rlx-cuda 0.2.2

NVIDIA CUDA backend — cuBLAS for matmul + NVRTC-compiled kernels for everything else, via the pure-Rust `cudarc` crate.
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
139
140
141
142
143
144
145
146
147
148
149

// RLX — versatile ML compiler + runtime.
// Copyright (C) 2026 Eugene Hauptmann, Nataliya Kosmyna.
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, version 3.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.
//! Native CUDA Gaussian splat forward (RLX-owned kernel).

#![cfg(feature = "native-splat")]

use crate::kernels::{dispatch_grid_2d, gaussian_splat_rasterize_kernel};
use cudarc::driver::{CudaSlice, CudaStream, LaunchConfig};
use slang_splat_ref::RenderParams;
use slang_splat_ref::native_prep::{
    PreparedRaster, camera_and_background_from_meta, prepare_raster, scene_from_slices,
};
use std::sync::Arc;

fn dispatch_prepared(
    stream: &Arc<CudaStream>,
    arena: &mut CudaSlice<f32>,
    prep: &PreparedRaster,
    dst_off: u32,
) {
    let d_ca = stream
        .memcpy_stod(&prep.color_alpha)
        .expect("splat native: stod color_alpha");
    let d_valid = stream
        .memcpy_stod(&prep.valid)
        .expect("splat native: stod valid");
    let d_pl = stream
        .memcpy_stod(&prep.pos_local)
        .expect("splat native: stod pos_local");
    let d_inv = stream
        .memcpy_stod(&prep.inv_scale)
        .expect("splat native: stod inv_scale");
    let d_quat = stream
        .memcpy_stod(&prep.quat)
        .expect("splat native: stod quat");
    let d_sorted = stream
        .memcpy_stod(&prep.sorted_values)
        .expect("splat native: stod sorted");
    let d_ranges = stream
        .memcpy_stod(&prep.tile_ranges)
        .expect("splat native: stod ranges");
    let d_rays = stream
        .memcpy_stod(&prep.rays)
        .expect("splat native: stod rays");
    let params = prep.params;

    let kernel = gaussian_splat_rasterize_kernel(stream.context());
    let (grid, block) = dispatch_grid_2d(params.width, params.height, 8, 8);
    let cfg = LaunchConfig {
        grid_dim: grid,
        block_dim: block,
        shared_mem_bytes: 0,
    };
    let mut launcher = stream.launch_builder(&kernel.function);
    launcher
        .arg(arena)
        .arg(&dst_off)
        .arg(&d_ca)
        .arg(&d_valid)
        .arg(&d_pl)
        .arg(&d_inv)
        .arg(&d_quat)
        .arg(&d_sorted)
        .arg(&d_ranges)
        .arg(&d_rays)
        .arg(&params);
    unsafe {
        launcher
            .launch(cfg)
            .expect("rlx-cuda: gaussian_splat_rasterize launch failed");
    }
}

#[allow(clippy::too_many_arguments)]
pub fn run_gaussian_splat_render_native(
    stream: &Arc<CudaStream>,
    buffer: &mut CudaSlice<f32>,
    arena_size_bytes: usize,
    positions_off: usize,
    positions_len: usize,
    scales_off: usize,
    scales_len: usize,
    rotations_off: usize,
    rotations_len: usize,
    opacities_off: usize,
    opacities_len: usize,
    colors_off: usize,
    colors_len: usize,
    sh_coeffs_off: usize,
    sh_coeffs_len: usize,
    meta_off: usize,
    dst_off: usize,
    width: u32,
    height: u32,
    tile_size: u32,
    radius_scale: f32,
    alpha_cutoff: f32,
    max_splat_steps: u32,
    transmittance_threshold: f32,
    max_list_entries: u32,
) {
    let n_f32 = arena_size_bytes / 4;
    stream
        .synchronize()
        .expect("rlx-cuda splat native: pre-sync failed");
    let mut host = vec![0f32; n_f32];
    stream
        .memcpy_dtoh(&buffer.slice(..), &mut host)
        .expect("rlx-cuda splat native: dtoh failed");

    let sl = |off: usize, len: usize| -> &[f32] { &host[off..off + len] };
    let meta_len = 23.min(host.len().saturating_sub(meta_off));
    let scene = scene_from_slices(
        sl(positions_off, positions_len),
        sl(scales_off, scales_len),
        sl(rotations_off, rotations_len),
        sl(opacities_off, opacities_len),
        sl(colors_off, colors_len),
        sl(sh_coeffs_off, sh_coeffs_len),
    );
    let (camera, background) = camera_and_background_from_meta(sl(meta_off, meta_len));
    let render_params = RenderParams {
        width,
        height,
        tile_size,
        radius_scale,
        alpha_cutoff,
        max_splat_steps,
        transmittance_threshold,
        max_list_entries,
    };
    let prep = prepare_raster(&scene, &camera, background, &render_params);
    dispatch_prepared(stream, buffer, &prep, dst_off as u32);
    stream
        .synchronize()
        .expect("rlx-cuda splat native: post-sync failed");
}