[TIR][Schedule] Improve cache_index to cache common subexpressions #13700
Conversation
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
Collaborator
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.
Generated by tvm-bot |
multiverstack-intellif
force-pushed
the
cache
branch
from
9b75cbc to
9bde9d8
Compare
multiverstack-intellif
changed the title
multiverstack-intellif
force-pushed
the
cache
branch
5 times, most recently
from
66f3b2d to
2a804ef
Compare
Hzfengsy
reviewed
Jan 13, 2023
python/tvm/tir/schedule/schedule.py
Outdated
| self, | ||
| block: Union[BlockRV, str], | ||
| storage_scope: str, | ||
| cse_thresh: int, |
Member
There was a problem hiding this comment.
Can we set a default value for cse_thresh? e.g. 0/-1 for not enabling cse?
Contributor
Author
There was a problem hiding this comment.
Can we set a default value for
cse_thresh? e.g. 0/-1 for not enabling cse?
OK, I'll set 0 as caching all the index computation for the target block, should be useful for simple cases and expensive in complex cases.
multiverstack-intellif
force-pushed
the
cache
branch
from
2a804ef to
6a329c2
Compare
Hzfengsy
approved these changes
Jan 18, 2023
wrongtest-intellif
approved these changes
Jan 18, 2023
Much of the index computation is duplicated, so this PR is to do common subexpression analyze when performing cache_index. A threshold number is needed when identifying a common sub expr, since we only want to cache the most frequently appeared sub expressions. And an arith helper function is added to help analyze the target expr. A common use case may be calling tvm.arith.detect_common_subexpr() and determining based on its result whether to perform cache_index and its corresponding threshold.
multiverstack-intellif
force-pushed
the
cache
branch
from
6a329c2 to
c62b8d2
Compare
fzi-peccia
pushed a commit
to fzi-peccia/tvm
that referenced
this pull request
Mar 27, 2023
…pache#13700) Much of the index computation is duplicated, so this PR is to do common subexpression analyze when performing cache_index. A threshold number is needed when identifying a common sub expr, since we only want to cache the most frequently appeared sub expressions. And an arith helper function is added to help analyze the target expr. A common use case may be calling tvm.arith.detect_common_subexpr() and determining based on its result whether to perform cache_index and its corresponding threshold. Co-authored-by: Min Chen <chen.min@intellif.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Much of the index computation is duplicated, so this PR is to do common subexpression analyze when performing cache_index.
And because all block var related computation could be cached, buffer-based cache is somehow too restrictive. So here make it block-based.
A threshold number is needed when identifying a common sub expr, since we only want to cache the most frequently appeared sub expressions. A common use case may be like this: calling tvm.arith.detect_common_subexpr() and determining based on its result whether to perform cache_index and its corresponding threshold.
cc @wrongtest-intellif @Hzfengsy