[ExecuTorch][WebGPU] Fuse the SwiGLU elementwise chain

[JCNTH]

Stack from ghstack (oldest at bottom):

• [ExecuTorch][WebGPU] Lift FastVLM dispatch limits and harden graph construction #21602
• [ExecuTorch][WebGPU] Cover Conv1d routes and dynamic manifests #21601
• [ExecuTorch][WebGPU] Support general groups=1 Conv1d #21600
• [ExecuTorch][WebGPU] Test byte-packed BOOL tails #21599
• [ExecuTorch][WebGPU] Support byte-packed BOOL storage #21598
• [ExecuTorch][Vulkan] Preserve persistent buffer mutations #21597
• [ExecuTorch][WebGPU] Preserve Linear and Q4 embedding params across dynamic resize #21483
• [ExecuTorch][WebGPU] Generate logical and arithmetic binary shader variants #21451
• [ExecuTorch][WebGPU] Generate extrema and unary shader variants #21450
• [ExecuTorch][WebGPU] Generate typed to_copy shader variants #21449
• [ExecuTorch][WebGPU] Make WGSL generation rollback-safe #21448
• [ExecuTorch][WebGPU] Add Qwen3 K16 attention and fp16 KV cache #21136
• [ExecuTorch][WebGPU] Add HuggingFace rotate-half RoPE #21135
• [ExecuTorch][WebGPU] Repair dynamic SDPA routing, add attestation #21134
• [ExecuTorch][WebGPU] Select emdawnwebgpu for Emscripten builds #21133
• [ExecuTorch][WebGPU] Add Llama K16 online causal attention #21132
• [ExecuTorch][WebGPU] Replace broad QKV fusion with BK64 kernel #21131
• [ExecuTorch][WebGPU] Route exact prefill shapes to the BK64 GEMM #21130
• -> [ExecuTorch][WebGPU] Fuse the SwiGLU elementwise chain #21129
• [ExecuTorch][WebGPU] Coordinate dynamic dispatch routes #21128
• [ExecuTorch][WebGPU] Add certified scoped output suppression #21127

The SwiGLU feed-forward activation lowered to separate gate and up projections
followed by a sigmoid and two elementwise multiplies, launching several
dispatches over the same tensors. This fuses the sigmoid and the two multiplies
into a single dynamically-resized 2D dispatch that computes
(gate * sigmoid(gate)) * up in one pass. Fusion applies only when the pattern
is exact and its scratch is planner-owned; graph outputs, extra consumers, and
separate inputs fall back to the unfused chain, and the landed broad-QKV
behavior is left untouched until its dedicated corrective diff. No Vulkan
analogue: the Vulkan backend has no fused SiLU/SwiGLU shader and composes it
from a sigmoid unary op plus an elementwise multiply at graph level.

Key changes:

• runtime/ops/mul/silu_mul_fused.wgsl (+ generated header): one-thread-per-
  element SiLU-gated multiply of gate and up into a single output.
• WebGPUGraph.cpp: detect the exact gate/up/sigmoid/mul pattern, own the gate
  scratch, size the 2D dispatch, and fall back conservatively.
  @exported-using-ghexport

Differential Revision: D113171742

Differential Revision: D113171742

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21129

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 9455347 with merge base ad3a71f ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[SS-JIA]

Review automatically exported from Phabricator review in Meta.