perf(allocator): increase initial chunk size from 512B to 16KB

[Boshen]

Summary

• Increase FIRST_ALLOCATION_GOAL from 512 bytes to 16 KB to reduce malloc calls
• Fix Bump::with_capacity() to respect exact capacity requested (otherwise with_capacity(1024) would allocate 16KB)

Malloc Reduction

Tested against 310K real-world JS/TS files:

Dataset	Files	Before (avg mallocs)	After	Reduction
oxlint-ecosystem-ci	256K	10.99	7.72	30%
monitor-oxc npm	53K	11.56	8.51	26%

Memory overhead increases by ~37% in allocated capacity, but actual memory used remains the same since AST size does not change.

Parser Benchmark (vs main)

File	Size	Change
RadixUIAdoptionSection.jsx	2.5 KB	-5.14% ✅
cal.com.tsx	1 MB	-4.85% ✅
react.development.js	84 KB	-0.19%
binder.ts	193 KB	-0.53%
checker.ts	2.9 MB	-0.03%

Smaller/medium files show ~5% speedup from fewer malloc calls.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR optimizes memory allocation performance by increasing the initial chunk size from 512 bytes to 16 KB, reducing malloc calls by approximately 30% based on benchmarking against 310K real-world JavaScript/TypeScript files.

Changes:

• Increased FIRST_ALLOCATION_GOAL constant from 512B (1 << 9) to 16KB (1 << 14)
• Updated documentation to explain the rationale with benchmark data
• Fixed Bump::with_capacity() to use the requested capacity as the starting point

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codspeed-hq]

CodSpeed Performance Report

Merging this PR will not alter performance

_{Comparing perf/allocator-initial-size-16kb (e6c8c44) with main (f1e2dc0)}

Summary

✅ 42 untouched benchmarks
⏩ 3 skipped benchmarks¹

Footnotes

1. 3 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. ↩

[Boshen]

Codspeed should not reveal any performance improvements because the allocator is always created outside the benchmark.

just allocs reveals less mallocs.

[Boshen]

Not seeing any downsides of this change, merging.

[Boshen]

Merge activity

• Jan 19, 2:36 PM UTC: The merge label '0-merge' was detected. This PR will be added to the Graphite merge queue once it meets the requirements.
• Jan 19, 2:36 PM UTC: Boshen added this pull request to the Graphite merge queue.
• Jan 19, 2:52 PM UTC: Merged by the Graphite merge queue.