cdrustanoia/crates/core/examples/bench_paranoia.rs

//! Small benchmark harness for the paranoia read path and sample index.
//!
//! Run with:
//!   cargo run --release --example bench_paranoia
//!   cargo run --release --example bench_paranoia -- 5000

use cdrustanoia_core::device::image::ImageDrive;
use cdrustanoia_core::device::samples_to_bytes;
use cdrustanoia_core::paranoia::sort::SortIndex;
use cdrustanoia_core::paranoia::{Paranoia, PARANOIA_MODE_FULL};
use std::collections::HashMap;
use std::time::Instant;

fn make_samples(sectors: usize) -> Vec<i16> {
    let mut out = Vec::with_capacity(sectors * 1176);
    let mut x = 0x1234_5678_9abc_def0u64;
    for i in 0..sectors * 1176 {
        x ^= x << 13;
        x ^= x >> 7;
        x ^= x << 17;
        let v = if i % 5000 < 700 {
            0
        } else {
            (x as i16).wrapping_add((i as i16).wrapping_mul(11))
        };
        out.push(v);
    }
    out
}

fn bench_index(samples: &[i16]) {
    let mut index = SortIndex::new(samples.len());
    let started = Instant::now();
    index.setup(samples, 0, 0, samples.len() as i64);
    let setup = started.elapsed();

    let mut found = 0usize;
    let started = Instant::now();
    for i in 0..10_000i64 {
        let post = (i * 7919) % samples.len() as i64;
        let value = samples[post as usize];
        if index.getmatch(samples, post, 512, value).is_some() {
            found += 1;
            while index.nextmatch(samples).is_some() {
                found += 1;
            }
        }
    }
    let queries = started.elapsed();
    println!(
        "SortIndex: setup={:?} 10k queries={:?} matches={} ({} samples)",
        setup,
        queries,
        found,
        samples.len()
    );
}

struct OldIndex {
    buckets: HashMap<i16, Vec<u32>>,
    size: usize,
    cursor: Option<(i16, usize, usize)>,
}

impl OldIndex {
    fn new() -> Self {
        Self {
            buckets: HashMap::new(),
            size: 0,
            cursor: None,
        }
    }
    fn setup(&mut self, samples: &[i16]) {
        self.buckets.clear();
        self.size = samples.len();
        for (i, &sample) in samples.iter().enumerate() {
            self.buckets.entry(sample).or_default().push(i as u32);
        }
    }
    fn getmatch(&mut self, post: i64, overlap: i64, value: i16) -> Option<usize> {
        let post = post.clamp(0, self.size as i64);
        let lo = (post - overlap).max(0) as usize;
        let hi = ((post + overlap).min(self.size as i64)).max(0) as usize;
        self.cursor = None;
        let positions = self.buckets.get(&value)?;
        let idx = positions.partition_point(|p| (*p as usize) < lo);
        if idx >= positions.len() {
            return None;
        }
        let pos = positions[idx] as usize;
        if pos >= hi {
            return None;
        }
        self.cursor = Some((value, idx + 1, hi));
        Some(pos)
    }
    fn nextmatch(&mut self) -> Option<usize> {
        let (value, idx, hi) = self.cursor?;
        let positions = self.buckets.get(&value)?;
        if idx >= positions.len() {
            self.cursor = None;
            return None;
        }
        let pos = positions[idx] as usize;
        if pos >= hi {
            self.cursor = None;
            return None;
        }
        self.cursor = Some((value, idx + 1, hi));
        Some(pos)
    }
}

fn bench_old_index(samples: &[i16]) {
    let mut index = OldIndex::new();
    let started = Instant::now();
    index.setup(samples);
    let setup = started.elapsed();
    let mut found = 0usize;
    let started = Instant::now();
    for i in 0..10_000i64 {
        let post = (i * 7919) % samples.len() as i64;
        let value = samples[post as usize];
        if index.getmatch(post, 512, value).is_some() {
            found += 1;
            while index.nextmatch().is_some() {
                found += 1;
            }
        }
    }
    let queries = started.elapsed();
    println!(
        "Old HashMap index: setup={:?} 10k queries={:?} matches={}",
        setup, queries, found
    );
}

fn main() {
    let sectors: usize = std::env::args()
        .nth(1)
        .and_then(|s| s.parse().ok())
        .unwrap_or(2000);
    let samples = make_samples(sectors);
    bench_index(&samples);
    bench_old_index(&samples);

    let bytes = samples_to_bytes(&samples);
    let drive = ImageDrive::from_bytes(bytes);
    let mut p = Paranoia::new(Box::new(drive));
    p.modeset(PARANOIA_MODE_FULL);
    let mut cb = |_: i64, _: i32| {};
    let started = Instant::now();
    let mut got = 0usize;
    for _ in 0..sectors {
        let sector = p.read(&mut cb).expect("read sector");
        got += sector.len();
    }
    let elapsed = started.elapsed();
    let bytes_read = (got * 2) as f64;
    println!(
        "Read: {} sectors, {} samples, {:?}, {:.1} MB/s",
        sectors,
        got,
        elapsed,
        bytes_read / elapsed.as_secs_f64() / 1_000_000.0
    );
    assert_eq!(got, sectors * 1176);
}