fusevm 0.9.9

Language-agnostic bytecode VM with fused superinstructions and Cranelift JIT path
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136

//! Find the crossover point where JIT beats interpreter.
//!
//! Run: cargo bench --features jit --bench jit_crossover

use criterion::{black_box, criterion_group, criterion_main, BenchmarkId, Criterion};
use fusevm::{ChunkBuilder, JitCompiler, Op, VM};

fn build_int_add_chain(n: usize) -> fusevm::Chunk {
    let mut b = ChunkBuilder::new();
    b.emit(Op::LoadInt(0), 1);
    for _ in 0..n {
        b.emit(Op::LoadInt(1), 1);
        b.emit(Op::Add, 1);
    }
    b.build()
}

fn build_bitwise_chain(n: usize) -> fusevm::Chunk {
    let mut b = ChunkBuilder::new();
    b.emit(Op::LoadInt(0x_DEAD_BEEF), 1);
    for _ in 0..n {
        b.emit(Op::LoadInt(0xFF), 1);
        b.emit(Op::BitAnd, 1);
        b.emit(Op::LoadInt(3), 1);
        b.emit(Op::Shl, 1);
        b.emit(Op::LoadInt(0x_1234), 1);
        b.emit(Op::BitXor, 1);
    }
    b.build()
}

fn bench_int_add_crossover(c: &mut Criterion) {
    let jit = JitCompiler::new();
    let mut group = c.benchmark_group("int_add_crossover");

    for &n in &[10, 50, 100, 500, 1000, 5000, 10000] {
        let chunk = build_int_add_chain(n);
        // warm JIT cache
        let _ = jit.try_run_linear(&chunk, &[]);

        group.bench_with_input(BenchmarkId::new("interp", n), &n, |b, _| {
            b.iter(|| {
                let mut vm = VM::new(chunk.clone());
                let r = vm.run();
                black_box(r);
            })
        });
        group.bench_with_input(BenchmarkId::new("jit", n), &n, |b, _| {
            b.iter(|| {
                let r = jit.try_run_linear(black_box(&chunk), &[]);
                black_box(r);
            })
        });
    }
    group.finish();
}

fn bench_bitwise_crossover(c: &mut Criterion) {
    let jit = JitCompiler::new();
    let mut group = c.benchmark_group("bitwise_crossover");

    for &n in &[10, 50, 100, 500, 1000, 5000] {
        let chunk = build_bitwise_chain(n);
        let _ = jit.try_run_linear(&chunk, &[]);

        group.bench_with_input(BenchmarkId::new("interp", n), &n, |b, _| {
            b.iter(|| {
                let mut vm = VM::new(chunk.clone());
                let r = vm.run();
                black_box(r);
            })
        });
        group.bench_with_input(BenchmarkId::new("jit", n), &n, |b, _| {
            b.iter(|| {
                let r = jit.try_run_linear(black_box(&chunk), &[]);
                black_box(r);
            })
        });
    }
    group.finish();
}

/// Isolate JIT overhead: just the cache lookup, no execution
fn bench_jit_overhead(c: &mut Criterion) {
    let jit = JitCompiler::new();
    let mut group = c.benchmark_group("jit_overhead");

    // Tiny chunk: 3 ops
    let tiny = {
        let mut b = ChunkBuilder::new();
        b.emit(Op::LoadInt(1), 1);
        b.emit(Op::LoadInt(2), 1);
        b.emit(Op::Add, 1);
        b.build()
    };
    let _ = jit.try_run_linear(&tiny, &[]);

    group.bench_function("cache_lookup_3ops", |b| {
        b.iter(|| {
            let r = jit.try_run_linear(black_box(&tiny), &[]);
            black_box(r);
        })
    });

    // Medium chunk: 100 ops
    let medium = build_int_add_chain(50);
    let _ = jit.try_run_linear(&medium, &[]);

    group.bench_function("cache_lookup_100ops", |b| {
        b.iter(|| {
            let r = jit.try_run_linear(black_box(&medium), &[]);
            black_box(r);
        })
    });

    // Large chunk: 2000 ops
    let large = build_int_add_chain(1000);
    let _ = jit.try_run_linear(&large, &[]);

    group.bench_function("cache_lookup_2000ops", |b| {
        b.iter(|| {
            let r = jit.try_run_linear(black_box(&large), &[]);
            black_box(r);
        })
    });

    group.finish();
}

criterion_group!(
    benches,
    bench_jit_overhead,
    bench_int_add_crossover,
    bench_bitwise_crossover,
);
criterion_main!(benches);