[justinchuby]

Context

PR #762 merged into main at e9c0ab6a4 and already carries the fix for the
CI-wide compile break introduced by #807: crates/onnx-runtime-cuda-memory/src/lib.rs
gates pub mod test_support; on #[cfg(feature = "gpu-tests")] only (previously
#[cfg(any(test, feature = "gpu-tests"))]), and the one integration test that
uses TestStream (tests/vmm_kv_layout_residency_gpu.rs) cfg-gates its import
and body accordingly.

Root cause recap: cfg(test) is only set when a crate compiles as its own
unit-test binary. Integration tests under tests/*.rs link the crate as a
regular (non---test) dependency, so cfg(test) never applies to them. The
old gate any(test, feature = "gpu-tests") therefore never actually exposed
test_support to tests/vmm_kv_layout_residency_gpu.rs unless gpu-tests was
passed explicitly — which is exactly what broke CI's honesty-checker base
(cuda-only, no gpu-tests) build for 7 of 9 failed jobs on #762's own CI run.

What this PR does

Since the functional fix already reached main via #762, this PR is a
minimal, standalone doc correction: test_support.rs's module doc still said
the module was compiled "behind the gpu-tests feature (and the crate's own
cfg(test))" — stale wording that contradicts the actual (correct) gate in
lib.rs and could mislead a future reader into re-introducing the same bug by
assuming cfg(test) alone helps here. Updated the doc to explicitly state the
gate is gpu-tests-only and explain why cfg(test) doesn't apply to
integration tests.

No functional/behavioral change. No .squad/ state included.

Delta vs. Opus-approved fix (afbd2c1 on #762)

git diff afbd2c131 origin/main -- crates/onnx-runtime-cuda-memory/src/lib.rs crates/onnx-runtime-cuda-memory/tests/vmm_kv_layout_residency_gpu.rs
is empty — main already has byte-identical content for both files via the
#762 merge. The only delta introduced by this branch beyond what Opus already
reviewed is the test_support.rs doc-comment correction (7 insertions / 3
deletions), which was never part of the reviewed diff. Flagging this so a tiny
follow-up look is optional but available before merge.

Validation (local, CPU-only host, matching CI's commands)

• cargo test --locked -p onnx-runtime-cuda-memory --features cuda --tests --no-run — compiles clean (base/no-gpu-tests config)
• cargo test --locked -p onnx-runtime-cuda-memory --features cuda,gpu-tests --tests --no-run — compiles clean (gpu-tests config)
• python .github/scripts/verify_cuda_test_honesty.py — passes: 416 tests/66 targets identical in both configs (416 ignored without gpu-tests; 412 fail-loud/4 ignored/0 passed with gpu-tests on this no-CUDA host)
• cargo clippy --locked -p onnx-runtime-cuda-memory --features cuda --all-targets -- -D warnings — clean
• cargo clippy --locked -p onnx-runtime-cuda-memory --features cuda,gpu-tests --all-targets -- -D warnings — clean
• cargo check --locked -p onnx-runtime-ep-cuda -p onnx-runtime-python --features onnx-runtime-python/cuda — clean
• cargo test --locked -p onnx-runtime-ep-cuda --features cuda --test capture_sync_contract — 1 passed
• cargo clippy -p onnx-runtime-ep-cuda --features cuda -- -D warnings — clean
• cargo fmt --all -- --check — clean

Notes

• EP plugin export: run our Rust execution providers inside upstream ONNX Runtime #762 (branch squad/ep-plugin-parity-cuda) temporarily carried this same
  cfg-gating fix directly on its own branch (to unblock its own CI) and has
  now merged into main, so main's CI should already be green going forward.
  This standalone branch exists per explicit request so the repo-wide CI
  unblock has its own paper trail, separate from EP plugin export: run our Rust execution providers inside upstream ONNX Runtime #762's larger diff.
• Kept scope minimal: no CI workflow changes, no other files touched.
• Leaving as Draft until CI is confirmed green on this PR's own run.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.47%. Comparing base (e9c0ab6) to head (dc00350).
⚠️ Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #818      +/-   ##
==========================================
- Coverage   78.20%   77.47%   -0.74%     
==========================================
  Files         350      350              
  Lines      141311   141311              
  Branches   141311   141311              
==========================================
- Hits       110519   109475    -1044     
- Misses      26389    27437    +1048     
+ Partials     4403     4399       -4     

