a/flang/test/Lower/OpenMP/loop-compound.f90 b/flang/test/Lower/OpenMP/loop-compound.f90
index 5012008b07671..383a3716a9439 100644
--- a/flang/test/Lower/OpenMP/loop-compound.f90
+++ b/flang/test/Lower/OpenMP/loop-compound.f90
@@ -23,6 +23,7 @@ program main
! DO SIMD
! ----------------------------------------------------------------------------
! CHECK: omp.wsloop
+ ! CHECK-NEXT: omp.simd
! CHECK-NEXT: omp.loop_nest
!$omp do simd
do i = 1, 10
@@ -34,6 +35,7 @@ program main
! ----------------------------------------------------------------------------
! CHECK: omp.parallel
! CHECK: omp.wsloop
+ ! CHECK-NEXT: omp.simd
! CHECK-NEXT: omp.loop_nest
!$omp parallel do simd
do i = 1, 10
@@ -57,6 +59,7 @@ program main
! CHECK: omp.target
! CHECK: omp.parallel
! CHECK: omp.wsloop
+ ! CHECK-NEXT: omp.simd
! CHECK-NEXT: omp.loop_nest
!$omp target parallel do simd
do i = 1, 10
diff --git a/flang/test/Lower/OpenMP/wsloop-schedule.f90 b/flang/test/Lower/OpenMP/wsloop-schedule.f90
new file mode 100644
index 0000000000000..1df67474d65e3
--- /dev/null
+++ b/flang/test/Lower/OpenMP/wsloop-schedule.f90
@@ -0,0 +1,37 @@
+! This test checks lowering of OpenMP DO Directive(Worksharing) with
+! simd schedule modifier.
+
+! RUN: bbc -fopenmp -emit-hlfir %s -o - | FileCheck %s
+
+program wsloop_dynamic
+ integer :: i
+!CHECK-LABEL: func @_QQmain()
+
+!$OMP PARALLEL
+!CHECK: omp.parallel {
+
+!$OMP DO SCHEDULE(simd: runtime)
+!CHECK: %[[WS_LB:.*]] = arith.constant 1 : i32
+!CHECK: %[[WS_UB:.*]] = arith.constant 9 : i32
+!CHECK: %[[WS_STEP:.*]] = arith.constant 1 : i32
+!CHECK: omp.wsloop schedule(runtime, simd) nowait {
+!CHECK-NEXT: omp.loop_nest (%[[I:.*]]) : i32 = (%[[WS_LB]]) to (%[[WS_UB]]) inclusive step (%[[WS_STEP]]) {
+!CHECK: fir.store %[[I]] to %[[STORE:.*]]#1 : !fir.ref<i32>
+
+ do i=1, 9
+ print*, i
+!CHECK: %[[RTBEGIN:.*]] = fir.call @_FortranAioBeginExternalListOutput
+!CHECK: %[[LOAD:.*]] = fir.load %[[STORE]]#0 : !fir.ref<i32>
+!CHECK: fir.call @_FortranAioOutputInteger32(%[[RTBEGIN]], %[[LOAD]]) {{.*}}: (!fir.ref<i8>, i32) -> i1
+!CHECK: fir.call @_FortranAioEndIoStatement(%[[RTBEGIN]]) {{.*}}: (!fir.ref<i8>) -> i32
+ end do
+!CHECK: omp.yield
+!CHECK: }
+!CHECK: omp.terminator
+!CHECK: }
+!CHECK: omp.terminator
+!CHECK: }
+
+!$OMP END DO NOWAIT
+!$OMP END PARALLEL
+end
diff --git a/flang/test/Lower/OpenMP/wsloop-simd.f90 b/flang/test/Lower/OpenMP/wsloop-simd.f90
index 1df67474d65e3..e331f1ba15f0e 100644
--- a/flang/test/Lower/OpenMP/wsloop-simd.f90
+++ b/flang/test/Lower/OpenMP/wsloop-simd.f90
@@ -1,37 +1,47 @@
-! This test checks lowering of OpenMP DO Directive(Worksharing) with
-! simd schedule modifier.
+! This test checks lowering of OpenMP DO SIMD composite constructs.
