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

Enable Bert with Semi Structure Sparsity on ROCm #133934

Draft
wants to merge 53 commits into
base: main
Choose a base branch
from
Draft

Enable Bert with Semi Structure Sparsity on ROCm #133934

wants to merge 53 commits into from
+207 −17

Conversation

petrex
Copy link
Contributor

@petrex petrex commented Aug 19, 2024 •
edited
Loading

Draft PR. Goal is to Enable Bert with Semi Structure Sparsity on ROCm.
Refer to https://pytorch.org/tutorials/advanced/semi_structured_sparse.html

require gfx942+ and rocm 6.4

Note that this PR is built on top of #124320

cc @jeffdaily @sunway513 @jithunnair-amd @pruthvistony @ROCmSupport @dllehr-amd @jataylo @hongxiayang

@pytorch-bot PyTorch Bot
Copy link

pytorch-bot bot commented Aug 19, 2024 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/133934

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit ac130cb with merge base 9b76449 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@pytorch-bot pytorch-bot bot added module: rocm AMD GPU support for Pytorch release notes: sparse release notes category labels Aug 19, 2024
@linux-foundation-easycla Linux Foundation: EasyCLA
Copy link

linux-foundation-easycla bot commented Aug 19, 2024 •
edited
Loading

CLA Signed

The committers listed above are authorized under a signed CLA.

@petrex petrex marked this pull request as draft August 19, 2024 23:50
@petrex
Copy link
Contributor Author

petrex commented Sep 9, 2024

@pytorchbot rebase

@pytorchmergebot
Copy link
Collaborator

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

@pytorchmergebot
Copy link
Collaborator

Rebase failed due to Command git -C /home/runner/work/pytorch/pytorch rebase refs/remotes/origin/viable/strict pull/133934/head returned non-zero exit code 1

Rebasing (1/79)
Rebasing (2/79)
Rebasing (3/79)
Rebasing (4/79)
Rebasing (5/79)
Rebasing (6/79)
Rebasing (7/79)
Rebasing (8/79)
Rebasing (9/79)
Rebasing (10/79)
Rebasing (11/79)
Rebasing (12/79)
Rebasing (13/79)
Rebasing (14/79)
Rebasing (15/79)
Rebasing (16/79)
Rebasing (17/79)
Rebasing (18/79)
Rebasing (19/79)
Rebasing (20/79)
Rebasing (21/79)
Rebasing (22/79)
Rebasing (23/79)
Rebasing (24/79)
Rebasing (25/79)
Rebasing (26/79)
Rebasing (27/79)
Rebasing (28/79)
Rebasing (29/79)
Rebasing (30/79)
Rebasing (31/79)
Rebasing (32/79)
Rebasing (33/79)
Rebasing (34/79)
Rebasing (35/79)
Auto-merging test/test_sparse_semi_structured.py
CONFLICT (content): Merge conflict in test/test_sparse_semi_structured.py
error: could not apply 305106de63... add mi300x check, cusparselt/hipsparselt backend
hint: Resolve all conflicts manually, mark them as resolved with
hint: "git add/rm <conflicted_files>", then run "git rebase --continue".
hint: You can instead skip this commit: run "git rebase --skip".
hint: To abort and get back to the state before "git rebase", run "git rebase --abort".
hint: Disable this message with "git config advice.mergeConflict false"
Could not apply 305106de63... add mi300x check, cusparselt/hipsparselt backend

Raised by https://github.com/pytorch/pytorch/actions/runs/10781630866

@petrex
Update CUDA to HIP mappings for cusparseLt in torch/utils/hipify/cuda…
2209da1 
…_to_hip_mappings.py
@petrex
Update CUDA to HIP mappings for cusparseLt in torch/utils/hipify/cuda…
3aa1085 
…_to_hip_mappings.py
@petrex
Update CUDA to HIP mappings for cusparseLt in torch/utils/hipify/cuda…
a6d402c 
…_to_hip_mappings.py
@petrex
Update CUDA to HIP mappings for cusparseLt in torch/utils/hipify/cuda…
170dca7 
…_to_hip_mappings.py
@petrex
hipSparseLt support compute type HIPSPARSELT_COMPUTE_32F
e1fa334
@petrex
Add support for hipsparseLt compute types look-up
cbbb05e
@petrex
Update CUDA to HIP mappings for cusparseLt compute types
72e0735
@petrex
hipsparseLT does not support R_32F compute type
17122ea
@petrex
add USE_ROCM && ROCM_VERSION
e2fc842
@petrex
add check for R32_F data type
1cd6de1
@petrex
Update lookup table to retrieve the datatype
00ba49e
@petrex
linter
3dc5555
@petrex
fix hipspraseLtDatatype_t
bd591ae
@petrex
drop duplicate const
ca11930
@petrex
lint
970edc3
@petrex
fix typo
ecae4ba
@petrex
build hipsparselt by default
361e58b
@petrex
Add support for hipsparselt package if ROCM version is 6.1.0 or greater
c6c1699
@petrex
Fix duplicate call to find_package_and_print_version(hipsparselt) in …
22aafff 
…LoadHIP.cmake
@petrex
use compressed_A.device().index() for checking hipSparseLtArch support
2bbfd79
@petrex
Refactor hipSparseLtArch support check in cuSPARSELtOps.cpp
abae1f9
@petrex
fix computeType mapping/lib dependency/refactor
ff349e4
@petrex
disable hipsparseLt check for cuda
886546d
@petrex
drop ishipSparseLt from cuda build
2533191
@petrex
Refactor hipSparseLtArch support check in cuSPARSELtOps.cpp
7f01afc
@petrex
fix typo
c30eb03
@petrex
add mi300x check, cusparselt/hipsparselt backend
e0f727d
@petrex
hipsparselt: HIP backend only support HIPSPARSE_ORDER_COL
9dde1e4
@petrex
drop float32 from hipsparselt inference type
788ff19
@petrex
fix leading dimension
3ee740e
@petrex
Seperate result buffer for hipSparseLtMatMul()
49be454
@petrex
skip tests that are not supported in ROCm atm
eae3b2c
@petrex
stress CI
fb35929 
remove duplicate mapping
@petrex
update to support row_major_order and unified hip enum
802f968 
update support check for hipSparseLt

cusparseLtDenseDescriptorInit update for hipsparselt
@petrex
fix cusparselt/hipsparselt enable
264944e
@petrex
lint
7d6bba6
@petrex
fix typo function name
86c8849
@petrex
enable cutlass/ck path for rocm
389a7ff
@petrex
add composable kernel submodule
5eba581
@petrex
add ck layout
61391c5
@petrex
Add meta cpu ref / reorder cpu ref with CK wrapper
220b26f 
add reorder_meta function for CK wrapper

add check for ROCm platform

chore: Remove temporary function used for testing

add TODO and fix typo
@petrex
optimize hipsparselt check
319473d
@petrex
add a check for hipsparselt
8a84a66
@petrex
fix test util
24e12d2
@petrex
drop seperate res buffer (update for new hipsparselt)
ac130cb
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@eqy eqy Awaiting requested review from eqy eqy is a code owner

@syed-ahmed syed-ahmed Awaiting requested review from syed-ahmed syed-ahmed is a code owner

@jeffdaily jeffdaily Awaiting requested review from jeffdaily jeffdaily is a code owner

@jithunnair-amd jithunnair-amd Awaiting requested review from jithunnair-amd jithunnair-amd is a code owner

Assignees
No one assigned
Labels
module: rocm AMD GPU support for Pytorch open source release notes: sparse release notes category
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@petrex @pytorchmergebot @pytorchbot