tlopex commented on code in PR #18547:
URL: https://github.com/apache/tvm/pull/18547#discussion_r2616505867
##########
src/meta_schedule/postproc/rewrite_parallel_vectorize_unroll.cc:
##########
@@ -425,25 +425,30 @@ class RewriteParallelVectorizeUnrollNode : public
PostprocNode {
tir::ParsedAnnotation parsed = parsed_root;
tir::AdjustParallelVectorize(sch, block_rv, loop_rvs, &parsed);
const int loops_num = loop_rvs.size();
- if (parsed.num_parallel_loops == loops_num &&
parsed.num_vectorize_loops == loops_num) {
- // Fuse, split, vectorize and parallelize
- tir::RewriteFuseSplitParallelVectorize(sch, &loop_rvs,
parsed.max_vectorize_extent);
- } else {
- // Parallel
- if (parsed.num_parallel_loops > 0) {
- tir::RewriteParallel(sch, parsed.num_parallel_loops, &loop_rvs);
+ try {
+ if (parsed.num_parallel_loops == loops_num &&
parsed.num_vectorize_loops == loops_num) {
+ // Fuse, split, vectorize and parallelize
+ tir::RewriteFuseSplitParallelVectorize(sch, &loop_rvs,
parsed.max_vectorize_extent);
+ } else {
+ // Parallel
+ if (parsed.num_parallel_loops > 0) {
+ tir::RewriteParallel(sch, parsed.num_parallel_loops, &loop_rvs);
+ }
+ // Vectorize
+ if (parsed.num_vectorize_loops > 0) {
+ tir::RewriteVectorize(sch, parsed.num_vectorize_loops,
&loop_rvs);
+ }
}
- // Vectorize
- if (parsed.num_vectorize_loops > 0) {
- tir::RewriteVectorize(sch, parsed.num_vectorize_loops, &loop_rvs);
+ // AutoUnroll
+ if (parsed.unroll_explicit != -1 || parsed.unroll_implicit != -1) {
+ ICHECK(parsed.unroll_explicit == -1 || parsed.unroll_implicit ==
-1);
+ int unroll_explicit = parsed.unroll_explicit != -1;
+ int max_step = parsed.unroll_explicit + parsed.unroll_implicit + 1;
+ tir::RewriteUnroll(sch, unroll_explicit, max_step, block_rv,
loop_rvs[0]);
}
- }
- // AutoUnroll
- if (parsed.unroll_explicit != -1 || parsed.unroll_implicit != -1) {
- ICHECK(parsed.unroll_explicit == -1 || parsed.unroll_implicit == -1);
- int unroll_explicit = parsed.unroll_explicit != -1;
- int max_step = parsed.unroll_explicit + parsed.unroll_implicit + 1;
- tir::RewriteUnroll(sch, unroll_explicit, max_step, block_rv,
loop_rvs[0]);
+ } catch (const tir::ScheduleError& e) {
+ DLOG(WARNING) << "Failed to apply parallelization/vectorization: "
<< e.what();
Review Comment:
Here I think catching the error and continuing may leave the schedule
partially mutated. It would be safer to return false on ScheduleError and let
the candidate be discarded, unless rollback semantics are guaranteed. cc
@tqchen
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
