add mobench support to ProveKit v1

[dcbuild3r]

Reopened from #430 using the new dcbuilder/* branch naming.

Old PR: #430
Replacement branch: dcbuilder/pr430-mobench-v1-browserstack
Old head branch: dcbuild3r/mobench-v1-browserstack
Head SHA: 94786155017ac2b11505a012ddb181f9e8a1a1c6

Previous PR CI / benchmark runs from #430:

• Mobile Bench PR Auto run 26368378198 — source for latest benchmark numbers
• Cargo Build & Test run 26368378803 — success
• Go CI run 26368378804 — success
• End-to-end CI run 26368378798 — cancelled

Summary

• add v1 mobench BrowserStack wiring for passport age-check and OPRF prove benchmarks
• run the same full Android/iOS triad matrix as PR add mobench support to ProveKit main #429
• fix Android prove timing so fixture clone/setup is outside the measured prove span
• use jemalloc as the Android fallback allocator under the native C FFI allocator
• harden Android incomplete-fixture reporting with timeout/build/kill-evidence fields

Android mobench fix note

The earlier failing run was https://github.com/worldfnd/provekit/actions/runs/26002840796. The missing Vivo Y21 monolithic cell did not recover a BrowserStack session payload or summary.json; available artifacts only show the BrowserStack fetch timeout after 7200s for build b1ebace919a1b8cad3861313180ac9a18a7e461d.

Before/failure memory from that run:

• Vivo Y21 OPRF: 291 MB process peak, 9.614s
• Vivo Y21 fragmented age check: 1206 MB process peak, 44.896s
• Vivo Y21 monolithic age check: no row; BrowserStack timeout/no summary
• S24 OPRF: 365 MB process peak, 2.216s

What changed:

• Android and iOS now use the same semantic boundary for prove benches: per-iteration fixture setup happens before the measured closure, and profile_phase("prove") wraps only the prover entry point. Previously prepared.clone() ran inside profile_phase("prove"), so clone cost and peak memory were charged to proving.
• Android native C FFI fallback allocation now routes through jemalloc instead of Bionic malloc when no host callback/mmap allocator is active.
• If BrowserStack returns no summary, failure.json now records attempts, fetch timeout seconds, build id, and any LMK/OOM/SIGKILL lines recovered from attempt/device logs.

Latest benchmark numbers were reposted as a PR comment from #430.

Validation

• cargo fmt --all
• cargo test -p bench-mobile --lib
• cargo test -p bench-mobile --test examples_smoke
• cargo test -p bench-mobile --test passport_smoke
• cargo check -p provekit-ffi --target aarch64-linux-android with NDK 26.1 aarch64-linux-android34-clang
• ruby -e 'require "yaml"; YAML.load_file(".github/workflows/mobile-bench-reusable.yml")'
• git diff --check

[dcbuild3r]

Latest mobile benchmark numbers

Reposted from #430 comment #430 (comment).

Source: Mobile Bench PR Auto run 26368378198 on dcbuild3r/mobench-v1-browserstack at 94786155.
Mobench version: 0.1.42. Artifact summaries report 2 measured runs per function/device plus 1 warmup run.

iOS

Benchmark	Device	Runs	Warmup	Mean/run	Median	p95	Min	Max	Wall total	CPU median/run	CPU total	CPU/wall	Peak growth	Peak memory	Process peak
OPRF	iPhone 16 Pro Max	2	1	510.137ms	510.137ms	513.350ms	506.923ms	513.350ms	1.020s	1.214s	2.429s	238.1%	12.20 MB	12.20 MB	355.25 MB
OPRF	iPhone 14	2	1	1.323s	1.323s	1.334s	1.312s	1.334s	2.646s	2.176s	4.353s	164.5%	10.86 MB	10.86 MB	147.92 MB
OPRF	iPhone SE 2022	2	1	956.647ms	956.647ms	1.053s	860.161ms	1.053s	1.913s	1.784s	3.569s	186.5%	11.02 MB	11.02 MB	148.77 MB
Passport fragmented	iPhone 16 Pro Max	2	1	3.085s	3.085s	3.124s	3.046s	3.124s	6.170s	12.666s	25.333s	410.6%	186.78 MB	186.78 MB	1014.66 MB
Passport fragmented	iPhone 14	2	1	4.215s	4.215s	4.243s	4.187s	4.243s	8.429s	16.325s	32.650s	387.3%	247.11 MB	247.11 MB	730.53 MB
Passport fragmented	iPhone SE 2022	2	1	4.345s	4.345s	4.389s	4.302s	4.389s	8.691s	16.986s	33.973s	390.9%	247.08 MB	247.08 MB	731.81 MB
Passport complete	iPhone 16 Pro Max	2	1	3.344s	3.344s	3.368s	3.320s	3.368s	6.688s	14.587s	29.174s	436.2%	359.48 MB	359.48 MB	1485.80 MB
Passport complete	iPhone 14	2	1	4.843s	4.843s	5.060s	4.625s	5.060s	9.685s	20.282s	40.565s	418.8%	489.75 MB	489.75 MB	1062.66 MB
Passport complete	iPhone SE 2022	2	1	4.842s	4.842s	5.069s	4.615s	5.069s	9.684s	20.403s	40.807s	421.4%	487.70 MB	487.70 MB	1057.06 MB

