Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

skip inner broadcast simplification for layout #3771

Closed
wants to merge 2 commits into from
Closed

skip inner broadcast simplification for layout #3771

wants to merge 2 commits into from

Conversation

shivadbhavsar
Copy link
Contributor

This is more of a temporary fix for #3649 since its causing multiple regressions. This simplification does not really make much sense to me with the layout op anyways.

I think the real issue is, should there even be a layout inserted here? Seems wrong that a "layout_convolution" pass is adding an op in a graph that has no convolutions?

@shivadbhavsar shivadbhavsar requested review from lakhinderwalia, pfultz2 and CharlieL7 and removed request for causten January 21, 2025 00:02
@shivadbhavsar shivadbhavsar self-assigned this Jan 21, 2025
@shivadbhavsar shivadbhavsar linked an issue Jan 21, 2025 that may be closed by this pull request
Fail in find_inner_broadcast due to preserve_output_layout #3649
Open
@shivadbhavsar shivadbhavsar added the bugfix Fixes a bug found in the code. label Jan 21, 2025
@codecov Codecov
Copy link

codecov bot commented Jan 21, 2025 •
edited
Loading

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.28%. Comparing base (50c6848) to head (47194f6).

Additional details and impacted files 
@@           Coverage Diff            @@
##           develop    #3771   +/-   ##
========================================
  Coverage    92.28%   92.28%           
========================================
  Files          519      519           
  Lines        22227    22228    +1     
========================================
+ Hits         20512    20513    +1     
  Misses        1715     1715

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@kahmed10
Copy link
Collaborator

I've already added this check to the recently merged PR.

@migraphx-bot
Copy link
Collaborator

Test Batch Rate new
e7c7c4		 Rate old
976ae7		 Diff Compare
torchvision-resnet50 64 3,252.99 3,250.58 0.07% ✅
torchvision-resnet50_fp16 64 6,924.39 6,923.22 0.02% ✅
torchvision-densenet121 32 2,453.80 2,454.68 -0.04% ✅
torchvision-densenet121_fp16 32 4,184.47 4,162.95 0.52% ✅
torchvision-inceptionv3 32 1,629.66 1,629.37 0.02% ✅
torchvision-inceptionv3_fp16 32 2,716.58 2,714.44 0.08% ✅
cadene-inceptionv4 16 763.13 762.87 0.03% ✅
cadene-resnext64x4 16 812.45 812.57 -0.01% ✅
slim-mobilenet 64 7,457.56 7,455.05 0.03% ✅
slim-nasnetalarge 64 208.60 208.54 0.03% ✅
slim-resnet50v2 64 3,443.38 3,446.59 -0.09% ✅
bert-mrpc-onnx 8 1,148.13 1,147.30 0.07% ✅
bert-mrpc-tf 1 487.53 475.38 2.56% ✅
pytorch-examples-wlang-gru 1 479.37 479.10 0.06% ✅
pytorch-examples-wlang-lstm 1 446.95 446.88 0.02% ✅
torchvision-resnet50_1 1 798.91 806.66 -0.96% ✅
cadene-dpn92_1 1 426.06 427.99 -0.45% ✅
cadene-resnext101_1 1 384.42 384.95 -0.14% ✅
onnx-taau-downsample 1 373.26 373.59 -0.09% ✅
dlrm-criteoterabyte 1 33.34 33.31 0.10% ✅
dlrm-criteoterabyte_fp16 1 52.62 52.40 0.43% ✅
agentmodel 1 8,504.78 8,333.33 2.06% ✅
unet_fp16 2 58.45 58.42 0.05% ✅
resnet50v1_fp16 1 980.68 1,034.28 -5.18% 🔴
resnet50v1_int8 1 1,026.18 1,034.34 -0.79% ✅
bert_base_cased_fp16 64 1,180.95 1,181.27 -0.03% ✅
bert_large_uncased_fp16 32 365.16 365.29 -0.04% ✅
bert_large_fp16 1 201.02 198.45 1.29% ✅
distilgpt2_fp16 16 2,226.99 2,225.12 0.08% ✅
yolov5s 1 530.80 524.28 1.24% ✅
tinyllama 1 43.55 43.59 -0.08% ✅
vicuna-fastchat 1 174.08 175.96 -1.07% ✅
whisper-tiny-encoder 1 418.16 418.14 0.00% ✅
whisper-tiny-decoder 1 432.47 431.26 0.28% ✅

This build is not recommended to merge 🔴

@migraphx-bot
Copy link
Collaborator


     ✅ bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

     ✅ bert-mrpc-tf: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

     ✅ torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     ✅ dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     ✅ agentmodel: PASSED: MIGraphX meets tolerance

     ✅ unet: PASSED: MIGraphX meets tolerance

     ✅ resnet50v1: PASSED: MIGraphX meets tolerance

     ✅ bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output


     ✅ bert_large: PASSED: MIGraphX meets tolerance

     ✅ yolov5s: PASSED: MIGraphX meets tolerance

     ✅ tinyllama: PASSED: MIGraphX meets tolerance

     ✅ vicuna-fastchat: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-encoder: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-decoder: PASSED: MIGraphX meets tolerance

     ✅ distilgpt2_fp16: PASSED: MIGraphX meets tolerance

@CharlieL7
Copy link
Collaborator

Already in develop branch

@CharlieL7 CharlieL7 closed this Jan 21, 2025
@shivadbhavsar
Copy link
Contributor Author

shivadbhavsar commented Jan 21, 2025 •
edited
Loading

wait can you please link the PR, I am still hitting this issue with devlop branch nvm,

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@lakhinderwalia lakhinderwalia Awaiting requested review from lakhinderwalia

@pfultz2 pfultz2 Awaiting requested review from pfultz2

@CharlieL7 CharlieL7 Awaiting requested review from CharlieL7

Assignees

@shivadbhavsar shivadbhavsar

Labels
bugfix Fixes a bug found in the code.
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Fail in find_inner_broadcast due to preserve_output_layout
4 participants
@shivadbhavsar @kahmed10 @migraphx-bot @CharlieL7