There was an error while loading. Please reload this page.
8a12a8c
v0.3.4: SIMD string scanning, zero-copy decode, fast path - SIMD string scanning: AVX2 (32B/iter, x86_64), SSE2 (16B/iter, x86_64), NEON (16B/iter, aarch64), 8-byte scalar fallback for other targets. Runtime AVX2 detection with #[target_feature] for correct codegen. - Zero-copy sub-binary references for non-escaped strings >= 64 bytes. - Decode fast path: decode!/1 bypasses option parsing for no-options calls. - 2-4x faster on string-heavy payloads, 15-30% faster on small payloads vs v0.3.3.