Crate fgumi_simd_fastq 

SIMD-accelerated FASTQ parsing using Helicase-style bitmask operations.

This crate provides high-throughput FASTQ parsing by processing 64 bytes at a time through SIMD registers (NEON on ARM, AVX2 on x86_64), classifying newline characters via bitmask operations and finding record boundaries without per-byte branching.

Architecture

1. Lexer: Loads 64-byte blocks into SIMD registers, produces a u64 bitmask where bit i is set if byte i is a newline (\n).
2. Parser: Walks the newline bitmask with trailing_zeros() to find record boundaries. Every 4th newline marks the end of a FASTQ record.

Example

use fgumi_simd_fastq::{find_record_offsets, parse_records};

let fastq = b"@r1\nACGT\n+\nIIII\n@r2\nTTTT\n+\nJJJJ\n";
let offsets = find_record_offsets(fastq);
assert_eq!(offsets, vec![0, 16, 32]);

let records: Vec<_> = parse_records(fastq).collect();
assert_eq!(records.len(), 2);
assert_eq!(records[0].name, b"r1");
assert_eq!(records[0].sequence, b"ACGT");

Structs

FastqBitmask

Bitmask output from lexing a 64-byte block.

FastqRecord

A borrowed FASTQ record with zero-copy slices into the input buffer.

SimdFastqReader

Buffered FASTQ reader that uses SIMD-accelerated record boundary detection.

Functions

find_record_offsets

Find FASTQ record boundary offsets in a byte buffer.

lex_block_full

Lex a 64-byte block, producing newline bitmask, ACGT bitmask, and 2-bit encoding.

parse_records

Iterator over zero-copy FASTQ records parsed from a byte buffer.