Skip to content

[Flang][OpenMP] Use simdloop operation only for omp simd pragma #79559

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

Jump to bottom
Merged
merged 2 commits into from
Jan 29, 2024
Merged

[Flang][OpenMP] Use simdloop operation only for omp simd pragma #79559

merged 2 commits into from
Jan 29, 2024

Conversation

DominikAdamski
Copy link
Contributor

OpenMP standard differentiates between omp simd (2.9.3.1) and omp do/for simd (2.9.3.2 for OpenMP 5.0 standard) pragmas. The first one describes the loop which needs to be vectorized. The second pragma describes the loop which needs to be workshared between existing threads. Each thread can use SIMD instructions to execute its chunk of the loop.

That's why we need to model

!$omp simd
  do-loop

as omp.simdloop operation and add compiler hints for vectorization.

The worksharing loop:
!$omp do simd
do-loop
should be represented as worksharing loop (omp.wsloop).

Currently Flang denotes both type of OpenMP pragmas by omp.simdloop operation. In consequence we cannot differentiate between:

!$omp parallel simd
   do-loop

and

!$omp parallel do simd
   do-loop

The second loop should be workshared between multiple threads. The first one describes the loop which needs to be redundantly executed by multiple threads. Current Flang implementation does not perform worksharing for !$omp do simd pragma and generates valid code only for first case.

@skatrak will propose detailed MLIR representation of !$omp do simd pragma

@DominikAdamski
[Flang][OpenMP] Use simdloop operation only for omp simd pragma
5630ec5 
OpenMP standard differentiates between omp simd (2.9.3.1) and
omp do/for simd (2.9.3.2 for OpenMP 5.0 standard) pragmas.
The first one describes the loop which needs to be vectorized.
The second pragma describes the loop which needs to be workshared
between existing threads. Each thread can use SIMD instructions
to execute its chunk of the loop.

That's why we need to model
!$omp simd
  do-loop
as omp.simdloop operation and add compiler hints for vectorization.

The worksharing loop:
!$omp do simd
  do-loop
should be represented as worksharing loop.

Currently Flang denotes both type of OpenMP pragmas by omp.simdloop
operation. In consequence we cannot differentiate between:
!$omp parallel simd
   do-loop
and
!$omp parallel do simd
   do-loop
The second loop should be workshared between multiple threads.
The first one describes the loop which needs to be redundantly
executed by multiple threads. Current Flang implementation
does not perform worksharing for `!$omp do simd` pragma and
generates valid code only for first case.
@llvmbot llvmbot added flang Flang issues not falling into any other category flang:fir-hlfir flang:openmp labels Jan 26, 2024
@llvmbot
Copy link
Member

llvmbot commented Jan 26, 2024 •
edited
Loading

@llvm/pr-subscribers-flang-fir-hlfir

@llvm/pr-subscribers-flang-openmp

Author: Dominik Adamski (DominikAdamski)

Changes

OpenMP standard differentiates between omp simd (2.9.3.1) and omp do/for simd (2.9.3.2 for OpenMP 5.0 standard) pragmas. The first one describes the loop which needs to be vectorized. The second pragma describes the loop which needs to be workshared between existing threads. Each thread can use SIMD instructions to execute its chunk of the loop.

That's why we need to model

!$omp simd
  do-loop

as omp.simdloop operation and add compiler hints for vectorization.

The worksharing loop:
!$omp do simd
do-loop
should be represented as worksharing loop (omp.wsloop).

Currently Flang denotes both type of OpenMP pragmas by omp.simdloop operation. In consequence we cannot differentiate between:

!$omp parallel simd
   do-loop

and

!$omp parallel do simd
   do-loop

The second loop should be workshared between multiple threads. The first one describes the loop which needs to be redundantly executed by multiple threads. Current Flang implementation does not perform worksharing for !$omp do simd pragma and generates valid code only for first case.

