[ET-VK][Ops] aten.convolution (Bias=False)#2887
Closed
junpi3 wants to merge 6 commits intogh/jorgep31415/47/basefrom
Closed
[ET-VK][Ops] aten.convolution (Bias=False)#2887junpi3 wants to merge 6 commits intogh/jorgep31415/47/basefrom
junpi3 wants to merge 6 commits intogh/jorgep31415/47/basefrom
Conversation
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/2887
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f2ca00c with merge base 99c4f4e ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
This was referenced Apr 5, 2024
junpi3
pushed a commit
that referenced
this pull request
Apr 5, 2024
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
ghstack-source-id: 221526247
Pull Request resolved: #2887
Closed
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
junpi3
pushed a commit
that referenced
this pull request
Apr 8, 2024
Pull Request resolved: #2887 The final touches to get ET-VK convolution on-par with ATen-VK's convolution. ## Idea In our shaders, we add the bias to our sum. ``` ${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0); ``` To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum. ## Issue If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`. ## Solution If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights, (2) allocate a StagingBuffer of that size, and (3) `memset` its data to zero. Failure to do (3) will result in undefined behavior. ghstack-source-id: 221721748 @exported-using-ghexport Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
SS-JIA
approved these changes
Apr 9, 2024
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
## Idea
In our shaders, we add the bias to our sum.
```
${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0);
```
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
## Issue
If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`.
## Solution
If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.
Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
[ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D55814589 |
Contributor
|
This pull request has been merged in 8aaf2c5. |
Closed
kedarnath03
pushed a commit
to kedarnath03/executorch
that referenced
this pull request
Jun 25, 2025
Pull Request resolved: pytorch/executorch#2887 The final touches to get ET-VK convolution on-par with ATen-VK's convolution. ## Idea In our shaders, we add the bias to our sum. ``` ${VEC4_T[DTYPE]} sum = texelFetch(bias_in, ivec2(pos.z, 0), 0); ``` To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum. ## Issue If `Bias=False`, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime as `TypeTag::NONE`, not `TypeTag::TENSORREF`. ## Solution If `TypeTag::NONE` is given, (1) create the `vTensor` using the `out_channels` value from the weights, (2) allocate a StagingBuffer of that size, and (3) `memset` its data to zero. Failure to do (3) will result in undefined behavior. ghstack-source-id: 221926167 @exported-using-ghexport Differential Revision: [D55814589](https://our.internmc.facebook.com/intern/diff/D55814589/)
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):
The final touches to get ET-VK convolution on-par with ATen-VK's convolution.
Idea
In our shaders, we add the bias to our sum.
To keep our shaders as is, we implement having no bias by allocating a buffer of zeros. Then, our shader adds zero to our sum.
Issue
If
Bias=False, dummy buffer of zeros is not serialized with the graph. The bias ValueRef is deserialized in the runtime asTypeTag::NONE, notTypeTag::TENSORREF.Solution
If
TypeTag::NONEis given, (1) create thevTensorusing theout_channelsvalue from the weights and (2) allocate a StagingBuffer of that size. The StagingBuffer will be transferred to GPU memory and initialized to zeros.Differential Revision: D55814589