! RUN: bbc -fopenmp -emit-hlfir %s -o - | FileCheck %s
+! RUN: %flang_fc1 -fopenmp -emit-hlfir %s -o - | FileCheck %s
-program wsloop_dynamic
- integer :: i
-!CHECK-LABEL: func @_QQmain()
+! CHECK-LABEL: func.func @_QPdo_simd_aligned(
+subroutine do_simd_aligned(A)
+ use iso_c_binding
+ type(c_ptr) :: A
+
+ ! CHECK: omp.wsloop
+ ! CHECK-NOT: aligned({{.*}})
+ ! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: aligned({{.*}})
+ !$omp do simd aligned(A)
+ do index_ = 1, 10
+ call c_test_call(A)
+ end do
+ !$omp end do simd
+end subroutine do_simd_aligned
-!$OMP PARALLEL
-!CHECK: omp.parallel {
+! CHECK-LABEL: func.func @_QPdo_simd_safelen(
+subroutine do_simd_safelen()
+ ! CHECK: omp.wsloop
+ ! CHECK-NOT: safelen({{.*}})
+ ! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: safelen({{.*}})
+ !$omp do simd safelen(4)
+ do index_ = 1, 10
+ end do
+ !$omp end do simd
+end subroutine do_simd_safelen
-!$OMP DO SCHEDULE(simd: runtime)
-!CHECK: %[[WS_LB:.*]] = arith.constant 1 : i32
-!CHECK: %[[WS_UB:.*]] = arith.constant 9 : i32
-!CHECK: %[[WS_STEP:.*]] = arith.constant 1 : i32
-!CHECK: omp.wsloop schedule(runtime, simd) nowait {
-!CHECK-NEXT: omp.loop_nest (%[[I:.*]]) : i32 = (%[[WS_LB]]) to (%[[WS_UB]]) inclusive step (%[[WS_STEP]]) {
-!CHECK: fir.store %[[I]] to %[[STORE:.*]]#1 : !fir.ref<i32>
-
- do i=1, 9
- print*, i
-!CHECK: %[[RTBEGIN:.*]] = fir.call @_FortranAioBeginExternalListOutput
-!CHECK: %[[LOAD:.*]] = fir.load %[[STORE]]#0 : !fir.ref<i32>
-!CHECK: fir.call @_FortranAioOutputInteger32(%[[RTBEGIN]], %[[LOAD]]) {{.*}}: (!fir.ref<i8>, i32) -> i1
-!CHECK: fir.call @_FortranAioEndIoStatement(%[[RTBEGIN]]) {{.*}}: (!fir.ref<i8>) -> i32
- end do
-!CHECK: omp.yield
-!CHECK: }
-!CHECK: omp.terminator
-!CHECK: }
-!CHECK: omp.terminator
-!CHECK: }
-
-!$OMP END DO NOWAIT
-!$OMP END PARALLEL
-end
+! CHECK-LABEL: func.func @_QPdo_simd_simdlen(
+subroutine do_simd_simdlen()
+ ! CHECK: omp.wsloop
+ ! CHECK-NOT: simdlen({{.*}})
+ ! CHECK-SAME: {
+ ! CHECK-NEXT: omp.simd
+ ! CHECK-SAME: simdlen({{.*}})
+ !$omp do simd simdlen(4)
+ do index_ = 1, 10
+ end do
+ !$omp end do simd
+end subroutine do_simd_simdlen
diff --git a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
index 35971fbacbf91..0c9c699a1f390 100644
--- a/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -899,6 +899,9 @@ static LogicalResult
convertOmpWsloop(Operation &opInst, llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation) {
auto wsloopOp = cast<omp::WsloopOp>(opInst);
+ // FIXME: Here any other nested wrappers (e.g. omp.simd) are skipped, so
+ // codegen for composite constructs like 'DO/FOR SIMD' will be the same as for
+ // 'DO/FOR'.
auto loopOp = cast<omp::LoopNestOp>(wsloopOp.getWrappedLoop());
llvm::ArrayRef<bool> isByRef = getIsByRef(wsloopOp.getReductionVarsByref());
More information about the flang-commits
mailing list