Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Support for fp16 in SpM x DnsM on GPU #18930

Merged
merged 2 commits into from
Aug 28, 2020

Conversation

access2rohit
Copy link
Contributor

@access2rohit access2rohit commented Aug 15, 2020

Description

enables fp16 DType suport in following case

dot(csr, dense1) = dense2 and dot(csr.T, dense1) = dense2

Will add test case shortly

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

  • Changes are complete (i.e. I finished coding on this PR)
  • All changes have test coverage:
  • Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
  • Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
  • Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
  • To the best of my knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Testing

pytest) ubuntu@ip-172-31-90-243 ~/workspace/incubator-mxnet (spgemm_fp16) $ python -m pytest -s --exitfirst --verbose tests/python/gpu/test_operator_gpu.py::test_fp16_spmm
=============================================================================================== test session starts ===============================================================================================
platform linux -- Python 3.6.10, pytest-5.3.5, py-1.8.2, pluggy-0.13.1 -- /home/ubuntu/anaconda3/envs/pytest/bin/python
cachedir: .pytest_cache
rootdir: /home/ubuntu/workspace/incubator-mxnet, inifile: pytest.ini
collected 1 item

tests/python/gpu/test_operator_gpu.py::test_fp16_spmm [09:52:41] ../src/base.cc:84: Upgrade advisory: this mxnet has been built against cuDNN lib version 7501, which is older than the oldest version tested by CI (7600).  Set MXNET_CUDNN_LIB_CHECKING=0 to quiet this warning.
[09:52:43] ../src/storage/storage.cc:198: Using Pooled (Naive) StorageManager for GPU
PASSED

@mxnet-bot
Copy link

Hey @access2rohit , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

  • To trigger all jobs: @mxnet-bot run ci [all]
  • To trigger specific jobs: @mxnet-bot run ci [job1, job2]

CI supported jobs: [windows-cpu, clang, centos-gpu, unix-cpu, website, miscellaneous, edge, windows-gpu, centos-cpu, unix-gpu, sanity]


Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

@access2rohit
Copy link
Contributor Author

@mxnet-label-bot add [pr-work-in-progress]

@lanking520 lanking520 added the pr-work-in-progress PR is still work in progress label Aug 15, 2020
@access2rohit
Copy link
Contributor Author

@ptrendx Can you take an initial look ? I am yet to add test case though

@ptrendx
Copy link
Member

ptrendx commented Aug 17, 2020

Looks good :-)

@access2rohit access2rohit changed the title [WIP]Support for fp16 in SpM x DnsM Support for fp16 in SpM x DnsM on GPU Aug 27, 2020
@access2rohit
Copy link
Contributor Author

access2rohit commented Aug 27, 2020

@mxnet-label-bot update [pr-awaiting-review]

@lanking520 lanking520 added pr-awaiting-review PR is waiting for code review and removed pr-work-in-progress PR is still work in progress labels Aug 27, 2020
@access2rohit
Copy link
Contributor Author

@ptrendx I have added test case. Can you help with review and merge

@sandeep-krishnamurthy
Copy link
Contributor

Merging per previous review by @ptrendx

@sandeep-krishnamurthy sandeep-krishnamurthy merged commit 29c90ee into apache:master Aug 28, 2020
access2rohit added a commit to access2rohit/incubator-mxnet that referenced this pull request Sep 2, 2020
* Support for fp16 in SpGeMM

* adding test for GPU spmm

Co-authored-by: Rohit Kumar Srivastava <srivastava.141@buckeyemail.osu.edu>
samskalicky pushed a commit that referenced this pull request Sep 3, 2020
Backported #18930

* Support for fp16 in SpGeMM

* adding test for GPU spmm

Co-authored-by: Rohit Kumar Srivastava <srivastava.141@buckeyemail.osu.edu>

Co-authored-by: Rohit Kumar Srivastava <srivastava.141@buckeyemail.osu.edu>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
pr-awaiting-review PR is waiting for code review
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants