[ET-VK] Fake u16vecn for devserver#6675
Merged
facebook-github-bot merged 3 commits intogh/SS-JIA/138/basefrom Nov 7, 2024
Merged
[ET-VK] Fake u16vecn for devserver#6675facebook-github-bot merged 3 commits intogh/SS-JIA/138/basefrom
facebook-github-bot merged 3 commits intogh/SS-JIA/138/basefrom
Conversation
## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/6675
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 481b41e with merge base 03b1ef2 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
SS-JIA
added a commit
that referenced
this pull request
Nov 5, 2024
## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/) ghstack-source-id: 251970874 Pull Request resolved: #6675
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65501674 |
junpi3
approved these changes
Nov 5, 2024
## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/) [ghstack-poisoned]
SS-JIA
added a commit
that referenced
this pull request
Nov 5, 2024
Pull Request resolved: #6675 ## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. ghstack-source-id: 251990621 @exported-using-ghexport Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/)
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65501674 |
## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/) [ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65501674 |
SS-JIA
added a commit
that referenced
this pull request
Nov 7, 2024
Pull Request resolved: #6675 ## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. ghstack-source-id: 252234981 @exported-using-ghexport Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/)
SS-JIA
added a commit
that referenced
this pull request
Nov 7, 2024
Pull Request resolved: #6675 ## Context Copy-pasted from the newly added `maybe_fake_u16vec3` function in the codegen script: > There is a latency benefit to using u16vecn variables to store texture position variables instead of ivecn, likely due to reduced register pressure. However, SwiftShader does not support 16 bit integer types in shaders, so this is a crude way to fallback to using ivecn to store texture positions so that testing with SwiftShader is still possible. ghstack-source-id: 252234981 @exported-using-ghexport Differential Revision: [D65501674](https://our.internmc.facebook.com/intern/diff/D65501674/) Co-authored-by: Stephen Jia <ssjia@meta.com>
SS-JIA
added a commit
that referenced
this pull request
Nov 12, 2024
As title. This is a follow up from #6675. Replace `uint16_t` with `int` as well as 16 bit vectorized types. Differential Revision: [D65824204](https://our.internmc.facebook.com/intern/diff/D65824204/) [ghstack-poisoned]
SS-JIA
added a commit
that referenced
this pull request
Nov 12, 2024
As title. This is a follow up from #6675. Replace `uint16_t` with `int` as well as 16 bit vectorized types. Differential Revision: [D65824204](https://our.internmc.facebook.com/intern/diff/D65824204/) ghstack-source-id: 253112193 Pull Request resolved: #6781
SS-JIA
added a commit
that referenced
this pull request
Nov 12, 2024
As title. This is a follow up from #6675. Replace `uint16_t` with `int` as well as 16 bit vectorized types. Differential Revision: [D65824204](https://our.internmc.facebook.com/intern/diff/D65824204/) ghstack-source-id: 253112193 Pull Request resolved: #6781 Co-authored-by: Stephen Jia <ssjia@meta.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):
Context
Copy-pasted from the newly added
maybe_fake_u16vec3function in the codegen script:Differential Revision: D65501674