diffusionx 0.12.0

A multi-threaded crate for random number generation and stochastic process simulation, with optional GPU acceleration.
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
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171

use crate::{
    FloatExt, XResult,
    gpu::{BM_PTX, CUDA_CTX, GPUMoment},
    simulation::continuous::Bm,
};
use cudarc::{
    driver::{CudaFunction, CudaModule},
    nvrtc::Ptx,
};
use std::sync::{Arc, LazyLock};

static MODULE: LazyLock<XResult<Arc<CudaModule>>> = LazyLock::new(|| {
    let ctx = CUDA_CTX.as_ref()?;
    let module = ctx.load_module(Ptx::from(BM_PTX))?;
    Ok(module)
});

static MEAN_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("mean")?;
    Ok(kernel)
});

static MSD_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("msd")?;
    Ok(kernel)
});

static RAW_MOMENT_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("raw_moment")?;
    Ok(kernel)
});

static FRAC_RAW_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("frac_raw_moment")?;
    Ok(kernel)
});

static CENTRAL_MOMENT_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("central_moment")?;
    Ok(kernel)
});

static FRAC_CENTRAL_KERNEL: LazyLock<XResult<CudaFunction>> = LazyLock::new(|| {
    let module = MODULE.as_ref()?;
    let kernel = module.load_function("frac_central_moment")?;
    Ok(kernel)
});

subscribe_gpu_function!(MODULE, mean, MEAN_KERNEL, (start_position: f32, diffusivity: f32, duration: f32, time_step: f32));

subscribe_gpu_function!(MODULE, msd, MSD_KERNEL, (start_position: f32, diffusivity: f32, duration: f32, time_step: f32));

subscribe_gpu_function!(MODULE, raw_moment, RAW_MOMENT_KERNEL, (start_position: f32, diffusivity: f32, order: i32, duration: f32, time_step: f32));

subscribe_gpu_function!(MODULE, frac_raw_moment, FRAC_RAW_KERNEL, (start_position: f32, diffusivity: f32, order: f32, duration: f32, time_step: f32));

subscribe_central_moment_gpu_function!(MODULE, central_moment, CENTRAL_MOMENT_KERNEL, (start_position: f32, diffusivity: f32, duration: f32, time_step: f32), i32);

subscribe_central_moment_gpu_function!(MODULE, frac_central_moment, FRAC_CENTRAL_KERNEL, (start_position: f32, diffusivity: f32, duration: f32, time_step: f32), f32);

impl<T: FloatExt> GPUMoment for Bm<T> {
    fn central_moment_gpu(
        &self,
        duration: f32,
        order: i32,
        particles: usize,
        time_step: f32,
    ) -> XResult<f32> {
        central_moment(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            duration,
            time_step,
            order,
            particles,
        )
    }

    fn raw_moment_gpu(
        &self,
        duration: f32,
        order: i32,
        particles: usize,
        time_step: f32,
    ) -> XResult<f32> {
        raw_moment(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            order,
            duration,
            time_step,
            particles,
        )
    }

    fn mean_gpu(&self, duration: f32, particles: usize, time_step: f32) -> XResult<f32> {
        mean(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            duration,
            time_step,
            particles,
        )
    }

    fn msd_gpu(&self, duration: f32, particles: usize, time_step: f32) -> XResult<f32> {
        msd(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            duration,
            time_step,
            particles,
        )
    }

    fn frac_central_moment_gpu(
        &self,
        duration: f32,
        order: f32,
        particles: usize,
        time_step: f32,
    ) -> XResult<f32> {
        frac_central_moment(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            duration,
            time_step,
            order,
            particles,
        )
    }

    fn frac_raw_moment_gpu(
        &self,
        duration: f32,
        order: f32,
        particles: usize,
        time_step: f32,
    ) -> XResult<f32> {
        frac_raw_moment(
            self.get_start_position().to_f32().unwrap(),
            self.get_diffusion_coefficient().to_f32().unwrap(),
            order,
            duration,
            time_step,
            particles,
        )
    }
}

#[cfg(test)]
mod tests {
    use crate::gpu::GPUMoment;
    use crate::simulation::continuous::Bm;

    #[test]
    fn test_gpu_moment() {
        let bm = Bm::<f32>::default();
        bm.mean_gpu(1.0, 100, 0.1).unwrap();
        bm.msd_gpu(1.0, 100, 0.1).unwrap();
        bm.raw_moment_gpu(1.0, 2, 100, 0.1).unwrap();
        bm.frac_raw_moment_gpu(1.0, 1.4, 100, 0.1).unwrap();
        bm.central_moment_gpu(1.0, 2, 100, 0.1).unwrap();
        bm.frac_central_moment_gpu(1.0, 1.5, 100, 0.1).unwrap();
    }
}