style: rustfmt the four files main is currently unformatted in

[justinchuby]

What

cargo fmt --all --check fails on main (84a2765). Four files are unformatted:

file	why
crates/mlas-sys/src/lib.rs	QGEMM_PACKED_LIVE_BYTES.fetch_sub call wrapping (#1151)
crates/onnx-runtime-ep-cpu/src/kernels/governed_accumulator_budget.rs	assert! wrapping (#1151)
crates/onnx-runtime-ep-cpu/src/kernels/qlinear_matmul.rs	two call sites (#1133/#1151)
crates/onnx-runtime-ep-cpu/src/kernels/simd_activations.rs	trailing blank line (#1136/#1143)

Whitespace and call-site wrapping only — no token changes, no behaviour change.

Evidence

$ cargo fmt --all --check   # before: 4 files reported
$ cargo fmt --all
$ cargo fmt --all --check   # after: clean, exit 0

Split out of the no-defer PR so that one carries no unrelated churn. This is the
same shape as #1131 and #1144.

[github-actions]

🔴 Benchmark Regression Detected

Comparison of criterion micro-benchmarks: PR head vs merge-base, measured on the same runner in the same job (base first → PR second).

ℹ️ Absolute times are informational only — they vary with runner load. The % change column is the reliable signal because both sides ran under identical conditions.

Status	Scenario	Base	PR	Change
🔴	matmul/large_generic_f32_threads=8/32x1024x1024	5.26 ms	6.93 ms	+31.7%
⚠️	matmul/large_generic_bf16_threads=1/32x1024x1024	2.04 ms	2.54 ms	+24.3%
⚠️	matmul/large_generic_bf16_threads=8/32x1024x1024	1.90 ms	2.34 ms	+23.1%
⚠️	gather/large_bf16_threads=1-internal/131072	14.95 µs	17.97 µs	+20.2%
✅	block_quantized_matmul_cached_dense/mxfp4_preexpanded_dense_oncelock_like_proxy/1x1024x1024	95.87 µs	109.34 µs	+14.1%
✅	matmul/large_generic_f16_threads=8/32x1024x1024	102.05 µs	111.54 µs	+9.3%
✅	gather/large_f32_threads=1-internal/131072	36.58 µs	39.42 µs	+7.8%
✅	gather/large_f16_threads=1-internal/131072	13.01 µs	14.02 µs	+7.8%
✅	tokenization/encode_tokens_per_second	386.71 µs	415.83 µs	+7.5%
✅	qwen3_sampling_processors/top_k_partial_selection	143.30 µs	153.77 µs	+7.3%
✅	matmul/medium_generic_f32_threads=8/32x512x512	1.66 ms	1.78 ms	+7.2%
✅	reduce_mean/small_f32_threads=1-internal/4096	14.94 µs	15.87 µs	+6.2%
✅	matmul/medium_generic_f16_threads=1/32x512x512	36.11 µs	38.13 µs	+5.6%
✅	matmul/large_generic_f32_threads=1/32x1024x1024	9.55 ms	10.03 ms	+5.0%
✅	gather/small_f32_threads=1-internal/4096	653.6 ns	686.4 ns	+5.0%
✅	add/medium_bf16_threads=1-internal/262144	102.58 µs	106.30 µs	+3.6%
✅	add/medium_f16_threads=1-internal/262144	103.41 µs	107.12 µs	+3.6%
✅	tokenization/decode_tokens_per_second	6.35 ms	6.56 ms	+3.2%
✅	gather/small_f16_threads=1-internal/4096	482.4 ns	498.0 ns	+3.2%
✅	matmul/small_generic_f16_threads=1/1x256x256	36.87 µs	37.90 µs	+2.8%
✅	matmul/small_generic_f16_threads=8/1x256x256	36.78 µs	37.69 µs	+2.5%
✅	sampling_latency/min_p_per_token	208.29 µs	212.41 µs	+2.0%
✅	matmul/medium_generic_bf16_threads=8/32x512x512	633.03 µs	644.98 µs	+1.9%
✅	matmul/small_generic_f32_threads=8/1x256x256	49.37 µs	50.04 µs	+1.4%
✅	sampling_latency/top_k_per_token	52.98 µs	53.57 µs	+1.1%
✅	qwen3_sampling_processors/top_p_full_sort_after_top_k_baseline	3.57 ms	3.61 ms	+1.1%
✅	matmul/small_generic_bf16_threads=1/1x256x256	37.16 µs	37.50 µs	+0.9%
✅	qwen3_sampling_processors/top_k_full_sort_baseline	2.15 ms	2.16 ms	+0.7%
✅	matmul/medium_generic_f32_threads=1/32x512x512	2.35 ms	2.36 ms	+0.5%
✅	block_quantized_moe_cached_dense/mxfp4_cached_dense_expert_repeated_call/rows=1,H=256,I=256,E=4,top_k=1	174.73 µs	175.56 µs	+0.5%
✅	add/large_bf16_threads=1-internal/4194304	1.72 ms	1.73 ms	+0.4%
✅	block_quantized_moe_cached_dense/mxfp4_uncached_expert_dequant_each_call/rows=1,H=256,I=256,E=4,top_k=1	675.90 µs	677.34 µs	+0.2%
✅	qwen3_sampling_processors/top_k_top_p_full_sort_baseline	5.84 ms	5.85 ms	+0.2%
✅	add/small_f16_threads=1-internal/1024	461.0 ns	461.5 ns	+0.1%
✅	sampling_latency/top_p_per_token	399.01 µs	399.42 µs	+0.1%
✅	add/medium_f32_threads=1-internal/262144	25.03 µs	25.00 µs	-0.2%
✅	reduce_mean/large_f32_threads=1-internal/262144	1.00 ms	1.00 ms	-0.3%
✅	kv_cache/alloc_dealloc_pages	40.24 µs	40.08 µs	-0.4%
✅	add/large_f32_threads=1-internal/4194304	889.50 µs	884.58 µs	-0.6%
✅	matmul/small_generic_bf16_threads=8/1x256x256	38.85 µs	38.59 µs	-0.7%
✅	gather/medium_bf16_threads=1-internal/32768	2.40 µs	2.38 µs	-0.7%
✅	matmul/medium_generic_bf16_threads=1/32x512x512	541.37 µs	537.39 µs	-0.7%
✅	gather/medium_f16_threads=1-internal/32768	2.43 µs	2.41 µs	-0.8%
✅	add/small_bf16_threads=1-internal/1024	449.8 ns	445.1 ns	-1.0%
✅	sampling_latency/greedy_per_token	3.31 µs	3.27 µs	-1.2%
✅	grammar_masking/llguidance_compute_mask/32	79.65 µs	78.51 µs	-1.4%
✅	block_quantized_matmul_cached_dense/mxfp4_uncached_dequant_each_call/1x1024x1024	1.16 ms	1.14 ms	-2.0%
✅	gather/medium_f32_threads=1-internal/32768	4.08 µs	4.00 µs	-2.0%
✅	logit_processing/seven_processor_chain_per_step	332.34 µs	325.21 µs	-2.1%
✅	gather/small_bf16_threads=1-internal/4096	495.2 ns	484.5 ns	-2.2%
✅	qwen3_sampling_processors/top_p_fast_after_top_k	534.69 µs	521.89 µs	-2.4%
✅	qwen3_sampling_processors/top_k_top_p_fast	682.83 µs	663.99 µs	-2.8%
✅	reduce_mean/medium_f32_threads=1-internal/65536	257.22 µs	248.51 µs	-3.4%
✅	block_quantized_matmul_cached_dense/mxfp4_cached_dense_repeated_call/1x1024x1024	129.13 µs	124.16 µs	-3.8%
✅	matmul/small_generic_f32_threads=1/1x256x256	47.81 µs	45.72 µs	-4.4%
✅	matmul/large_generic_f16_threads=1/32x1024x1024	91.15 µs	86.99 µs	-4.6%
✅	matmul/medium_generic_f16_threads=8/32x512x512	43.76 µs	41.49 µs	-5.2%
✅	add/large_f16_threads=1-internal/4194304	1.83 ms	1.71 ms	-6.7%
🟢	add/small_f32_threads=1-internal/1024	252.7 ns	194.4 ns	-23.1%

