jxr-cuda 0.2.0

CUDA reconstruction backend for the JXR decoder
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

// SPDX-License-Identifier: MIT OR Apache-2.0

use cudarc::driver::CudaSlice;

use crate::DenseCudaBatchLayout;

/// Immutable completed CUDA-resident JPEG XR output.
pub struct ResidentCudaImage {
    pub(crate) buffer: CudaSlice<u8>,
    layout: jxr_core::SurfaceLayout,
    report: jxr_core::DecodeReport,
}

impl core::fmt::Debug for ResidentCudaImage {
    fn fmt(&self, formatter: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        formatter
            .debug_struct("ResidentCudaImage")
            .field("device_ordinal", &self.buffer.ordinal())
            .field("layout", &self.layout)
            .field("report", &self.report)
            .finish_non_exhaustive()
    }
}

impl ResidentCudaImage {
    pub(crate) fn from_buffer(
        buffer: CudaSlice<u8>,
        layout: jxr_core::SurfaceLayout,
        report: jxr_core::DecodeReport,
    ) -> Self {
        Self {
            buffer,
            layout,
            report,
        }
    }

    /// Validated device surface layout.
    #[must_use]
    pub const fn layout(&self) -> &jxr_core::SurfaceLayout {
        &self.layout
    }

    /// Route report for the completed decode.
    #[must_use]
    pub const fn report(&self) -> &jxr_core::DecodeReport {
        &self.report
    }

    /// CUDA device ordinal owning the allocation.
    #[must_use]
    pub fn device_ordinal(&self) -> usize {
        self.buffer.ordinal()
    }

    /// Read-only access to the completed device allocation.
    #[must_use]
    pub const fn device_buffer(&self) -> &CudaSlice<u8> {
        &self.buffer
    }
}

/// Completed homogeneous batch in one CUDA allocation.
pub struct CudaResidentBatch {
    pub(crate) buffer: CudaSlice<u8>,
    layout: DenseCudaBatchLayout,
    reports: Vec<jxr_core::DecodeReport>,
}

impl core::fmt::Debug for CudaResidentBatch {
    fn fmt(&self, formatter: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        formatter
            .debug_struct("CudaResidentBatch")
            .field("device_ordinal", &self.buffer.ordinal())
            .field("layout", &self.layout)
            .field("report_count", &self.reports.len())
            .finish_non_exhaustive()
    }
}

impl CudaResidentBatch {
    pub(crate) fn from_buffer(
        buffer: CudaSlice<u8>,
        layout: DenseCudaBatchLayout,
        reports: Vec<jxr_core::DecodeReport>,
    ) -> Self {
        Self {
            buffer,
            layout,
            reports,
        }
    }

    /// Validated dense layout.
    #[must_use]
    pub const fn layout(&self) -> &DenseCudaBatchLayout {
        &self.layout
    }

    /// Ordered decode reports.
    #[must_use]
    pub fn reports(&self) -> &[jxr_core::DecodeReport] {
        &self.reports
    }

    /// CUDA device ordinal owning the allocation.
    #[must_use]
    pub fn device_ordinal(&self) -> usize {
        self.buffer.ordinal()
    }

    /// Read-only access to the completed dense allocation.
    #[must_use]
    pub const fn device_buffer(&self) -> &CudaSlice<u8> {
        &self.buffer
    }
}