[ExecuTorch][WebGPU] Add update_cache op (llama.update_cache)#20202
Merged
Conversation
Pull Request resolved: #20083 Add `llama.update_cache.default`: an in-place KV-cache write. The shader scatters the new K/V (`[1,S,H,D]`) into the cache (`[1,Cmax,H,D]`) at `dst_offset = input_pos*n_heads*head_dim`, bounds-checked against the cache size. The handler validates shape (batch==1, matching n_heads/head_dim) and sizes the 1D dispatch from the device limit via `WebGPUUtils` before allocating. Mirrors the Vulkan `sdpa_kv_cache_update` reference. The export/delegation test is the follow-up diff stacked directly above. Authored with assistance from Claude. ghstack-source-id: 392019030 @exported-using-ghexport Differential Revision: [D107547308](https://our.internmc.facebook.com/intern/diff/D107547308/)
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20202
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
JulianCloudNTH
approved these changes
Jun 10, 2026
Pull Request resolved: #20084 Tests for `llama.update_cache.default`, stacked on the op diff below. `test/ops/sdpa/test_update_cache.py` lowers the op through `VulkanPartitioner` (asserting it delegates to VulkanBackend) and exports per-case `.pte`s; `test/native/test_update_cache.cpp` runs them on-GPU and checks an integer-exact scatter golden against the returned cache. Coverage mirrors the Vulkan KV-cache test (`VulkanSDPATest`): single-shot writes at varied shapes/offsets, plus a multi-step advancing-input_pos replay that threads the returned cache across steps over the same GQA param sets (incl. llama3 head_dim=128). Comparing the cache directly is stronger than Vulkan, which checks it only indirectly via the SDPA output. Authored with assistance from Claude. ghstack-source-id: 391979582 @exported-using-ghexport Differential Revision: [D107547307](https://our.internmc.facebook.com/intern/diff/D107547307/)
…ymint Pull Request resolved: #20085 Adds the dynamic-scalar (SymInt) mechanism to the WebGPU graph as a standalone enabler, ahead of the SDPA op that consumes it. Mirrors the Vulkan delegate's SymInt = live uniform-buffer design: a `ValueType::SymInt` backed by a 16-byte `Uniform|CopyDst` buffer, `set_symint`/`read_symint`/`symint_buffer` accessors with dirty-tracking, a `SymIntSource` + `add_symint_source`/`update_symints_from_inputs` host-read path, and an `add_resize_hook`/`propagate_resize`/`dispatch_at` recompute plumbing. `WebGPUBackend::execute` calls `propagate_resize` after refreshing the SymInts from the runtime inputs. The `et_vk.select_as_symint` op handler records `out SymInt = x[index]` along a dim at build time. This diff has no in-graph consumer yet — the SDPA op (stacked above) reads the SymInt value via `read_symint()` for dynamic `input_pos`. Building it as its own diff keeps the enabler separate from the op, matching the update_cache → mechanism → SDPA layering. Authored with assistance from Claude. ghstack-source-id: 391979584 @exported-using-ghexport Differential Revision: [D107584280](https://our.internmc.facebook.com/intern/diff/D107584280/)
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.
This PR was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #20083 by @JulianCloudNTH
^ 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/JulianCloudNTH/16/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/JulianCloudNTH/16/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/main
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/JulianCloudNTH/16/orig
@diff-train-skip-merge