Visual flags: ⚠️ ≥ 15% slower, 🔴 ≥ 30% slower — calibrated against measured runner noise (~27% worst-case on multi-threaded matmul)

Host info

CPU: Apple M1 (Virtual)
Cores: 3
OS: Darwin 25.5.0 arm64
Rust: rustc 1.97.1 (8bab26f4f 2026-07-14)
Load avg: { 3.31 3.08 3.67 }

What this cannot catch

• Regressions in code paths not covered by these benchmarks (e.g., end-to-end decode with a real model)
• Sub-threshold regressions that compound over multiple PRs
• Performance changes that only manifest under GPU execution
• Latency changes in the ORT integration path (these benchmarks exercise the native Rust kernels)

[justinchuby]

Superseded by #1172, which landed the same four files. Closing.

[codecov]

Codecov Report

❌ Patch coverage is 88.88889% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 79.84%. Comparing base (84a2765) to head (06b6417).
⚠️ Report is 13 commits behind head on main.

Files with missing lines	Patch %	Lines
.../onnx-runtime-ep-cpu/src/kernels/qlinear_matmul.rs	75.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1168   +/-   ##
=======================================
  Coverage   79.84%   79.84%           
=======================================
  Files         359      359           
  Lines      157243   157245    +2     
  Branches   157243   157245    +2     
=======================================
+ Hits       125549   125552    +3     
+ Misses      27136    27135    -1     
  Partials     4558     4558           

Flag	Coverage Δ
mlas	85.39% <100.00%> (+0.03%)	⬆️
offline	79.73% <83.33%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
crates/mlas-sys/src/lib.rs	84.78% <100.00%> (+<0.01%)	⬆️
...-ep-cpu/src/kernels/governed_accumulator_budget.rs	88.67% <100.00%> (+0.10%)	⬆️
...nnx-runtime-ep-cpu/src/kernels/simd_activations.rs	88.58% <ø> (ø)
.../onnx-runtime-ep-cpu/src/kernels/qlinear_matmul.rs	83.60% <75.00%> (ø)

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.