@skatrak will propose detailed MLIR representation of !$omp do simd pragma

Full diff: https://github.com/llvm/llvm-project/pull/79559.diff

9 Files Affected:

  • (modified) flang/lib/Lower/OpenMP.cpp (+47-3)
  • (modified) flang/test/Lower/OpenMP/FIR/if-clause.f90 (+13-21)
  • (modified) flang/test/Lower/OpenMP/FIR/loop-combined.f90 (+3-3)
  • (added) flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90 (+16)
  • (added) flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90 (+14)
  • (added) flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90 (+14)
  • (added) flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90 (+14)
  • (modified) flang/test/Lower/OpenMP/if-clause.f90 (+13-21)
  • (modified) flang/test/Lower/OpenMP/loop-combined.f90 (+3-3)
diff --git a/flang/lib/Lower/OpenMP.cpp b/flang/lib/Lower/OpenMP.cpp
index d2215f4d1bf1ce..0888b25f1c59d5 100644
--- a/flang/lib/Lower/OpenMP.cpp
+++ b/flang/lib/Lower/OpenMP.cpp
@@ -3311,6 +3311,43 @@ static void createWsLoop(Fortran::lower::AbstractConverter &converter,
                                       /*outer=*/false, &dsp);
 }
 
+static void createSimdWsLoop(
+    Fortran::lower::AbstractConverter &converter,
+    Fortran::lower::pft::Evaluation &eval, llvm::omp::Directive ompDirective,
+    const Fortran::parser::OmpClauseList &beginClauseList,
+    const Fortran::parser::OmpClauseList *endClauseList, mlir::Location loc) {
+  ClauseProcessor cp(converter, beginClauseList);
+  cp.processTODO<
+      Fortran::parser::OmpClause::Aligned, Fortran::parser::OmpClause::Allocate,
+      Fortran::parser::OmpClause::Linear, Fortran::parser::OmpClause::Safelen,
+      Fortran::parser::OmpClause::Simdlen, Fortran::parser::OmpClause::Order>(
+      loc, ompDirective);
+  // 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.
+  createWsLoop(converter, eval, ompDirective, beginClauseList, endClauseList,
+               loc);
+}
+
+static bool isWorkshareSimdConstruct(llvm::omp::Directive ompDirective) {
+  switch (ompDirective) {
+  default:
+    return false;
+  case llvm::omp::OMPD_distribute_parallel_do_simd:
+  case llvm::omp::OMPD_do_simd:
+  case llvm::omp::OMPD_parallel_do_simd:
+  case llvm::omp::OMPD_target_parallel_do_simd:
+  case llvm::omp::OMPD_target_teams_distribute_parallel_do_simd:
+  case llvm::omp::OMPD_teams_distribute_parallel_do_simd:
+    return true;
+  }
+}
+
 static void genOMP(Fortran::lower::AbstractConverter &converter,
                    Fortran::lower::SymMap &symTable,
                    Fortran::semantics::SemanticsContext &semanticsContext,
@@ -3377,10 +3414,17 @@ static void genOMP(Fortran::lower::AbstractConverter &converter,
                               ")");
   }
 
-  // 2.9.3.1 SIMD construct
   if (llvm::omp::allSimdSet.test(ompDirective)) {
-    createSimdLoop(converter, eval, ompDirective, loopOpClauseList,
-                   currentLocation);
+    if (isWorkshareSimdConstruct(ompDirective)) {
+      // 2.9.3.2 Workshare SIMD construct
+      createSimdWsLoop(converter, eval, ompDirective, loopOpClauseList,
+                       endClauseList, currentLocation);
+
+    } else {
+      // 2.9.3.1 SIMD construct
+      createSimdLoop(converter, eval, ompDirective, loopOpClauseList,
+                     currentLocation);
+    }
   } else {
     createWsLoop(converter, eval, ompDirective, loopOpClauseList, endClauseList,
                  currentLocation);
diff --git a/flang/test/Lower/OpenMP/FIR/if-clause.f90 b/flang/test/Lower/OpenMP/FIR/if-clause.f90
index ef98a00f10dbd2..a1235be8e61ea2 100644
--- a/flang/test/Lower/OpenMP/FIR/if-clause.f90
+++ b/flang/test/Lower/OpenMP/FIR/if-clause.f90
@@ -28,7 +28,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! DO SIMD
   ! ----------------------------------------------------------------------------
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp do simd
@@ -36,15 +36,13 @@ program main
   end do
   !$omp end do simd
 