Android

Benchmark	Device	Runs	Warmup	Mean/run	Median	p95	Min	Max	Wall total	CPU median/run	CPU total	CPU/wall	Peak growth	Peak memory	Process peak
OPRF	Samsung Galaxy S24	2	1	2.096s	2.096s	2.104s	2.088s	2.104s	4.192s	5.343s	10.687s	254.9%	39.83 MB	39.83 MB	430.64 MB
OPRF	Google Pixel 7	2	1	5.686s	5.686s	5.915s	5.457s	5.915s	11.372s	7.901s	15.802s	139.0%	2.85 MB	2.85 MB	201.77 MB
OPRF	Samsung Galaxy M32	2	1	6.116s	6.116s	6.717s	5.515s	6.717s	12.232s	10.126s	20.252s	165.6%	1.80 MB	1.80 MB	286.83 MB
Passport fragmented	Samsung Galaxy S24	2	1	13.130s	13.130s	13.281s	12.978s	13.281s	26.260s	41.271s	82.543s	314.3%	263.05 MB	263.05 MB	984.10 MB
Passport fragmented	Google Pixel 7	2	1	17.882s	17.882s	17.919s	17.845s	17.919s	35.764s	45.644s	91.289s	255.3%	192.33 MB	192.33 MB	716.82 MB
Passport fragmented	Samsung Galaxy M32	2	1	30.928s	30.928s	30.973s	30.884s	30.973s	61.857s	85.175s	170.350s	275.4%	308.50 MB	308.50 MB	854.13 MB
Passport complete	Samsung Galaxy S24	2	1	9.409s	9.409s	9.424s	9.395s	9.424s	18.819s	34.250s	68.501s	364.0%	442.15 MB	442.15 MB	1205.77 MB
Passport complete	Google Pixel 7	2	1	16.684s	16.684s	16.891s	16.477s	16.891s	33.368s	47.877s	95.755s	287.0%	378.04 MB	378.04 MB	956.37 MB
Passport complete	Samsung Galaxy M32	2	1	28.874s	28.874s	29.144s	28.604s	29.144s	57.748s	89.707s	179.414s	310.7%	586.32 MB	586.32 MB	1335.73 MB

Memory growth is the benchmark-reported peak growth above baseline; process peak is the reported process RSS peak.

[github-actions]

Mobench Benchmark Results

Benchmark Summary

• Generated: 2026-06-21T07:29:27.88932Z
• Target: iOS
• Function: bench_mobile::bench_oprf_prove
• Iterations/Warmup: 5 / 1
• Devices: iPhone SE 2022-15, iPhone 14-16, iPhone 16 Pro Max-18

Device	Function	Samples	Warmup	Wall mean / iter	Wall total	CPU median / iter	CPU total	CPU / wall	Peak growth	Process peak
iPhone 14	bench_mobile::bench_oprf_prove	5	1	1.355s	6.776s	2.174s	10.797s	159.3%	11.33 MB	154.83 MB
iPhone 16 Pro Max	bench_mobile::bench_oprf_prove	5	1	502.288ms	2.511s	1.204s	6.029s	240.1%	12.23 MB	355.47 MB
iPhone SE 2022	bench_mobile::bench_oprf_prove	5	1	1.186s	5.930s	1.993s	9.901s	167.0%	11.17 MB	156.75 MB

Note: memory growth excludes warmup/baseline retained before the measured iteration.

Device Comparison Plots

oprf-prove

Ios Sina Plot

Benchmark Summary

• Generated: 2026-06-21T07:52:35.709455Z
• Target: Android
• Function: bench_mobile::bench_oprf_prove
• Iterations/Warmup: 5 / 1
• Devices: Samsung Galaxy S24-14.0

Device	Function	Samples	Warmup	Wall mean / iter	Wall total	CPU median / iter	CPU total	CPU / wall	Peak growth	Process peak
Samsung Galaxy S24	bench_mobile::bench_oprf_prove	5	1	2.201s	11.007s	5.656s	28.203s	256.2%	45.08 MB	433.46 MB

Note: memory growth excludes warmup/baseline retained before the measured iteration.

Device Comparison Plots

oprf-prove

Android Sina Plot

---

Posted by mobench at 2026-06-21 11:20 UTC

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
docs-provekit-org	Error		Jun 21, 2026 4:12am

[dcbuild3r]

@Bisht13 ready for review once the fresh rerun settles.

I pushed ae3d6c09 to add the five-sample mobile trigger and kicked off full CI with [mobench:both:5]: https://github.com/worldfnd/provekit/actions/runs/27893152094