Neko 0.9.99
A portable framework for high-order spectral element flow simulations
Loading...
Searching...
No Matches
pipecg_kernel.h
Go to the documentation of this file.
1#ifndef __KRYLOV_PIPECG_KERNEL_H__
2#define __KRYLOV_PIPECG_KERNEL_H__
3/*
4 Copyright (c) 2021-2022, The Neko Authors
5 All rights reserved.
6
7 Redistribution and use in source and binary forms, with or without
8 modification, are permitted provided that the following conditions
9 are met:
10
11 * Redistributions of source code must retain the above copyright
12 notice, this list of conditions and the following disclaimer.
13
14 * Redistributions in binary form must reproduce the above
15 copyright notice, this list of conditions and the following
16 disclaimer in the documentation and/or other materials provided
17 with the distribution.
18
19 * Neither the name of the authors nor the names of its
20 contributors may be used to endorse or promote products derived
21 from this software without specific prior written permission.
22
23 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
24 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
25 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
26 FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
27 COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
28 INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
29 BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
30 LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
31 CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
33 ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 POSSIBILITY OF SUCH DAMAGE.
35*/
36
37#include <math/bcknd/device/cuda/math_kernel.h>
38
42template< typename T >
43__global__ void cg_update_xp_kernel(T * __restrict__ x,
44 T * __restrict__ p,
45 T ** __restrict__ u,
46 const T * alpha,
47 const T * beta,
48 const int p_cur,
49 const int p_space,
50 const int n) {
51
52 const int idx = blockIdx.x * blockDim.x + threadIdx.x;
53 const int str = blockDim.x * gridDim.x;
54
55
56 for (int i = idx; i < n; i+= str) {
57 T tmp = 0.0;
58 int p_prev = p_space;
59 for (int j = 0; j < p_cur; j ++) {
60 p[i] = beta[j]*p[i] + u[p_prev][i];
61 tmp += alpha[j]*p[i];
62 p_prev = j;
63 }
64 x[i] += tmp;
65 u[p_space][i] = u[p_space-1][i];
66 }
67}
71template< typename T >
72__global__ void pipecg_vecops_kernel(T * __restrict__ p,
73 T * __restrict__ q,
74 T * __restrict__ r,
75 T * __restrict__ s,
76 T * __restrict__ u1,
77 T * __restrict__ u2,
78 T * __restrict__ w,
79 T * __restrict__ z,
80 T * __restrict__ ni,
81 T * __restrict__ mi,
82 const T alpha,
83 const T beta,
84 const T * mult,
85 T * buf_h1,
86 T * buf_h2,
87 T * buf_h3,
88 const int n) {
89
90 const int idx = blockIdx.x * blockDim.x + threadIdx.x;
91 const int str = blockDim.x * gridDim.x;
92
93 const unsigned int lane = threadIdx.x % warpSize;
94 const unsigned int wid = threadIdx.x / warpSize;
95
96 __shared__ T buf1[32];
97 __shared__ T buf2[32];
98 __shared__ T buf3[32];
99 T tmp1 = 0.0;
100 T tmp2 = 0.0;
101 T tmp3 = 0.0;
102
103 for (int i = idx; i < n; i+= str) {
104 z[i] = beta * z[i] + ni[i];
105 q[i] = beta * q[i] + mi[i];
106 s[i] = beta * s[i] + w[i];
107 r[i] = r[i] - alpha * s[i];
108 u2[i] = u1[i] - alpha * q[i];
109 w[i] = w[i] - alpha * z[i];
110 tmp1 = tmp1 + r[i] * mult[i] * u2[i];
111 tmp2 = tmp2 + w[i] * mult[i] * u2[i];
112 tmp3 = tmp3 + r[i] * mult[i] * r[i];
113
114 }
115
116 tmp1 = reduce_warp<T>(tmp1);
117 tmp2 = reduce_warp<T>(tmp2);
118 tmp3 = reduce_warp<T>(tmp3);
119 if (lane == 0) {
120 buf1[wid] = tmp1;
121 buf2[wid] = tmp2;
122 buf3[wid] = tmp3;
123 }
124 __syncthreads();
125
126 tmp1 = (threadIdx.x < blockDim.x / warpSize) ? buf1[lane] : 0;
127 tmp2 = (threadIdx.x < blockDim.x / warpSize) ? buf2[lane] : 0;
128 tmp3 = (threadIdx.x < blockDim.x / warpSize) ? buf3[lane] : 0;
129 if (wid == 0) {
130 tmp1 = reduce_warp<T>(tmp1);
131 tmp2 = reduce_warp<T>(tmp2);
132 tmp3 = reduce_warp<T>(tmp3);
133 }
134
135 if (threadIdx.x == 0) {
136 buf_h1[blockIdx.x] = tmp1;
137 buf_h2[blockIdx.x] = tmp2;
138 buf_h3[blockIdx.x] = tmp3;
139 }
140
141}
142
143#endif // __KRYLOV_PIPECG_KERNEL_H__
w
__global__ void T *__restrict__ T *__restrict__ const T *__restrict__ const T *__restrict__ const T *__restrict__ w
Definition ax_helm_full_kernel.h:48
i
const int i
Definition ax_helm_full_kernel.h:95
u
__global__ void T *__restrict__ T *__restrict__ const T *__restrict__ u
Definition ax_helm_full_kernel.h:46
j
const int j
Definition ax_helm_full_kernel.h:94
__syncthreads
__syncthreads()
x
__global__ void const T *__restrict__ x
Definition cdtp_kernel.h:106
dirichlet_apply_scalar_kernel
__global__ void dirichlet_apply_scalar_kernel(const int *__restrict__ msk, T *__restrict__ x, const T g, const int m)
Definition dirichlet_kernel.h:42
math_kernel.h
pipecg_vecops_kernel
__global__ void pipecg_vecops_kernel(T *__restrict__ p, T *__restrict__ q, T *__restrict__ r, T *__restrict__ s, T *__restrict__ u1, T *__restrict__ u2, T *__restrict__ w, T *__restrict__ z, T *__restrict__ ni, T *__restrict__ mi, const T alpha, const T beta, const T *mult, T *buf_h1, T *buf_h2, T *buf_h3, const int n)
Definition pipecg_kernel.h:72
cg_update_xp_kernel
__global__ void cg_update_xp_kernel(T *__restrict__ x, T *__restrict__ p, T **__restrict__ u, const T *alpha, const T *beta, const int p_cur, const int p_space, const int n)
Definition pipecg_kernel.h:43