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp do simd if(.true.)
   do i = 1, 10
   end do
   !$omp end do simd
 
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp do simd if(simd: .true.)
   do i = 1, 10
   end do
@@ -103,7 +101,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp parallel do simd
@@ -113,8 +111,7 @@ program main
 
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(.true.)
   do i = 1, 10
   end do
@@ -122,8 +119,7 @@ program main
   
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(parallel: .true.) if(simd: .false.)
   do i = 1, 10
   end do
@@ -131,7 +127,7 @@ program main
   
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp parallel do simd if(parallel: .true.)
@@ -142,8 +138,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(simd: .true.)
   do i = 1, 10
   end do
@@ -306,7 +301,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp target parallel do simd
@@ -318,8 +313,7 @@ program main
   ! CHECK-SAME: if({{.*}})
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(.true.)
   do i = 1, 10
   end do
@@ -329,8 +323,7 @@ program main
   ! CHECK-SAME: if({{.*}})
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(target: .true.) if(parallel: .false.) &
   !$omp&                        if(simd: .true.)
   do i = 1, 10
@@ -342,7 +335,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp target parallel do simd if(target: .true.)
@@ -355,8 +348,7 @@ program main
   ! CHECK-SAME: {
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(parallel: .true.) if(simd: .false.)
   do i = 1, 10
   end do
diff --git a/flang/test/Lower/OpenMP/FIR/loop-combined.f90 b/flang/test/Lower/OpenMP/FIR/loop-combined.f90
index 117f7d625270ec..a6cec1beb49c86 100644
--- a/flang/test/Lower/OpenMP/FIR/loop-combined.f90
+++ b/flang/test/Lower/OpenMP/FIR/loop-combined.f90
@@ -23,7 +23,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! DO SIMD
   ! ----------------------------------------------------------------------------
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp do simd
   do i = 1, 10
   end do
@@ -33,7 +33,7 @@ program main
   ! PARALLEL DO SIMD
   ! ----------------------------------------------------------------------------
   ! CHECK: omp.parallel
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp parallel do simd
   do i = 1, 10
   end do
@@ -54,7 +54,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! CHECK: omp.target
   ! CHECK: omp.parallel
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp target parallel do simd
   do i = 1, 10
   end do
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90
new file mode 100644
index 00000000000000..b62c54182442ac
--- /dev/null
+++ b/flang/test/Lower/OpenMP/Todo/omp-do-simd-aligned.f90
@@ -0,0 +1,16 @@
+! This test checks lowering of OpenMP do simd aligned() pragma
+
+! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+subroutine testDoSimdAligned(int_array)
+        use iso_c_binding
+        type(c_ptr) :: int_array
+!CHECK: not yet implemented: Unhandled clause ALIGNED in DO SIMD construct
+!$omp do simd aligned(int_array)
+        do index_ = 1, 10
+          call c_test_call(int_array)
+        end do
+!$omp end do simd
+
+end subroutine testDoSimdAligned
+
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90
new file mode 100644
index 00000000000000..a9e0446ec8c34e
--- /dev/null
+++ b/flang/test/Lower/OpenMP/Todo/omp-do-simd-linear.f90
@@ -0,0 +1,14 @@
+! This test checks lowering of OpenMP do simd linear() pragma
+
+! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+subroutine testDoSimdLinear(int_array)
+        integer :: int_array(*)
+!CHECK: not yet implemented: Unhandled clause LINEAR in DO SIMD construct
+!$omp do simd linear(int_array)
+        do index_ = 1, 10
+        end do
+!$omp end do simd
+
+end subroutine testDoSimdLinear
+
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90
new file mode 100644
index 00000000000000..054eb52ea170ac
--- /dev/null
+++ b/flang/test/Lower/OpenMP/Todo/omp-do-simd-safelen.f90
@@ -0,0 +1,14 @@
+! This test checks lowering of OpenMP do simd safelen() pragma
+
+! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+subroutine testDoSimdSafelen(int_array)
+        integer :: int_array(*)
+!CHECK: not yet implemented: Unhandled clause SAFELEN in DO SIMD construct
+!$omp do simd safelen(4)
+        do index_ = 1, 10
+        end do
+!$omp end do simd
+
+end subroutine testDoSimdSafelen
+
diff --git a/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90 b/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90
new file mode 100644
index 00000000000000..bd00b6f336c931
--- /dev/null
+++ b/flang/test/Lower/OpenMP/Todo/omp-do-simd-simdlen.f90
@@ -0,0 +1,14 @@
+! This test checks lowering of OpenMP do simd simdlen() pragma
+
+! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+subroutine testDoSimdSimdlen(int_array)
+        integer :: int_array(*)
+!CHECK: not yet implemented: Unhandled clause SIMDLEN in DO SIMD construct
+!$omp do simd simdlen(4)
+        do index_ = 1, 10
+        end do
+!$omp end do simd
+
+end subroutine testDoSimdSimdlen
+
diff --git a/flang/test/Lower/OpenMP/if-clause.f90 b/flang/test/Lower/OpenMP/if-clause.f90
index 032009add31535..f982bf67b07225 100644
--- a/flang/test/Lower/OpenMP/if-clause.f90
+++ b/flang/test/Lower/OpenMP/if-clause.f90
@@ -28,7 +28,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! DO SIMD
   ! ----------------------------------------------------------------------------
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp do simd
@@ -36,15 +36,13 @@ program main
   end do
   !$omp end do simd
 
