[Vulkan] Enable Codegen ShaderInfo Registry from GLSLT + Params YAML files (conv2d_pw) #91916
Conversation
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
…files (conv2d_pw)
@bypass-github-export-checks
This diff allows for adding entries to the shader registry by specifying which op names and registry keys should map to a template codegen Shader in the codegen Shader's glslt and params yaml files.
This can be done by
- adding a REGISTER_FOR entry which maps to either a tuple of (op name, list of registry keys) or null to the YAML file, and
- adding a ```REGISTER_FOR = $REGISTER_FOR``` line to the ShaderInfo comment in the glslt file
Ex.
YAML File:
```
conv2d_pw:
parameter_names_with_default_values:
...
REGISTER_FOR:
- !!python/tuple ["conv2d_pw", ["catchall"]]
parameter_values:
- ...
REGISTER_FOR: null
```
GLSLT File:
```
...
* REGISTER_FOR = $REGISTER_FOR
...
```
This diff also registers the conv2d_pw_2x2 Shader under ```'conv2d_pw → 'catchall'``` in the registry and uses ```VK_REGISTRY_KERNEL``` to retrieve the shader by look up in the registry
The shader registry generated in spv.cpp now looks like
```
ShaderRegistry shader_registry = {
{"conv2d", {{"catchall", "conv2d"}}},
{"conv2d_pw", {{"catchall", "conv2d_pw_2x2"}}}};
```
and the generated conv2d_p2_KxK.glsl files look like:
K=1
```
...
/*
* TILE_SIZE = (1, 1, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = None
*/
...
```
K=2
```
...
/*
* TILE_SIZE = (2, 2, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = ('conv2d_pw', ['catchall'])
*/
...
```
Differential Revision: [D42198560](https://our.internmc.facebook.com/intern/diff/D42198560/)
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/91916
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 2 FailuresAs of commit 32b7714: NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Jan 9, 2023
mcr229
approved these changes
Jan 9, 2023
…arams YAML files (conv2d_pw)"
bypass-github-export-checks
This diff allows for adding entries to the shader registry by specifying which op names and registry keys should map to a template codegen Shader in the codegen Shader's glslt and params yaml files.
This can be done by
- adding a REGISTER_FOR entry which maps to either a tuple of (op name, list of registry keys) or null to the YAML file, and
- adding a ```REGISTER_FOR = $REGISTER_FOR``` line to the ShaderInfo comment in the glslt file
Ex.
YAML File:
```
conv2d_pw:
parameter_names_with_default_values:
...
REGISTER_FOR:
- !!python/tuple ["conv2d_pw", ["catchall"]]
parameter_values:
- ...
REGISTER_FOR: null
```
GLSLT File:
```
...
* REGISTER_FOR = $REGISTER_FOR
...
```
This diff also registers the conv2d_pw_2x2 Shader under ```'conv2d_pw → 'catchall'``` in the registry and uses ```VK_REGISTRY_KERNEL``` to retrieve the shader by look up in the registry
The shader registry generated in spv.cpp now looks like
```
ShaderRegistry shader_registry = {
{"conv2d", {{"catchall", "conv2d"}}},
{"conv2d_pw", {{"catchall", "conv2d_pw_2x2"}}}};
```
and the generated conv2d_p2_KxK.glsl files look like:
K=1
```
...
/*
* TILE_SIZE = (1, 1, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = None
*/
...
```
K=2
```
...
/*
* TILE_SIZE = (2, 2, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = ('conv2d_pw', ['catchall'])
*/
...
```
Differential Revision: [D42198560](https://our.internmc.facebook.com/intern/diff/D42198560/)
[ghstack-poisoned]
…arams YAML files (conv2d_pw)"
bypass-github-export-checks
This diff allows for adding entries to the shader registry by specifying which op names and registry keys should map to a template codegen Shader in the codegen Shader's glslt and params yaml files.
This can be done by
- adding a REGISTER_FOR entry which maps to either a tuple of (op name, list of registry keys) or null to the YAML file, and
- adding a ```REGISTER_FOR = $REGISTER_FOR``` line to the ShaderInfo comment in the glslt file
Ex.
YAML File:
```
conv2d_pw:
parameter_names_with_default_values:
...
REGISTER_FOR:
- !!python/tuple ["conv2d_pw", ["catchall"]]
parameter_values:
- ...
REGISTER_FOR: null
```
GLSLT File:
```
...
* REGISTER_FOR = $REGISTER_FOR
...
```
This diff also registers the conv2d_pw_2x2 Shader under ```'conv2d_pw → 'catchall'``` in the registry and uses ```VK_REGISTRY_KERNEL``` to retrieve the shader by look up in the registry
The shader registry generated in spv.cpp now looks like
```
ShaderRegistry shader_registry = {
{"conv2d", {{"catchall", "conv2d"}}},
{"conv2d_pw", {{"catchall", "conv2d_pw_2x2"}}}};
```
and the generated conv2d_p2_KxK.glsl files look like:
K=1
```
...
/*
* TILE_SIZE = (1, 1, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = None
*/
...
```
K=2
```
...
/*
* TILE_SIZE = (2, 2, 1)
* WEIGHT_STORAGE = TEXTURE_2D
* WEIGHT_STORAGE_LAYOUT = OC4,IC4,4ic,4oc
* BIAS_STORAGE = TEXTURE_2D
* REGISTER_FOR = ('conv2d_pw', ['catchall'])
*/
...
```
Differential Revision: [D42198560](https://our.internmc.facebook.com/intern/diff/D42198560/)
[ghstack-poisoned]
|
@pytorchbot merge -f 'Landed internally' (Initiating merge automatically since Phabricator Diff has merged, using force because this PR might not pass merge_rules.json but landed internally) |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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):
@bypass-github-export-checks
This diff allows for adding entries to the shader registry by specifying which op names and registry keys should map to a template codegen Shader in the codegen Shader's glslt and params yaml files.
This can be done by
REGISTER_FOR = $REGISTER_FORline to the ShaderInfo comment in the glslt fileEx.
YAML File:
GLSLT File:
This diff also registers the conv2d_pw_2x2 Shader under
'conv2d_pw → 'catchall'in the registry and usesVK_REGISTRY_KERNELto retrieve the shader by look up in the registryThe shader registry generated in spv.cpp now looks like
and the generated conv2d_p2_KxK.glsl files look like:
K=1
K=2
Differential Revision: D42198560