feat: support running TypeScript benchmarks with tsx

[adriencaccia]

Make tsx safe to benchmark with CodSpeed, so the docs can recommend it instead of the unmaintained esbuild-register, which cannot run a top-level await. Under CodSpeed, a preloaded loader turns off the esbuild options tsx forces (keepNames ran __name() calls inside the measured code, up to +80% on closure-heavy benchmarks, and hid arrow function names in flamegraphs) and the --ping of esbuild's service, whose handling every second was counted in random simulation measurements; scripts started with the tsx CLI rerun themselves with the runner's node flags, and tinybench now lets promise jobs left by module loading run before its first measurement. On Node 22.13 to 26.8, simulation instruction counts, walltime and memory results match esbuild-register.

Refs COD-3900

[codspeed]

Merging this PR will regress 4 benchmarks

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ 14 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• wait 500ms
• wait 500ms
• wait 1sec
• wait 1sec
• wait 1ms
• test_recursive_fibo_10
• switch 1
• wait 1ms
• recursive fibo 10
• test_recursive_fibo_10
• …and 4 more

ℹ️ 22 benchmarks have no flame graph

The flamegraph could not be generated because too much time was spent in runtime internals, making the visualization not useful for performance analysis.

• test_recursive_cached_fibo_10
• test sync baz 100
• test_recursive_cached_fibo_20
• test sync baz 10
• switch 2
• test_recursive_cached_fibo_10
• test_iterative_fibo_10
• test_iterative_fibo_100
• wait 1ms
• test_recursive_fibo_10
• …and 12 more

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 36 improved benchmarks
❌ 4 regressed benchmarks
✅ 223 untouched benchmarks
🆕 3 new benchmarks
⏩ 57 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Memory	test_recursive_cached_fibo_10	784 B	1,440 B	-45.56%
❌	Simulation	recursive fibo 15	350.7 µs	635.6 µs	-44.83%
❌	Memory	test sync baz 10	784 B	1,312 B	-40.24%
❌	WallTime	switch 2	84 ns	108 ns	-22.22%
⚡	Memory	switch 2	13,384 B	656 B	×20
⚡	WallTime	test_recursive_cached_fibo_10	2,544 ns	612 ns	×4.2
⚡	WallTime	test_recursive_cached_fibo_20	4 µs	1.7 µs	×2.4
⚡	WallTime	test_recursive_cached_fibo_30	4.7 µs	2.2 µs	×2.1
⚡	Simulation	test_recursive_cached_fibo_10	85.3 µs	45.7 µs	+86.71%
⚡	Simulation	recursive fibo 15 with hooks	644.1 µs	362 µs	+77.94%
⚡	Simulation	recursive fibo 15 with hooks	644.2 µs	362.2 µs	+77.89%
⚡ ⚠️	Simulation	wait 500ms	47.9 ms	27.8 ms	+72.52%
⚡ ⚠️	Simulation	wait 500ms	47.1 ms	28.1 ms	+67.77%
⚡ ⚠️	Simulation	wait 1sec	85.1 ms	56.2 ms	+51.46%
⚡ ⚠️	Simulation	wait 1sec	83.7 ms	56.9 ms	+47.07%
⚡	Simulation	test_recursive_cached_fibo_20	81.9 µs	56.7 µs	+44.52%
⚡	Simulation	test_recursive_cached_fibo_30	94.8 µs	69.7 µs	+36.06%
⚡	Simulation	test_recursive_cached_fibo_20	109.8 µs	81.8 µs	+34.23%
⚡ ℹ️	Simulation	test_recursive_cached_fibo_10	112 µs	83.5 µs	+34.22%
⚡	Simulation	test_recursive_cached_fibo_30	117.6 µs	89.6 µs	+31.25%
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing cod-3900-use-tsx-instead-of-esbuild-register-for-nodejs-typescript (9fab634) with main (d15b97d)}

Footnotes

1. 57 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