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp do simd if(.true.)
   do i = 1, 10
   end do
   !$omp end do simd
 
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp do simd if(simd: .true.)
   do i = 1, 10
   end do
@@ -103,7 +101,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp parallel do simd
@@ -113,8 +111,7 @@ program main
 
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(.true.)
   do i = 1, 10
   end do
@@ -122,8 +119,7 @@ program main
   
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(parallel: .true.) if(simd: .false.)
   do i = 1, 10
   end do
@@ -131,7 +127,7 @@ program main
   
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp parallel do simd if(parallel: .true.)
@@ -142,8 +138,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp parallel do simd if(simd: .true.)
   do i = 1, 10
   end do
@@ -306,7 +301,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp target parallel do simd
@@ -318,8 +313,7 @@ program main
   ! CHECK-SAME: if({{.*}})
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(.true.)
   do i = 1, 10
   end do
@@ -329,8 +323,7 @@ program main
   ! CHECK-SAME: if({{.*}})
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(target: .true.) if(parallel: .false.) &
   !$omp&                        if(simd: .true.)
   do i = 1, 10
@@ -342,7 +335,7 @@ program main
   ! CHECK:      omp.parallel
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
-  ! CHECK:      omp.simdloop
+  ! CHECK:      omp.wsloop
   ! CHECK-NOT:  if({{.*}})
   ! CHECK-SAME: {
   !$omp target parallel do simd if(target: .true.)
@@ -355,8 +348,7 @@ program main
   ! CHECK-SAME: {
   ! CHECK:      omp.parallel
   ! CHECK-SAME: if({{.*}})
-  ! CHECK:      omp.simdloop
-  ! CHECK-SAME: if({{.*}})
+  ! CHECK:      omp.wsloop
   !$omp target parallel do simd if(parallel: .true.) if(simd: .false.)
   do i = 1, 10
   end do
diff --git a/flang/test/Lower/OpenMP/loop-combined.f90 b/flang/test/Lower/OpenMP/loop-combined.f90
index 960e9518127d75..70488b6a769ce4 100644
--- a/flang/test/Lower/OpenMP/loop-combined.f90
+++ b/flang/test/Lower/OpenMP/loop-combined.f90
@@ -23,7 +23,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! DO SIMD
   ! ----------------------------------------------------------------------------
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp do simd
   do i = 1, 10
   end do
@@ -33,7 +33,7 @@ program main
   ! PARALLEL DO SIMD
   ! ----------------------------------------------------------------------------
   ! CHECK: omp.parallel
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp parallel do simd
   do i = 1, 10
   end do
@@ -54,7 +54,7 @@ program main
   ! ----------------------------------------------------------------------------
   ! CHECK: omp.target
   ! CHECK: omp.parallel
-  ! CHECK: omp.simdloop
+  ! CHECK: omp.wsloop
   !$omp target parallel do simd
   do i = 1, 10
   end do

