ci: reuse the Gradle daemon for benchmark builds

[mrousavy]

The paired Android benchmark build currently starts and stops a single-use Gradle daemon for each revision. Remove --no-daemon so compatible head and base builds in the same job can reuse the daemon. The existing Gradle build cache stays enabled; device measurements run in a separate job.

All checks passed on 41bfd38dd663f12f33ef625e6bc10a9867cc6fcd, including both native builds and both measurement jobs in run 34154006746. Bash syntax, ShellCheck, artifact selection and app-isolation tests also passed.

Build timing review:

Configuration	Head Gradle build	Base Gradle build	Build/package APK step
Daemon reuse, current head	71 s	61 s	144 s
Unchanged Android build, #1629	63 s	62 s	135 s
Unchanged Android build, #1630	67 s	64 s	143 s
Daemon reuse, before rebase	67 s	55 s	133 s

The current head and two controls share the same main base, runner class and restored Gradle cache. Their task counts match: head executes 140 tasks with 64 cached; base executes 135 with 69 cached. Both daemon runs start one reusable JVM, while the controls start two single-use daemons. The earlier run's Android build passed before the workflow was cancelled by the rebase.

The second build benefits from reuse, but a reduction in total build time is not demonstrated by these separate hosted runs. Their build/package timings overlap. Keep the PR open pending stronger timing evidence; current results do not justify claiming an overall speedup.

[vercel]

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

Project	Deployment	Actions	Updated
nitro-docs	Ready	Preview	Sep 9, 2026 12:31pm UTC

[mrousavy]

@nitro-modules-bot please test performance

[mrousavy]

@nitro-modules-bot please test performance

[mrousavy]

@nitro-modules-bot please test performance

[mrousavy]

https://github.com/nitro-modules-bot please test performance

[mrousavy]

@nitro-modules-bot please test performance

[mrousavy]

@nitro-modules-bot please test performance

[nitro-modules-bot]

Performance Report

iOS

Benchmark	Before	After	Difference
C++ immediatePromise()	775.6 ns	819.1 ns	🔴 +5.61% slower

All Benchmarks

Benchmark	Before	After	Difference
JavaScript addNumbers()	23.3 ns	24.3 ns	🔴 +4.16% slower
C++ bounce(1 MiB)	234.6 ns	234.9 ns	🔴 +0.14% slower
C++ bounce(4 KiB)	228.6 ns	233.1 ns	🔴 +1.98% slower
C++ bounce native-owned buffer (1 MiB)	3.67 µs	3.62 µs	🟢 -1.49% faster
C++ bounce native-owned buffer (4 KiB)	351.5 ns	354.3 ns	🔴 +0.8% slower
C++ copy(1 MiB)	63.51 µs	65.15 µs	🔴 +2.58% slower
C++ copy(4 KiB)	960.2 ns	964.7 ns	🔴 +0.46% slower
C++ large array (1,024)	48.90 µs	48.95 µs	🔴 +0.1% slower
C++ small array (16)	913.7 ns	911.7 ns	🟢 -0.22% faster
C++ synchronousCallback()	198.6 ns	196.7 ns	🟢 -0.97% faster
C++ create()	421.8 ns	428.0 ns	🔴 +1.46% slower
C++ returnExisting()	192.0 ns	191.9 ns	🟢 -0.04% faster
C++ typed map (8 entries)	1.87 µs	1.83 µs	🟢 -1.71% faster
C++ optional trailing string	142.1 ns	144.1 ns	🔴 +1.4% slower
C++ addNumbers()	82.3 ns	82.3 ns	🔴 +0.01% slower
C++ simpleFunc()	65.4 ns	64.6 ns	🟢 -1.12% faster
C++ deferred worker Promise (includes trigger call)	21.86 µs	21.13 µs	🟢 -3.36% faster
C++ number property get/set	130.4 ns	129.3 ns	🟢 -0.85% faster
C++ short ASCII string	138.3 ns	137.4 ns	🟢 -0.61% faster
C++ Unicode string	276.6 ns	272.4 ns	🟢 -1.54% faster
C++ nested Car struct	2.41 µs	2.38 µs	🟢 -1.14% faster
C++ number | string variant	125.8 ns	124.1 ns	🟢 -1.37% faster
Swift bounce(1 MiB)	220.9 ns	210.4 ns	🟢 -4.78% faster
Swift bounce(4 KiB)	225.3 ns	230.0 ns	🔴 +2.05% slower
Swift bounce native-owned buffer (1 MiB)	3.29 µs	3.31 µs	🔴 +0.44% slower
Swift bounce native-owned buffer (4 KiB)	342.8 ns	342.7 ns	🟢 -0.03% faster
Swift copy(1 MiB)	63.17 µs	60.90 µs	🟢 -3.58% faster
Swift copy(4 KiB)	904.6 ns	892.7 ns	🟢 -1.32% faster
Swift large array (1,024)	45.01 µs	45.61 µs	🔴 +1.35% slower
Swift small array (16)	933.4 ns	958.7 ns	🔴 +2.71% slower
Swift synchronousCallback()	232.9 ns	226.4 ns	🟢 -2.79% faster
Swift create()	24.99 µs	23.88 µs	🟢 -4.45% faster
Swift returnExisting()	237.5 ns	236.8 ns	🟢 -0.28% faster
Swift typed map (8 entries)	2.39 µs	2.39 µs	🔴 +0.3% slower
Swift optional trailing string	154.8 ns	149.1 ns	🟢 -3.7% faster
Swift addNumbers()	68.3 ns	69.3 ns	🔴 +1.55% slower
Swift simpleFunc()	57.7 ns	58.0 ns	🔴 +0.6% slower
Swift deferred worker Promise (includes trigger call)	23.92 µs	24.15 µs	🔴 +0.97% slower
Swift immediatePromise()	1.28 µs	1.24 µs	🟢 -3.4% faster
Swift number property get/set	128.5 ns	125.1 ns	🟢 -2.62% faster
Swift short ASCII string	157.5 ns	157.8 ns	🔴 +0.18% slower
Swift Unicode string	330.9 ns	324.4 ns	🟢 -1.96% faster
Swift nested Car struct	2.26 µs	2.26 µs	🟢 -0.15% faster
Swift number | string variant	188.0 ns	191.9 ns	🔴 +2.11% slower
TurboModule addNumbers()	905.0 ns	907.2 ns	🔴 +0.25% slower

