Reverting tilizing f32 on device #1669
Merged
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
sdjordjevicTT
requested review from
jnie-TT,
kmabeeTT,
AleksKnezevic,
pilkicTT,
svuckovicTT,
mtopalovicTT,
nobradovictt,
jserbedzijaTT and
azecevicTT
as code owners
pilkicTT
approved these changes
Dec 26, 2024
jserbedzijaTT
approved these changes
Dec 26, 2024
sdjordjevicTT
force-pushed
the
sdjordjevic/reverting_tilizing_f32_on_device
branch
from
40d79fe to
602fa4e
Compare
mtopalovicTT
approved these changes
Dec 26, 2024
|
Since this involves regression and blocks uplift to Forge-FE, I will merge this immediately. |
sdjordjevicTT
deleted the
sdjordjevic/reverting_tilizing_f32_on_device
branch
|
@sdjordjevicTT this seems to be a limitation of device tilize in general and not specific to fp32 tilize. Essentially if you change the data format in the llama matmul test to bf16 the same error will appear. This seems to happen when the tensor exceeds a certain size. We should probably sync with the ttnn folks to add support for this. Looking at the error message it seems to be a TODO item on their side |
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Outdated suggestions cannot be applied.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.
Reverting commit ad87a44 and PR #1647 that introduces llama matmul test failure in forge-fe:
tenstorrent/tt-forge-fe#954 - https://github.com/tenstorrent/tt-forge-fe/actions/runs/12501472126/job/34879043368?pr=954
Added a matmul repro test that fails before the revert, and passes after the revert.