skatrak
skatrak approved these changes Jan 26, 2024
View reviewed changes
Copy link
Member

@skatrak skatrak left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks Dominik, this looks good to me. I just have a small suggestion. Please wait for a day before merging, in case there are any concerns by other reviewers.

@DominikAdamski DominikAdamski mentioned this pull request Jan 26, 2024
Closed
mjklemm
mjklemm approved these changes Jan 26, 2024
View reviewed changes
Copy link
Contributor

@mjklemm mjklemm left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The reasoning for this PR is inline with what's needed for proper support of SIMD from an OpenMP spec perspective.

@DominikAdamski DominikAdamski merged commit ce80323 into llvm:main Jan 29, 2024
skatrak added a commit that referenced this pull request Jan 29, 2024
@skatrak
[MLIR][OpenMP] Add omp.simd operation
2c8e9c7 
This patch introduces the `omp.simd` operation. In contrast to the existing
`omp.simdloop` operation, it is intended to hold SIMD information within
worksharing loops, rather than representing a SIMD-only loop. Some examples
of such loops are "omp do/for simd", "omp distribute simd", "omp target teams
distribute parallel do/for simd", etc. For more context on this work, refer to
PR #79559.

This operation must always be nested within an `omp.wsloop` operation as its
only non-terminator child. It follows the same approach as the `omp.distribute`
operation, by serving as a simple wrapper operation holding clause information.
@skatrak skatrak mentioned this pull request Jan 29, 2024
Closed
@DominikAdamski
Copy link
Contributor Author

The PR for MLIR changes: #79843

DominikAdamski added a commit to DominikAdamski/aomp that referenced this pull request Feb 1, 2024
@DominikAdamski
[Flang] Remove -Dsimd= from babel test
26553f6 
Upstream Flang can generate valid workshare code
for `omp do` pragma.

See: llvm/llvm-project#79559
for more details.
@DominikAdamski DominikAdamski mentioned this pull request Feb 1, 2024
Merged
DominikAdamski added a commit to ROCm/aomp that referenced this pull request Feb 1, 2024
@DominikAdamski
[Flang] Remove -Dsimd= from babel test (#807)
09909f8 
Upstream Flang can generate valid workshare code
for `omp do` pragma.

See: llvm/llvm-project#79559
for more details.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@skatrak skatrak skatrak approved these changes

@mjklemm mjklemm mjklemm approved these changes

@jsjodin jsjodin Awaiting requested review from jsjodin

@kiranchandramohan kiranchandramohan Awaiting requested review from kiranchandramohan

@abidh abidh Awaiting requested review from abidh

@agozillon agozillon Awaiting requested review from agozillon

@dpalermo dpalermo Awaiting requested review from dpalermo

@kparzysz kparzysz Awaiting requested review from kparzysz

Assignees
No one assigned
Labels
flang:fir-hlfir flang:openmp flang Flang issues not falling into any other category
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@DominikAdamski @llvmbot @skatrak @mjklemm