Flag	Coverage Δ
cli-ort-linux	83.50% <ø> (ø)
cli-ort-windows	82.99% <ø> (-0.10%)	⬇️
mlas	80.92% <ø> (-0.42%)	⬇️
offline	77.17% <ø> (-0.78%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.
see 9 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[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
🔴	gather/large_f32_threads=1-internal/131072	25.69 µs	43.67 µs	+70.0%
🔴	matmul/small_generic_f32_threads=8/1x256x256	34.80 µs	55.28 µs	+58.8%
🔴	matmul/small_generic_f16_threads=1/1x256x256	31.72 µs	45.90 µs	+44.7%
🔴	gather/large_bf16_threads=1-internal/131072	11.13 µs	15.55 µs	+39.7%
🔴	matmul/small_generic_f16_threads=8/1x256x256	30.88 µs	42.21 µs	+36.7%
🔴	matmul/small_generic_f32_threads=1/1x256x256	37.85 µs	50.27 µs	+32.8%
⚠️	gather/medium_f32_threads=1-internal/32768	3.73 µs	4.74 µs	+27.2%
⚠️	gather/large_f16_threads=1-internal/131072	11.94 µs	15.04 µs	+26.0%
⚠️	matmul/small_generic_bf16_threads=1/1x256x256	33.22 µs	41.24 µs	+24.2%
⚠️	gather/medium_bf16_threads=1-internal/32768	2.43 µs	2.99 µs	+23.1%
⚠️	gather/medium_f16_threads=1-internal/32768	2.44 µs	3.00 µs	+22.9%
⚠️	gather/small_f16_threads=1-internal/4096	482.6 ns	574.3 ns	+19.0%
⚠️	add/medium_f16_threads=1-internal/262144	2.54 ms	2.95 ms	+16.1%
⚠️	gather/small_bf16_threads=1-internal/4096	479.7 ns	554.6 ns	+15.6%
⚠️	add/large_f16_threads=1-internal/4194304	41.43 ms	47.71 ms	+15.2%
✅	add/medium_bf16_threads=1-internal/262144	2.61 ms	3.00 ms	+14.9%
✅	add/medium_f32_threads=1-internal/262144	2.51 ms	2.87 ms	+14.4%
✅	logit_processing/seven_processor_chain_per_step	347.05 µs	395.15 µs	+13.9%
✅	reduce_mean/large_f32_threads=1-internal/262144	1.00 ms	1.13 ms	+12.9%
✅	matmul/small_generic_bf16_threads=8/1x256x256	36.33 µs	40.62 µs	+11.8%
✅	add/large_f32_threads=1-internal/4194304	42.59 ms	47.26 ms	+11.0%
✅	gather/small_f32_threads=1-internal/4096	678.8 ns	742.1 ns	+9.3%
✅	matmul/large_generic_f32_threads=1/32x1024x1024	10.48 ms	11.21 ms	+6.9%
✅	add/large_bf16_threads=1-internal/4194304	42.70 ms	45.42 ms	+6.4%
✅	kv_cache/alloc_dealloc_pages	42.72 µs	44.45 µs	+4.1%
✅	reduce_mean/small_f32_threads=1-internal/4096	15.49 µs	16.00 µs	+3.2%
✅	block_quantized_moe_cached_dense/mxfp4_cached_dense_expert_repeated_call/rows=1,H=256,I=256,E=4,top_k=1	82.57 µs	85.04 µs	+3.0%
✅	qwen3_sampling_processors/top_k_top_p_fast	697.05 µs	713.16 µs	+2.3%
✅	add/small_bf16_threads=1-internal/1024	13.28 µs	13.57 µs	+2.2%
