Advent of Code 2023 Day 6 Optimizations

fastest_day6_part2.rs

// time:   [1.8843 µs 1.8897 µs 1.8955 µs]
pub fn original(input: &[u8]) -> Option<usize> {
    let mut idx = 0;
    'outer: while idx + 13 < input.len() {
        let mut state = 0;
        for (next_idx, byte) in input[idx..idx + 14].iter().enumerate().rev() {
            let bit_idx = byte % 32;
            if state & (1 << bit_idx) != 0 {
                idx += next_idx + 1;
                continue 'outer;
            }
            state |= 1 << bit_idx;
        }
        return Some(idx);
    }
    return None;
}

// time:   [972.80 ns 980.40 ns 987.77 ns]
pub fn my_fastest(input: &[u8]) -> Option<usize> {
    let mut idx = 0;
    while let Some(slice) = input.get(idx..idx + 14) {
        let mut state = 0u32;

        if let Some(pos) = slice.iter().rposition(|byte| {
            let bit_idx = byte % 32;
            let ret = state & (1 << bit_idx) != 0;
            state |= 1 << bit_idx;
            ret
        }) {
            idx += pos + 1;
        } else {
            return Some(idx);
        }
    }
    return None;
}

// time:   [6.6444 µs 6.6834 µs 6.7255 µs]
pub fn benny(input: &[u8]) -> Option<usize> {
    let mut filter = 0u32;
    input
        .iter()
        .take(14 - 1)
        .for_each(|c| filter ^= 1 << (c % 32));
    input.windows(14).position(|w| {
        let first = w[0];
        let last = w[w.len() - 1];
        filter ^= 1 << (last % 32);
        let res = filter.count_ones() == 14 as _;
        filter ^= 1 << (first % 32);
        res
    })
}

//  time:   [2.7528 µs 2.7618 µs 2.7721 µs]
#[cfg_attr(target_arch = "x86_64", target_feature(enable = "popcnt"))]
pub unsafe fn benny_popcnt(input: &[u8]) -> Option<usize> {
    let mut filter = 0u32;
    input
        .iter()
        .take(14 - 1)
        .for_each(|c| filter ^= 1 << (c % 32));
    input.windows(14).position(|w| {
        let first = w[0];
        let last = w[w.len() - 1];
        filter ^= 1 << (last % 32);
        let res = filter.count_ones() == 14 as _;
        filter ^= 1 << (first % 32);
        res
    })
}

my_benchmark.rs

use std::time::Duration;

use aoc_2022::*;
use criterion::{criterion_group, criterion_main, Criterion};

