Split SDPA + KV cache operator into SDPA operator and KV cache update operator + Add RemoveAsserts pass and apply it during LlaMa export
#8075
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/8075
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 1 PendingAs of commit c2ac929 with merge base 2a1a583 ( 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: D68922404 |
SS-JIA
added a commit
to SS-JIA/executorch-1
that referenced
this pull request
Jan 30, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
8bf5e92 to
f171f94
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
SS-JIA
added a commit
to SS-JIA/executorch-1
that referenced
this pull request
Jan 30, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
f171f94 to
6966c73
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
SS-JIA
added a commit
to SS-JIA/executorch-1
that referenced
this pull request
Jan 31, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Reviewed By: kimishpatel Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
6966c73 to
45894c7
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
SS-JIA
added a commit
to SS-JIA/executorch-1
that referenced
this pull request
Jan 31, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Reviewed By: kimishpatel Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
45894c7 to
95ffcdc
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
SS-JIA
added a commit
to SS-JIA/executorch-1
that referenced
this pull request
Jan 31, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Reviewed By: kimishpatel Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
95ffcdc to
51e195d
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
digantdesai
approved these changes
Jan 31, 2025
… operator + Add `RemoveAsserts` pass and apply it during LlaMa export (pytorch#8075) Summary: **Note**: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with `ghexport`, which was having some problems exporting the second diff, as well as the fact that both diffs are needed for `export_llama` to work so it makes more sense to just have a single diff. ## Context Recent changes split the `sdpa_with_kv_cache` operator into two separate operators, `update_cache` and `custom_sdpa` to decouple the cache update step from the actual SDPA computation. As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split `sdpa_with_kv_cache` into two operators. Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned. ## Context Recently, some assertion ops were added to the Llama source code. Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported. To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export. Reviewed By: kimishpatel, digantdesai Differential Revision: D68922404
SS-JIA
force-pushed
the
export-D68922404
branch
from
51e195d to
c2ac929
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D68922404 |
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.
Summary:
Note: This diff is a combination of D68919676 and D68919678. I decided to combine the two because of problems with
ghexport, which was having some problems exporting the second diff, as well as the fact that both diffs are needed forexport_llamato work so it makes more sense to just have a single diff.Context
Recent changes split the
sdpa_with_kv_cacheoperator into two separate operators,update_cacheandcustom_sdpato decouple the cache update step from the actual SDPA computation.As a result, SDPA is no longer being delegated on Vulkan because of this interface change. To rectify this, Vulkan must also split
sdpa_with_kv_cacheinto two operators.Note that during this diff the new operators are not partitioned yet because of complications caused by assertion ops in the graph. The next diff adds a pass to remove such assertion ops which allows the new operators to be partitioned.
Context
Recently, some assertion ops were added to the Llama source code.
Unfortunately, this causes issues for the Vulkan delegate because runtime assertions are not yet supported in Vulkan and the assertion ops cause graph breaks due to not being supported.
To prevent graph breaks when delegating to Vulkan, apply a pass to remove assertion ops during the llama export.
Differential Revision: D68922404