✅	matmul/medium_generic_bf16_threads=1/32x512x512	568.35 µs	573.01 µs	+0.8%
✅	matmul/medium_generic_f32_threads=1/32x512x512	2.42 ms	2.43 ms	+0.7%
✅	grammar_masking/llguidance_compute_mask/32	87.48 µs	87.84 µs	+0.4%
✅	add/small_f16_threads=1-internal/1024	13.47 µs	13.48 µs	+0.0%
✅	matmul/medium_generic_f16_threads=1/32x512x512	32.94 µs	32.62 µs	-1.0%
✅	qwen3_sampling_processors/top_p_fast_after_top_k	583.24 µs	576.76 µs	-1.1%
✅	sampling_latency/top_p_per_token	421.72 µs	414.34 µs	-1.8%
✅	reduce_mean/medium_f32_threads=1-internal/65536	278.03 µs	272.77 µs	-1.9%
✅	qwen3_sampling_processors/top_k_partial_selection	165.23 µs	161.17 µs	-2.5%
✅	add/small_f32_threads=1-internal/1024	213.1 ns	206.9 ns	-2.9%
✅	qwen3_sampling_processors/top_p_full_sort_after_top_k_baseline	4.00 ms	3.87 ms	-3.2%
✅	matmul/medium_generic_f32_threads=8/32x512x512	1.04 ms	992.68 µs	-4.6%
✅	matmul/medium_generic_f16_threads=8/32x512x512	33.02 µs	31.36 µs	-5.0%
✅	sampling_latency/top_k_per_token	58.81 µs	55.18 µs	-6.2%
✅	qwen3_sampling_processors/top_k_top_p_full_sort_baseline	6.66 ms	6.16 ms	-7.5%
✅	sampling_latency/min_p_per_token	227.09 µs	208.71 µs	-8.1%
✅	qwen3_sampling_processors/top_k_full_sort_baseline	2.42 ms	2.22 ms	-8.5%
✅	sampling_latency/greedy_per_token	3.78 µs	3.45 µs	-8.6%
✅	tokenization/decode_tokens_per_second	8.25 ms	7.45 ms	-9.7%
✅	tokenization/encode_tokens_per_second	472.51 µs	418.92 µs	-11.3%
✅	matmul/medium_generic_bf16_threads=8/32x512x512	460.87 µs	400.03 µs	-13.2%
🟢	matmul/large_generic_f16_threads=1/32x1024x1024	108.91 µs	83.15 µs	-23.7%
🟢	matmul/large_generic_f32_threads=8/32x1024x1024	5.47 ms	3.97 ms	-27.4%
🟢	block_quantized_moe_cached_dense/mxfp4_uncached_expert_dequant_each_call/rows=1,H=256,I=256,E=4,top_k=1	557.88 µs	403.20 µs	-27.7%
🟢	block_quantized_matmul_cached_dense/mxfp4_cached_dense_repeated_call/1x1024x1024	80.43 µs	58.04 µs	-27.8%
🟢	block_quantized_matmul_cached_dense/mxfp4_uncached_dequant_each_call/1x1024x1024	828.70 µs	596.51 µs	-28.0%
🟢	block_quantized_matmul_cached_dense/mxfp4_preexpanded_dense_oncelock_like_proxy/1x1024x1024	81.80 µs	56.44 µs	-31.0%
🟢	matmul/large_generic_bf16_threads=1/32x1024x1024	3.38 ms	2.13 ms	-37.1%
🟢	matmul/large_generic_f16_threads=8/32x1024x1024	142.38 µs	88.91 µs	-37.6%
🟢	matmul/large_generic_bf16_threads=8/32x1024x1024	3.10 ms	1.58 ms	-49.1%

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: { 4.96 3.80 5.33 }

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)