pub fn criterion_benchmark(c: &mut Criterion) {
    let input = "rnttlvtttmnmpmhpmmzvmmhpmmnrntnnsnrnndvnddmbbtptssjcczmmbwmbmwwmmflfggwzzhjjgppwrwdwqwbbbmwbwgggqccmlmdmgdgqgpqgpgzzndddgdbbsvvfsfppwjwzjjcnjccwrwrprgppbpddnccjggfrggqngqgdqgddvsshqhmqqfvqvlvwvnwwmccrpcczvcczzgsgwwlggqsggmdgglblpblltbbzrbzzcscgccwssbddsmddzzvvhjjwjwrjwrwfftmttplldnllqttdhddmvdmmhsmsmqsqwqpqbpqppqdppbnpnhhppjbbrzrbrggdbddnrdrzrdrpdppzzfzrzgzllwlwzlllwtltslttlmtmqqbbzmzhmmjjwrjrrnzrzhzmmtbbrppmmfddjhdjdsdllrbbpfphpzpwwjvwwdpwpzwwzrrfnfwwfbbpttzjtjptpvvsfvvljjrzzmdmsddpvpgvvdcdwcwhwghwggrjjhdjhddzhhwbhwbwwbrwwrfrmfrfvrfftzzqrzqrqtrqtrqtrqttgctclcwlwwvqqnlnqlnnndmnmggznzhzqzggzrzbbrwbrwrvvscvvmvhmvvbsbjsjfjpfpqpbbbbjtjptpprnndpnnrsswffmnnjhhpqhqwhhbwhbhgbbfrbrjjpmjmzzdzjdjffcvvtwttrrcscvcncqcfcwwpgpfptpcpspllcfcssgbglgdllcdcmcpcqclctcvtvvbhbbdhhczhzshzhbhchzchchlcllwswqqmpmbbqwwzgzhggczgzffsgggjjlcjjpnjjfqjfjvfvrfrlrvvpfpcfcbffnmfmbmbgggplldblddvcvjcvvmbmmmrvrvvnhhqttgdgvdgdqgddcsclljplpbbsvbsbfsffshsgslsllzztggttfccctwtswwrvwvnnfbnnvjnvvlgvvfppmvpmmvpmmcttjffgsffcllncntcncfclcgcppvdvpvrvbrrnnvrvhrvrzvrvnrvrdrnrqrnngsshqhcqqbfqqzmmmzjzqztznttjffqzfqfbqqsggclgclcddtqqdppbjppqvpqvvmqmqnqcnqcqcqvqjqzqbzzbrrgfgddgtgnnpjjbzjjbbtstzszwswggfffmnnpllfvvnwnpwpqwqwlllqlmmzcmcrcmrcmrcrffnrnssjjrdjjwwqgwqgqdggwzwjzzbfbpptctchthbbbqsqggrllldwdffvwwlrwwljwjnncmmjmvjmmqnnmrrgjgvvpqqbppqlpqllqnqdndwwlppcjcdjjcnnmddffgjjrwjrwjjhvhqvhhfssrtsrsgrrwjwhhbqqpzzrdrhrqhrqrprqqmllsszhzllcdllpmpbbmsmjjmwmssvlslwwmtwmmgvgcvggzcgzgddsjsjfsssftfcchhfmfcmcpcpvcppqggjddljjbttwrrhghqggpghgngzngznnzqnnssqlqmqbmmmqdqqhttnqqjhhphrhccbggmjjpwpbwpprdrvvjzvvtctzzmpphwpwdpdvvhbvvdvsdsjdjbdjdljjstswttnhhbfffmnngddlglqggvzvlzvzqqvbqbvbjbnjjsvjjwpjjzshrdtrjttvqnbltrfpvnztrwzrtgjpzgqdjfglqgjrgdzbhqpghdbfhlfjhbfjfjppfgljmgwljlsbmltgztthnzvdrgqlgddvqhzctdgcphfqvnpjjgzwqfvnhvzdrwtpgfdjpqnfshslqmplcprdntnhqqbqptwzvdddhcjcqrfhjqnjvpnhttblwgjwlfwntdchgfjmdbgtqtdgnzbqwzzcltwtmtqtdbvjtfvlzpcvgmrfqwfhwqmhvwhftzgmhshffnjwqbvztszsrrglqvhfpqmbnqjsfnwdwgdtmztbvqrmztfctmvptbwnfzfgdztjgnqsrsqqqnrpgzsqszzwwwgqnnnrdzhzdbqjgbvncprzcjqchzfgnclbrmphbsdwwpvwjwlbshhgjfbhjjtdqrmrcjfnrrhqrpsbglthzpvfglqspttdpwlljhnlrjpzchbrqcgtmcscjnwvpztfjdcwbnbgmbpgdthgnhbrtwftnscbsrndghbslflpcpjwbcjnhzcwdcslmzqbtrlnzmntlpjcsctnsqwtbffqlhfgcsflvfwnmczvsbflnnnzpfjfrcwhhcbtbjcbghtcwcgdbrwrgfgvpwtcwlwcmnmrtcrjbwtwlrfstztsghfvrfjzzpswpqfqpvqstvbhqfjlgmtdlhqrhwzqpnqpllnlgzwptbgftmblqcwfcllbwfzdhrndfrvdvwzqvhnghlzvhldnnvrgqvlpfdnpmcgddjmstzsqfvzwftflrwtzqwjbbqhjpfbdztdfsgsztvvrvslgspgpdcmwszdfsddqhpzpsjgqmgzqvhchlgrmcmzwzbtwfphvgcdmhfdczhffgmqpncdjszzgwfvwsqddvbcgngbjwhmphjsmjthvbthhfwdtqmjctcmdpqpsdrnrzdgzgzctbhwsgvtjgwjbsnnjmpmqgwrnqfqbpnrpddjsrsvmcshhthwfrwmqsrjhlsrgfzvwmdzhwrvchppqldghgzflrnwqnvntmtdwmrpgbdbzvcmnqstzntvllcgzsnvrhqzsfncznhgrggmvrmgsqmhbdsjbsqqhzppfcwdrgdvfjdscrvpwtsdmcnczwbbjhvddprwtzfwslcfdcrqfszcgmhtdfvlqzqtvwngzvmmqcrqpzwzhggjnphsrmnctnfhtppglspnvzrsqfgzdfrrwbzbqwvbvbnzgmdrqrnsvdpvlgcmnggsbmbtfwrvdjrtgtgcqscnfpgswgsngdqnnscffdcnlrcpdpcbpzvqcrtjhlwvgnfhhqmprthrtcvcjjwgprqqdwfbgmzlwttjpvcjzfwbdhvngsjpgtqsvbldbcvhjhzbjzblqtqhlnbzzqfcpnzdhbplztcgvzhbgshqbccgwzhftqvtwzbwmnfrsgphhhgtsmwlqhlcchtbtggqwmbdthhmqqjtfdvfpddfdrtfjbpmwtcbfnrhwcnpdqrdtsfdmjfzdwwgnftnwpssgqtlpdbwhnzcnfmppclsswbhcdghpnslwjznqszgdtrnpncsqsnbrplrfwpbnfnvttlzcjtvhzcpzmhfsfzfjlzqqnprpdvwbfthmrswqrcqqwrnwzmgjqqsnqdblssmhngjjvprmqbswtgzzvprwhrgjqshvmwzgrgfmzlgrtzbmdlzncwqdftfsndvdfmmplswdbjtbcbvcvtpjvrqpghczpqvvpqwfbfhllbpvrrfsmsjhqbldcrwvzvcvzfffqvplbfbdbwctbjsljlfwtbcnpsbtpmcqgdvltmztvrcfsprbnvwplwhncgsdnrdmqnmcvpvvrlmlwtgvrnwvzsfctdlcfvtgqmpnbwcbwvfpmqnbjvwffpjtvvgflhrnlngrzhhttdtdbvscftsqtvbrgzfgsjvwhzjcbsqcttlgwmhhvjhwhgmmdtflfdbvnhgcblqmwjfsqnngjqfbvdnsfbgwjfhpgdgvhpbvlrtpcpvhrbtqpbffzcqrvbqwvqmmrcwtnvcgwvsqzvrwdbcnjshbnbftmmvrmjvgfdwbsjvqfdwnnvqqhbmshcrclrwhfhbtnwqvmrrvdwgcwcsrhdbqndsthmrmbjhttjtzmlflbrmhlcgsbdjcjcvwcjffnqrntpflrgfcngpchtrzpnflwjvcgbwtsnjfqsggwmwhdvbzdpjmtwlmrslnjsndjtgjmmwmdgtnfrztppzqvqhbfzqpsdhvsshddlzwcmsndrpqhndsrjnngnmgmzrvchwlqgdnbssbhpbpgwpsrcnbphpslvqplhpgdhmrnwwjmhvnsfjmrfwtvjjrmgptvjffhbgpmfgmgrcjrwqhccssrqjpljbpwcvsfdtmbhzsmsjsgblgpcqszsttfclrjcnsslmngmbmwqfhddbvmbvwmrmvglsl".as_bytes();

    let mut group = c.benchmark_group("day6_part2");

    group.measurement_time(Duration::from_secs(10));

    group.bench_with_input("original", input, |b, s| b.iter(|| original(s)));

    group.bench_with_input("my_fastest", input, |b, s| b.iter(|| my_fastest(s)));
    
    group.bench_with_input("benny", input, |b, s| b.iter(|| benny(s)));

    group.bench_with_input("benny_popcnt", input, |b, s| b.iter(|| unsafe { benny_popcnt(s) }));

    group.finish();
}

criterion_group!(benches, criterion_benchmark);
criterion_main!(benches);

In line 26, do I have it right that subtracting by 96 instead of moduloing by 32 would achieve the same result? Is moduloing faster?

by using modulus, it can process both capital and small letters, when it comes to speed it is rather similar.