Skip to content

[ET-VK] Rearranging code in permute op shader to reduce heavy math ops and improve performance.#7014

Merged
facebook-github-bot merged 4 commits intogh/trivedivivek/13/basefrom
gh/trivedivivek/13/head
Nov 26, 2024
Merged

[ET-VK] Rearranging code in permute op shader to reduce heavy math ops and improve performance.#7014
facebook-github-bot merged 4 commits intogh/trivedivivek/13/basefrom
gh/trivedivivek/13/head

Conversation

@trviv
Copy link
Contributor

@trviv trviv commented Nov 21, 2024

Stack from ghstack (oldest at bottom):

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.

Differential Revision: D66174765

…s and improve performance.

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)

[ghstack-poisoned]
@pytorch-bot
Copy link

pytorch-bot bot commented Nov 21, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/7014

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

✅ No Failures

As of commit 8c171a0 with merge base 2967302 (image):
💚 Looks good so far! There are no failures yet. 💚

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

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Nov 21, 2024
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D66174765

…avy math ops and improve performance."

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)

[ghstack-poisoned]
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D66174765

…avy math ops and improve performance."

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)

[ghstack-poisoned]
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D66174765

…avy math ops and improve performance."

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)

[ghstack-poisoned]
trviv added a commit that referenced this pull request Nov 26, 2024
…s and improve performance.

Pull Request resolved: #7014

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.
ghstack-source-id: 255546339
@exported-using-ghexport

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D66174765

@trviv trviv added release tracker release notes: backends release notes: api Changes to public facing apis (any interfaces, pybinded runtime methods, etc.) and removed release tracker release notes: api Changes to public facing apis (any interfaces, pybinded runtime methods, etc.) labels Nov 26, 2024
@facebook-github-bot facebook-github-bot merged commit 005eb2e into gh/trivedivivek/13/base Nov 26, 2024
@facebook-github-bot facebook-github-bot deleted the gh/trivedivivek/13/head branch November 26, 2024 22:15
kirklandsign pushed a commit that referenced this pull request Nov 26, 2024
…s and improve performance. (#7095)

Pull Request resolved: #7014

The diff rearranges Permute op shader code in executorch vulkan backend to reduce heavy math operations and improve performance. The change also include adding an extension to support explicit arithmetic types and changing the data type of the position variable to u16vec3.
ghstack-source-id: 255546339
@exported-using-ghexport

Differential Revision: [D66174765](https://our.internmc.facebook.com/intern/diff/D66174765/)

Co-authored-by: Vivek Trivedi <5340687+trivedivivek@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. fb-exported

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants