[ET-VK] Migrate to use new q8ta_conv2d ops#17388
Merged
meta-codesync[bot] merged 8 commits intogh/SS-JIA/419/basefrom Feb 13, 2026
Merged
[ET-VK] Migrate to use new q8ta_conv2d ops#17388meta-codesync[bot] merged 8 commits intogh/SS-JIA/419/basefrom
meta-codesync[bot] merged 8 commits intogh/SS-JIA/419/basefrom
Conversation
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/17388
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit 4544e52 with merge base dcfd12d ( NEW FAILURE - The following job has failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This was referenced Feb 11, 2026
This PR needs a
|
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
manuelcandales
approved these changes
Feb 12, 2026
added 3 commits
February 12, 2026 15:34
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/) [ghstack-poisoned]
meta-codesync
Bot
merged commit 469a75e
into
gh/SS-JIA/419/base
194 of 197 checks passed
SS-JIA
pushed a commit
that referenced
this pull request
Feb 13, 2026
Pull Request resolved: #17388 This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace ghstack-source-id: 340983076 @exported-using-ghexport Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/)
SS-JIA
pushed a commit
that referenced
this pull request
Feb 13, 2026
Pull Request resolved: #17388 This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace ghstack-source-id: 340983076 @exported-using-ghexport Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/)
SS-JIA
pushed a commit
that referenced
this pull request
Feb 13, 2026
Pull Request resolved: #17388 This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace ghstack-source-id: 340983076 @exported-using-ghexport Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/)
SS-JIA
pushed a commit
that referenced
this pull request
Feb 13, 2026
Pull Request resolved: #17388 This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace ghstack-source-id: 340983076 @exported-using-ghexport Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/)
chizkiyahu
pushed a commit
to chizkiyahu/executorch
that referenced
this pull request
Feb 23, 2026
Pull Request resolved: pytorch#17388 This renames the quantized convolution custom ops from the old `conv2d_q8ta_q8csw_q8to` naming convention to the shorter `q8ta_conv2d` convention, and moves operator registrations from the `etvk` namespace to the standard `et_vk` namespace. A new `q8ta_conv2d_pw` op is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation. Changes: - custom_ops_lib.py: Rename functions and op names; add q8ta_conv2d_pw definition - op_registry.py: Split registration into separate PW and non-PW entries - quantized_convolution.py: Add pointwise conv detection and routing to q8ta_conv2d_pw - Q8taConv2d.cpp, Q8taConv2dDW.cpp, Q8taConv2dPW.cpp, Q8taConv2dIm2Col.cpp: Change VK_REGISTER_OP from etvk to et_vk namespace - TestQ8taConv2d.cpp: Update operator name strings to et_vk namespace ghstack-source-id: 340983076 @exported-using-ghexport Differential Revision: [D93000168](https://our.internmc.facebook.com/intern/diff/D93000168/)
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
2 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):
This renames the quantized convolution custom ops from the old
conv2d_q8ta_q8csw_q8tonaming convention to the shorterq8ta_conv2dconvention, and moves operator registrations from theetvknamespace to the standardet_vknamespace. A newq8ta_conv2d_pwop is also introduced for pointwise (1x1 stride-1 no-padding) convolutions, allowing the pattern matcher to route these to a specialized implementation.Changes:
Differential Revision: D93000168