Skip to content

Fix PyTorch fatbinary dumping #177

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
Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Fix PyTorch fatbinary dumping #177

wants to merge 3 commits into from

Conversation

@yuhc
Copy link
Member

@yuhc yuhc commented Jun 17, 2021

Description

The previous dump spec failed to dump fatbinaries with no kernel or no useful kernel. This diff handles this corner case.

Types of changes

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Document update (this change is mainly a documentation update)

Checklist:

  • My code passes format and lint checks.
  • My change requires a change to the documentation.
  • I have updated the documentation accordingly.
  • I have commented my code, particularly in hard-to-understand areas.
  • My changes generate no new warnings.
  • I have tested my code with a reasonable workload.
  • My code may break some other features.

@elementgreen @yuhc
Use ".nv.info." to match cuobjdump function name instead of .nv.info.…
00c59f8 
…_Z for cases where function does not start with _Z
@elementgreen @yuhc
Don't dump CUDA binaries which have no .nv.info CUDA_INFO argument in…
88e889a 
…formation

* Skip already processed functions by name by looking up in fatbin_funcs hash
* /tmp/fatbin-info.ava and ava_metadata(NULL)->fd_functions are only updated if the binary has functions
* Removed fatbin_num static integer in favor of ava_metadata(NULL)->num_fatbins
@elementgreen @yuhc
Use accomulated ht_name2idx function name hash for pytorch opt worker…
bd28a81 
… to handle duplicate function names between CUDA binaries

* Remove old __helper_load_function_arg_info since worker now uses the same function as guest
* Rename __helper_load_function_arg_info_guest to __helper_load_function_arg_info since it is now also used by worker
@yuhc yuhc added the bug Something isn't working label Jun 17, 2021
@yuhc yuhc mentioned this pull request Jun 17, 2021
Open
@photoszzt photoszzt mentioned this pull request Jun 30, 2021
Merged
11 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

bug Something isn't working

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@yuhc @elementgreen