structured-zstd 0.0.19

Pure Rust zstd implementation — managed fork of ruzstd. Dictionary decompression, no FFI.
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
137
138

#[inline(always)]
pub(crate) fn prefetch_slice(slice: &[u8]) {
    prefetch_slice_impl_l1(slice);
}

#[inline(always)]
pub(crate) fn prefetch_slice_t1(slice: &[u8]) {
    prefetch_slice_impl_t1(slice);
}

#[cfg(target_arch = "x86_64")]
#[inline(always)]
fn prefetch_slice_impl_l1(slice: &[u8]) {
    use core::arch::x86_64::_MM_HINT_T0;
    prefetch_stride_x86_64::<{ _MM_HINT_T0 }>(slice);
}

#[cfg(target_arch = "x86_64")]
#[inline(always)]
fn prefetch_slice_impl_t1(slice: &[u8]) {
    use core::arch::x86_64::_MM_HINT_T1;
    prefetch_stride_x86_64::<{ _MM_HINT_T1 }>(slice);
}

#[cfg(target_arch = "x86_64")]
#[inline(always)]
fn prefetch_stride_x86_64<const HINT: i32>(slice: &[u8]) {
    use core::arch::x86_64::_mm_prefetch;
    const CACHE_LINE: usize = 64;
    const MAX_LINES: usize = 4;

    if slice.len() < CACHE_LINE {
        return;
    }

    let line_count = slice.len().div_ceil(CACHE_LINE).min(MAX_LINES);
    let base = slice.as_ptr();
    for i in 0..line_count {
        let ptr = unsafe { base.add(i * CACHE_LINE) };
        unsafe { _mm_prefetch(ptr.cast(), HINT) };
    }
}

#[cfg(all(target_arch = "x86", target_feature = "sse"))]
#[inline(always)]
fn prefetch_slice_impl_l1(slice: &[u8]) {
    use core::arch::x86::_MM_HINT_T0;
    prefetch_stride_x86::<{ _MM_HINT_T0 }>(slice);
}

#[cfg(all(target_arch = "x86", target_feature = "sse"))]
#[inline(always)]
fn prefetch_slice_impl_t1(slice: &[u8]) {
    use core::arch::x86::_MM_HINT_T1;
    prefetch_stride_x86::<{ _MM_HINT_T1 }>(slice);
}

#[cfg(all(target_arch = "x86", target_feature = "sse"))]
#[inline(always)]
fn prefetch_stride_x86<const HINT: i32>(slice: &[u8]) {
    use core::arch::x86::_mm_prefetch;
    const CACHE_LINE: usize = 64;
    const MAX_LINES: usize = 4;

    if slice.len() < CACHE_LINE {
        return;
    }

    let line_count = slice.len().div_ceil(CACHE_LINE).min(MAX_LINES);
    let base = slice.as_ptr();
    for i in 0..line_count {
        let ptr = unsafe { base.add(i * CACHE_LINE) };
        unsafe { _mm_prefetch(ptr.cast(), HINT) };
    }
}

#[cfg(target_arch = "aarch64")]
#[inline(always)]
fn prefetch_slice_impl_l1(slice: &[u8]) {
    prefetch_stride_aarch64::<true>(slice);
}

#[cfg(target_arch = "aarch64")]
#[inline(always)]
fn prefetch_slice_impl_t1(slice: &[u8]) {
    prefetch_stride_aarch64::<false>(slice);
}

#[cfg(target_arch = "aarch64")]
#[inline(always)]
fn prefetch_stride_aarch64<const L1: bool>(slice: &[u8]) {
    use core::arch::asm;
    const CACHE_LINE: usize = 64;
    const MAX_LINES: usize = 4;

    if slice.len() < CACHE_LINE {
        return;
    }

    let line_count = slice.len().div_ceil(CACHE_LINE).min(MAX_LINES);
    let base = slice.as_ptr();
    for i in 0..line_count {
        let ptr = unsafe { base.add(i * CACHE_LINE) };
        if L1 {
            unsafe {
                asm!(
                    "prfm pldl1keep, [{ptr}]",
                    ptr = in(reg) ptr,
                    options(nostack, preserves_flags, readonly)
                )
            };
        } else {
            unsafe {
                asm!(
                    "prfm pldl2keep, [{ptr}]",
                    ptr = in(reg) ptr,
                    options(nostack, preserves_flags, readonly)
                )
            };
        }
    }
}

#[cfg(not(any(
    target_arch = "x86_64",
    all(target_arch = "x86", target_feature = "sse"),
    target_arch = "aarch64",
)))]
#[inline(always)]
fn prefetch_slice_impl_l1(_slice: &[u8]) {}

#[cfg(not(any(
    target_arch = "x86_64",
    all(target_arch = "x86", target_feature = "sse"),
    target_arch = "aarch64",
)))]
#[inline(always)]
fn prefetch_slice_impl_t1(_slice: &[u8]) {}