Android

Benchmark	Before	After	Difference
C++ create()	612.4 ns	559.7 ns	🟢 -8.6% faster
C++ typed map (8 entries)	1.61 µs	1.71 µs	🔴 +5.63% slower
Kotlin deferred worker Promise (includes trigger call)	32.05 µs	34.96 µs	🔴 +9.1% slower

All Benchmarks

Benchmark	Before	After	Difference
JavaScript addNumbers()	21.2 ns	21.1 ns	🟢 -0.35% faster
C++ bounce(1 MiB)	411.1 ns	404.9 ns	🟢 -1.51% faster
C++ bounce(4 KiB)	400.0 ns	397.4 ns	🟢 -0.67% faster
C++ bounce native-owned buffer (1 MiB)	3.17 µs	3.18 µs	🔴 +0.31% slower
C++ bounce native-owned buffer (4 KiB)	413.5 ns	407.1 ns	🟢 -1.54% faster
C++ copy(1 MiB)	26.57 µs	26.68 µs	🔴 +0.42% slower
C++ copy(4 KiB)	775.8 ns	808.4 ns	🔴 +4.2% slower
C++ large array (1,024)	38.10 µs	38.35 µs	🔴 +0.65% slower
C++ small array (16)	775.9 ns	771.8 ns	🟢 -0.53% faster
C++ synchronousCallback()	312.7 ns	310.9 ns	🟢 -0.56% faster
C++ returnExisting()	214.3 ns	212.4 ns	🟢 -0.89% faster
C++ optional trailing string	137.8 ns	134.4 ns	🟢 -2.44% faster
C++ addNumbers()	88.7 ns	86.0 ns	🟢 -3.09% faster
C++ simpleFunc()	69.9 ns	70.1 ns	🔴 +0.24% slower
C++ deferred worker Promise (includes trigger call)	22.30 µs	22.13 µs	🟢 -0.74% faster
C++ immediatePromise()	844.0 ns	859.1 ns	🔴 +1.79% slower
C++ number property get/set	151.2 ns	151.2 ns	🔴 +0.03% slower
C++ short ASCII string	126.3 ns	126.1 ns	🟢 -0.16% faster
C++ Unicode string	230.4 ns	229.2 ns	🟢 -0.52% faster
C++ nested Car struct	2.05 µs	2.06 µs	🔴 +0.54% slower
C++ number | string variant	124.6 ns	123.6 ns	🟢 -0.8% faster
Kotlin bounce(1 MiB)	1.01 µs	967.9 ns	🟢 -3.84% faster
Kotlin bounce(4 KiB)	935.9 ns	936.6 ns	🔴 +0.08% slower
Kotlin bounce native-owned buffer (1 MiB)	3.83 µs	3.91 µs	🔴 +2.11% slower
Kotlin bounce native-owned buffer (4 KiB)	1.10 µs	1.10 µs	🔴 +0.17% slower
Kotlin copy(1 MiB)	263.64 µs	272.33 µs	🔴 +3.3% slower
Kotlin copy(4 KiB)	4.49 µs	4.45 µs	🟢 -0.83% faster
Kotlin large array (1,024)	40.71 µs	39.77 µs	🟢 -2.3% faster
Kotlin small array (16)	1.00 µs	1.01 µs	🔴 +0.84% slower
Kotlin synchronousCallback()	991.4 ns	953.6 ns	🟢 -3.81% faster
Kotlin create()	1.50 µs	1.52 µs	🔴 +1.18% slower
Kotlin returnExisting()	549.1 ns	536.4 ns	🟢 -2.31% faster
Kotlin typed map (8 entries)	5.91 µs	5.89 µs	🟢 -0.23% faster
Kotlin optional trailing string	303.0 ns	307.7 ns	🔴 +1.56% slower
Kotlin addNumbers()	117.9 ns	118.4 ns	🔴 +0.46% slower
Kotlin simpleFunc()	91.9 ns	92.0 ns	🔴 +0.09% slower
Kotlin immediatePromise()	2.11 µs	2.10 µs	🟢 -0.7% faster
Kotlin number property get/set	217.3 ns	222.6 ns	🔴 +2.48% slower
Kotlin short ASCII string	402.6 ns	406.7 ns	🔴 +1.03% slower
Kotlin Unicode string	694.5 ns	690.7 ns	🟢 -0.55% faster
Kotlin nested Car struct	4.26 µs	4.22 µs	🟢 -1.14% faster
Kotlin number | string variant	388.2 ns	385.8 ns	🟢 -0.62% faster
TurboModule addNumbers()	213.8 ns	215.0 ns	🔴 +0.6% slower

Benchmarking Code Diff 201794bf...9351d8d9 (view raw output)

Raw measurements and artifacts

Raw measurements: performance-report-1 (JSON artifact). Run 34351445968, attempt 1. Download requires GitHub access.

Android: measurements, attempt 1, apps, attempt 1.
iOS: measurements, attempt 1, apps, attempt 1.