[ET-VK][ez] Fuse update_cache + custom_sdpa into sdpa_with_kv_cache#15645
[ET-VK][ez] Fuse update_cache + custom_sdpa into sdpa_with_kv_cache#15645
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15645
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 132 PendingAs of commit 97b86a0 with merge base e938fea ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Pull Request resolved: #15618 SDPA used to be handled by a custom op `sdpa_with_kv_cache`, but it was eventually split (D62301837) into update_cache and custom_sdpa ops. However, having a single fused op is useful for Vulkan since it allows more control over how the cache tensors are stored and represented. Essentially, it makes it easier to manage the cache tensors and opens up opportunities for future optimizations. This diff introduces a fusion pass that does 2 things: 1. Combine update_cache and custom_sdpa back into sdpa_with_kv_cache 2. Ensure all references to the cache_pos symint use the same node - this prevents the select_at_dim_as_symint op from being called every time it is used. ghstack-source-id: 321258710 @exported-using-ghexport Differential Revision: [D86340339](https://our.internmc.facebook.com/intern/diff/D86340339/)
12e48b2 to
97b86a0
Compare
This PR needs a
|
…ytorch#15645) SDPA used to be handled by a custom op `sdpa_with_kv_cache`, but it was eventually split (D62301837) into update_cache and custom_sdpa ops. However, having a single fused op is useful for Vulkan since it allows more control over how the cache tensors are stored and represented. Essentially, it makes it easier to manage the cache tensors and opens up opportunities for future optimizations. This diff introduces a fusion pass that does 2 things: 1. Combine update_cache and custom_sdpa back into sdpa_with_kv_cache 2. Ensure all references to the cache_pos symint use the same node - this prevents the select_at_dim_as_symint op from being called every time it is used. Differential Revision: [D86340339](https://our.internmc.facebook.com/intern/diff/D86340339/)
…ytorch#15645) SDPA used to be handled by a custom op `sdpa_with_kv_cache`, but it was eventually split (D62301837) into update_cache and custom_sdpa ops. However, having a single fused op is useful for Vulkan since it allows more control over how the cache tensors are stored and represented. Essentially, it makes it easier to manage the cache tensors and opens up opportunities for future optimizations. This diff introduces a fusion pass that does 2 things: 1. Combine update_cache and custom_sdpa back into sdpa_with_kv_cache 2. Ensure all references to the cache_pos symint use the same node - this prevents the select_at_dim_as_symint op from being called every time it is used. Differential Revision: [D86340339](https://our.internmc.facebook.com/intern/diff/D86340339/)
This PR was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #15618 by @SS-JIA
^ Please use this as the source of truth for the PR details, comments, and reviews
ghstack PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/354/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/354/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/353/orig
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/354/orig
Differential Revision: D86340339
@diff-train-skip-merge