refactor: replace old warp size api #705
Open
+522
−364
Conversation
This file contains 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
|
Still in draft as we're doing an internal review in parallel. @stanleytsang-amd @RobsonRLemos feel to already run this through the tests to speed things along as I don't think much will change. I'll probably open up a pair of PRs to hipCUB and rocThrust to fix the compiler warnings from the deprecations tomorrow. |
Naraenda
force-pushed
the
deprecate-old-warp-size
branch
from
8046d96 to
76f60f1
Compare
Naraenda
force-pushed
the
deprecate-old-warp-size
branch
from
76f60f1 to
7765574
Compare
Naraenda
requested review from
a team,
stanleytsang-amd,
umfranzw,
RobsonRLemos and
lawruble13
as code owners
|
While applying the migration to the new API on hipCUB, I've found an unrollable loop which is fixed in 7765574. @stanleytsang-amd @RobsonRLemos Apologies for the delay, there were a few oversights that took a bit longer to resolve. |
spolifroni-amd
approved these changes
Mar 19, 2025
…/warp algo types on host
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Required steps to get rid of
__AMDGCN_WAVEFRONT_SIZE. This PR should be included in ROCm 6.5.::rocprim::arch::wavefront::min_size():::rocprim::wavefront::size()felt strange as it really isn't a top level API (i.e. algorithm)::rocprim::arch::wavefront_size()was also a possibility, but we might add more wavefront related functions. So scoping it made more sense.::rocprim::archwas more appropriate so we can add more architecture related constants such as number of LDS banks, etc.min_size()andmax_size()are the same on device for now.