[Flang][OpenMP] Enable no-loop kernels #155818
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -2590,13 +2590,27 @@ convertOmpWsloop(Operation &opInst, llvm::IRBuilderBase &builder, | |
| } | ||
|
|
||
| builder.SetInsertPoint(*regionBlock, (*regionBlock)->begin()); | ||
|
|
||
| bool noLoopMode = false; | ||
| omp::TargetOp targetOp = wsloopOp->getParentOfType<mlir::omp::TargetOp>(); | ||
| if (targetOp) { | ||
| Operation *targetCapturedOp = targetOp.getInnermostCapturedOmpOp(); | ||
|
There was a problem hiding this comment. Should we check here that the captured op is the !$omp target teams distribute parallel do
do i=1,10
!$omp do
do j=1,i
end do
end doI'm not actually sure about what is the expected behavior of this, but I imagine that no-loop would just refer to the outer loop, as it's the one for which the trip count can be evaluated in the host. There was a problem hiding this comment. Hi Sergio,
Yes. There is assumption in
There are 2 issues:
There was a problem hiding this comment. Thank you for the clarifications, Dominik.
What I mean is adding something like this: Operation *targetCapturedOp = targetOp.getInnermostCapturedOmpOp();
if (*loopOp == targetCapturedOp) {
omp::TargetRegionFlags kernelFlags = targetOp.getKernelExecFlags(targetCapturedOp);
...
}We need that because, if not, There was a problem hiding this comment. Fixed in commit: d2e88db |
||
| omp::TargetRegionFlags kernelFlags = | ||
| targetOp.getKernelExecFlags(targetCapturedOp); | ||
| if (omp::bitEnumContainsAll(kernelFlags, | ||
| omp::TargetRegionFlags::spmd | | ||
| omp::TargetRegionFlags::no_loop) && | ||
| !omp::bitEnumContainsAny(kernelFlags, omp::TargetRegionFlags::generic)) | ||
| noLoopMode = true; | ||
| } | ||
|
|
||
| llvm::OpenMPIRBuilder::InsertPointOrErrorTy wsloopIP = | ||
| ompBuilder->applyWorkshareLoop( | ||
| ompLoc.DL, loopInfo, allocaIP, loopNeedsBarrier, | ||
| convertToScheduleKind(schedule), chunk, isSimd, | ||
| scheduleMod == omp::ScheduleModifier::monotonic, | ||
| scheduleMod == omp::ScheduleModifier::nonmonotonic, isOrdered, | ||
| workshareLoopType, noLoopMode); | ||
|
|
||
| if (failed(handleError(wsloopIP, opInst))) | ||
| return failure(); | ||
|
|
@@ -5365,6 +5379,12 @@ initTargetDefaultAttrs(omp::TargetOp targetOp, Operation *capturedOp, | |
| ? llvm::omp::OMP_TGT_EXEC_MODE_GENERIC_SPMD | ||
| : llvm::omp::OMP_TGT_EXEC_MODE_GENERIC | ||
| : llvm::omp::OMP_TGT_EXEC_MODE_SPMD; | ||
| if (omp::bitEnumContainsAll(kernelFlags, | ||
| omp::TargetRegionFlags::spmd | | ||
| omp::TargetRegionFlags::no_loop) && | ||
| !omp::bitEnumContainsAny(kernelFlags, omp::TargetRegionFlags::generic)) | ||
| attrs.ExecFlags = llvm::omp::OMP_TGT_EXEC_MODE_SPMD_NO_LOOP; | ||
|
|
||
| attrs.MinTeams = minTeamsVal; | ||
| attrs.MaxTeams.front() = maxTeamsVal; | ||
| attrs.MinThreads = 1; | ||
|
|
||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -800,10 +800,6 @@ template <typename Ty> class StaticLoopChunker { | |
|
|
||
| // If we know we have more threads than iterations we can indicate that to | ||
| // avoid an outer loop. | ||
| if (OneIterationPerThread) | ||
| ASSERT(NumThreads >= NumIters, "Broken assumption"); | ||
|
|
||
|
|
@@ -851,10 +847,6 @@ template <typename Ty> class StaticLoopChunker { | |
|
|
||
| // If we know we have more blocks than iterations we can indicate that to | ||
| // avoid an outer loop. | ||
| if (OneIterationPerThread) | ||
| ASSERT(NumBlocks >= NumIters, "Broken assumption"); | ||
|
|
||
|
|
@@ -914,11 +906,6 @@ template <typename Ty> class StaticLoopChunker { | |
|
|
||
| // If we know we have more threads (across all blocks) than iterations we | ||
| // can indicate that to avoid an outer loop. | ||
| if (OneIterationPerThread) | ||
| ASSERT(NumBlocks * NumThreads >= NumIters, "Broken assumption"); | ||
|
|
||
|
|
||
| Original file line number | Diff line number | Diff line change | ||||
|---|---|---|---|---|---|---|
| @@ -0,0 +1,43 @@ | ||||||
| ! Check if the first OpenMP GPU kernel is promoted to no-loop mode. | ||||||
|
||||||
| ! The second cannot be promoted due to the limit on the number of teams. | ||||||
| ! REQUIRES: flang, amdgpu | ||||||
|
||||||
| ! REQUIRES: flang, amdgpu | |
| ! REQUIRES: flang |
Shouldn't this still be correct with nvidia GPUs?
There was a problem hiding this comment.
Yes, it should be correct with Nvida GPU. I fixed the test case.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I find these HelpTexts more confusing than helpful. Proposal:
(please adapt as you see fit)
The name "assume_threads/teams_oversubscription" seems disconnected from this function at first, but I think it comes from allowing oversubscription, where the "subscription" is the number of available teams/threads as determined by
thread_limit,num_teams,OMP_THREAD_LIMIT, etc., and threfore allowing the optimzer to oversubscribe this amount. Maybe there is a way to make the description more connected to the flag name.Avoid "Flang" in the HelpText. It might also made available in Clang on day.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for the feedback. I changed the description.