Repository navigation
Qualcomm AI Engine Direct - Cast TopK int64 indices to int32 - #24054
Open
msluszniak wants to merge 1 commit into
Open
msluszniak wants to merge 1 commit into
msluszniak wants to merge 1 commit into
Conversation
QNN TopK writes int32 indices, but I64toI32 left the getitem of the indices declared int64, so a delegate output read them as int64 (half the values wrong) and a TopK left on CPU fed int64 into a delegated consumer. I64toI32 now casts the int64 outputs of multi-output ops (TopK) to int32 at their getitem.
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24054
Note: Links to docs will display an error until the docs builds have been completed. ❌ 4 New FailuresAs of commit ec55c2b with merge base e6cfbf8 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
executorch-triage
Bot
added
the
community: contribution
nil-is-all
added
partner: qualcomm
borys3kk
pushed a commit
to borys3kk/swm_executorch
that referenced
this pull request
Oct 22, 2026
QNN TopK writes int32 indices, but I64toI32 left the getitem of the indices declared int64, so a delegate output read them as int64 (half the values wrong) and a TopK left on CPU fed int64 into a delegated consumer. I64toI32 now casts the int64 outputs of multi-output ops (TopK) to int32 at their getitem. Backport of pytorch/executorch#24054. 1.4.1 has no rework pass tests.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
QNN TopK writes int32 indices, but
I64toI32kept theirgetitemint64 (the upstream TODO "support OP such as topK"), so delegated indices came back half zeros and CPU TopK indices broke a delegated consumer.I64toI32now casts the int64 outputs of TopK to int32 at thegetitem.Test plan
New
TestQNNFloatingPointModel::test_qnn_backend_topk_indicesandtest_qnn_backend_topk_on_cpu_and_indexfail on main and pass here (x86 simulator).test_passes.py,rework/passes/test.py(debug-handle maps updated for the new cast) give the same results as main;TestQNNFloatingPointModelpasses.cc @cccclai @winskuo-quic @shewu-quic @haowhsu-quic @DannyYuyang-quic @cbilgin @abhinaykukkadapu @psiddh @nil-is-all