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 GPU-only operations in CudaTensor class #42

Merged
merged 41 commits into from Jul 7, 2021
Merged

Enable GPU-only operations in CudaTensor class #42

merged 41 commits into from Jul 7, 2021

Conversation

mlxd
Copy link
Member

@mlxd mlxd commented Jul 1, 2021

Context: This PR removes the intermediate transfers to the Tensor class for slicing, as well as provides Transpose functionality via the cuTENSOR library.

Description of the Change: Via calls to the cuTENSOR library we enable permutation of the tensor class for a given set of indices. Additionally, via this permutation, we enable tensor slicing.

Benefits: We can avoid intermediate GPU-CPU-GPU transfers to perform tensor slicing.

Possible Drawbacks: Debugging on-device code can be more challenging.

Related GitHub Issues:

mlxd and others added 28 commits June 15, 2021 10:46
@mlxd
Add support to load datafile as column major for CUDA
52a2468
@mlxd
Ensure Tensor and CudaTensor supported for serializer
2bebe07
@mlxd
Fix conversion between Tensor and CudaTensor indices and sizes
3cf9050
@mlxd
Rename TBCC to TBC
7818dee
@mlxd
Additional renaming for TBCC to TBC
3204acb
@mlxd
Renaming tbcc to tbc and avoiding datatype ambiguity in class
3d8e24d
@mlxd
Refactor to generalise methods and interface between between Tensor a…
8bca4fa 
…nd CudaTensor
@mlxd
Add tests foir CudaTensor backed TN
34800f7
@mlxd
Intermediate slicing and rehsping support via Tensor for CudaTensor
15834ac
@mlxd
Add initial slice and reshape support for CudaTensor
59b305a
@mlxd
Allow CudaTensor network contractions
8c991b9
@mlxd
Merge branch 'main' into tbc_cuda
a0d831a
@mlxd
Add support for incremental CuTensor naming convention
2c12afe
@mlxd
Enable CudaTensor single GPU support for TBC
5e8cb46
@mlxd
Add CuTensor AddTensor support for CudaTensor class
3a1b9ad
@mlxd
Remove redundant calls for cutensorElementwiseBinary
dfb95cd
@mlxd
Extend TBC tests to CudaTensor
cffb452
@mlxd
Ensure testing compliance for CudaTensor with TBC
5d6a4a1
@mlxd
Add private equality op for CudaTensor
3915f38
@mlxd
Merge branch 'main' into tbc_cuda
f1ca92f
@mlxd
Add safety check for CUDA and fix GPU memory leak
5f6155a
@mlxd
Fix CudaTensor formatting
e45bcb6
@mlxd
Add compile-time removal of CUDA safety checks
fbab9f4
@mlxd
Add prelim support for removing CPU intermediate ops for CudaTensor
d81147d
@mlxd
Merge branch 'main' into cudatensor_nocpu
fd521b4
@mlxd
Fix pointer offsets in slicing
119c850
@mlxd
Add tranpose-permute to CudaTensor
6cee0f2
@mlxd
Esnure CudaTensor ordering preserved after slicing
1427f5e
@github-actions
Copy link

github-actions bot commented Jul 1, 2021 •
edited

Test Report (C++) on Ubuntu

    1 files  ±0      1 suites  ±0   0s ⏱️ ±0s
521 tests ±0  521 ✔️ ±0  0 💤 ±0  0 ❌ ±0 
870 runs  ±0  870 ✔️ ±0  0 💤 ±0  0 ❌ ±0 

Results for commit b44fdf3. ± Comparison against base commit b44fdf3.

♻️ This comment has been updated with latest results.

@github-actions
Copy link

github-actions bot commented Jul 1, 2021 •
edited

Test Report (C++) on MacOS

    1 files  ±0      1 suites  ±0   0s ⏱️ ±0s
521 tests ±0  521 ✔️ ±0  0 💤 ±0  0 ❌ ±0 
870 runs  ±0  870 ✔️ ±0  0 💤 ±0  0 ❌ ±0 

Results for commit 6b80040. ± Comparison against base commit 29805e4.

♻️ This comment has been updated with latest results.

Mandrenkov
Mandrenkov reviewed Jul 5, 2021
View reviewed changes
Copy link
Collaborator

@Mandrenkov Mandrenkov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks great! I especially like the implementation of SliceIndex() in terms of Transpose().

It's also a nice bonus that we don't need to change any of the tests!

.github/CHANGELOG.md Outdated Show resolved Hide resolved
CMakeLists.txt Outdated Show resolved Hide resolved
include/jet/TensorNetwork.hpp Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Outdated Show resolved Hide resolved
include/jet/CudaTensor.hpp Show resolved Hide resolved
@github-actions
Copy link

github-actions bot commented Jul 7, 2021 •
edited

Test Report (Python) on Ubuntu

    1 files  ±0      1 suites  ±0   7s ⏱️ ±0s
490 tests ±0  490 ✔️ ±0  0 💤 ±0  0 ❌ ±0 

Results for commit b44fdf3. ± Comparison against base commit b44fdf3.

♻️ This comment has been updated with latest results.

Mandrenkov
Mandrenkov approved these changes Jul 7, 2021
View reviewed changes
Copy link
Collaborator

@Mandrenkov Mandrenkov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me!

.github/CHANGELOG.md Outdated Show resolved Hide resolved
@mlxd @Mandrenkov
Update .github/CHANGELOG.md
13f5255 
Co-authored-by: Mikhail Andrenkov <Mandrenkov@users.noreply.github.com>
@mlxd mlxd merged commit b44fdf3 into main Jul 7, 2021
@mlxd mlxd deleted the cudatensor_nocpu branch July 7, 2021 15:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@Mandrenkov Mandrenkov Mandrenkov approved these changes

@trevor-vincent trevor-vincent Awaiting requested review from trevor-vincent

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
2 participants
@mlxd @Mandrenkov