WebGPU: Support Split-K with batch size > 1

[Jiawei-Shao]

Description

This patch adds the support of Split-K with batch size > 1 by
encoding both batch index and Split-K index in dispatch_z and
decompose them in the shader via:
batch = logical_global_id.z / num_k_splits
split_index = logical_global_id.z % num_k_splits

This patch also adds batch size to the criteria of using Split-K
as increasing batch size will also increasing the parallelism,
reducing the effectiveness of Split-K.

This patch also replaces consteval with constexpr in
ort_version_check.h to workaround a compilation error
about vs2022.

Motivation and Context

With this patch we can improve the performance of
sam-vit-b-decoder-static-fp16-demo (7.5%) on Intel PTL.

[Copilot]

Pull request overview

Adds WebGPU Split‑K support for MatMul/Conv|MatMul when batch_size > 1 by packing (batch, split_index) into dispatch_z, and updates Split‑K heuristics to consider batch size (plus a small MSVC workaround in version checking).

Changes:

• Encode both batch index and Split‑K index into dispatch_z and decode in WGSL using splits_per_batch.
• Add a batch-size threshold and incorporate batch size into Split‑K gating heuristics.
• Add new unit tests intended to exercise Split‑K with batched inputs; replace consteval with constexpr in version validation helpers.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
onnxruntime/core/providers/webgpu/math/gemm_utils.cc	WGSL Split‑K indexing updated to decode batch/split from logical_global_id.z.
onnxruntime/core/providers/webgpu/math/gemm_packed.cc	Minor callsite update (comment/arg annotation) for MakeMatMulPackedVec4Source.
onnxruntime/core/providers/webgpu/math/matmul.cc	Compute-side Split‑K dispatch now encodes (batch * splits_per_batch) and passes splits_per_batch as a uniform; batched fill-bias/zero init.
onnxruntime/core/providers/webgpu/math/matmul.h	Extend fill-bias/zero program factory to accept batch_size.
onnxruntime/core/providers/webgpu/math/matmul_packed.h	Add splits_per_batch and batch_size uniforms to relevant programs.
onnxruntime/core/providers/webgpu/math/matmul_packed.cc	Fill-bias/zero shader updated to initialize outputs across all batches.
onnxruntime/core/providers/webgpu/webgpu_utils.h	Add max_batch_size_ to Split‑K config.
onnxruntime/core/providers/webgpu/webgpu_utils.cc	Split‑K gating updated to include batch size and enforce a max batch threshold.
onnxruntime/core/session/ort_version_check.h	Switch consteval to constexpr to work around VS2022/MSVC issues.
onnxruntime/test/providers/cpu/math/matmul_test.cc	Add a batched MatMul test intended to hit the Split‑K path.
onnxruntime/test/providers/cpu/nn/conv_op_test.cc	Add batched Conv2D “matmul-like” tests for Split‑K (with/without bias).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[qjia7]

LGTM with one nit.