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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
|
__device__ unsigned int INNAME(lfc_retirementCount) = {0};
__global__ void INNAME(integrate_mul_kernel)(
const Tgpu *a_G, int nG,
const LFVolume_gpu *volume_W,
const int *volume_WMi_gpu,
const int *WMi_gpu,
int WMimax,
int q,
Tgpu *out, int block_out,
Tgpu *results, int Mcount, int nM, int nvec)
{
int yy = gridDim.y / Mcount;
int bloy = blockIdx.y / yy;
int block = blockIdx.y - bloy * yy;
unsigned int tid = threadIdx.x;
unsigned int gridSize = REDUCE_LFC_THREADS * gridDim.x;
unsigned int i_b = blockIdx.x * (REDUCE_LFC_THREADS) + tid;
extern __shared__ Tgpu Zgpu(sdata)[];
// perform first level of reduction,
// reading from global memory, writing to shared memory
a_G += nG * block;
for (int vv=0; vv < WMi_gpu[bloy]; vv++) {
const LFVolume_gpu *v = &volume_W[volume_WMi_gpu[bloy * WMimax
+ vv]];
int *nGBcum = v->nGBcum;
#ifdef GPU_USE_COMPLEX
Tgpu phase = v->phase_k[q];
#endif
int len_A_gm = v->len_A_gm;
Tgpu *out_t = out + v->M * block_out + block * nM * block_out;
int a_ind, ai=0, acum=0;
if (i_b < len_A_gm) {
int bi = v->nB;
int ci;
int bcum = nGBcum[bi];
int ccum;
while (bi - ai > 1) {
ci = ai + 1 + (bi - ai - 2) * (i_b - acum) / (bcum - acum);
ccum = nGBcum[ci];
if (ccum <= i_b) {
ai = ci;
acum = ccum;
} else {
bi = ci;
bcum = ccum;
}
}
a_ind = v->GB1[ai] + i_b - acum;
}
for (int i=0; i < nvec; i++) {
Tgpu a_Gv;
double *A_gm2 = v->A_gm;
Tgpu *out_t2 = out_t;
if (i_b < len_A_gm) {
#ifdef GPU_USE_COMPLEX
a_Gv = MULTT(a_G[i * nG + a_ind], phase);
#else
a_Gv = a_G[i * nG + a_ind];
#endif
}
for (int m=0; m < v->nm; m++) {
Tgpu mySum = MAKED(0);
if (i_b < len_A_gm) {
mySum = MULTD(a_Gv, A_gm2[i_b]);
}
if (len_A_gm > gridSize) {
unsigned int i_bb = i_b + gridSize;
int aai = ai;
int aacum = acum;
while (i_bb < len_A_gm) {
int bi = v->nB;
int ci;
int bcum = nGBcum[bi];
int ccum;
while (bi - aai > 1) {
ci = aai + 1 + (bi - aai - 2) * (i_bb - aacum)
/ (bcum - aacum);
ccum = nGBcum[ci];
if (ccum <= i_bb) {
aai = ci;
aacum = ccum;
} else {
bi = ci;
bcum = ccum;
}
}
#ifdef GPU_USE_COMPLEX
IADD(mySum, MULTD(MULTT(a_G[i * nG + v->GB1[aai]
+ i_bb - aacum], phase),
A_gm2[i_bb]));
#else
IADD(mySum, MULTD(a_G[i * nG + v->GB1[aai] + i_bb
- aacum],
A_gm2[i_bb]));
#endif
i_bb += gridSize;
}
}
Zgpu(sdata)[tid] = mySum;
__syncthreads();
if (REDUCE_LFC_THREADS >= 512) {
if (tid < 256) {
Zgpu(sdata)[tid] = mySum
= ADD(mySum,
Zgpu(sdata)[tid + 256]);
}
__syncthreads();
}
if (REDUCE_LFC_THREADS >= 256) {
if (tid < 128) {
Zgpu(sdata)[tid] = mySum
= ADD(mySum,
Zgpu(sdata)[tid + 128]);
}
__syncthreads();
}
if (REDUCE_LFC_THREADS >= 128) {
if (tid < 64) {
Zgpu(sdata)[tid] = mySum
= ADD(mySum,
Zgpu(sdata)[tid + 64]);
}
__syncthreads();
}
if (tid < 32) {
volatile Tgpu *smem = Zgpu(sdata);
#ifdef GPU_USE_COMPLEX
if (REDUCE_LFC_THREADS >= 64) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 32].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 32].y;
}
if (REDUCE_LFC_THREADS >= 32) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 16].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 16].y;
}
if (REDUCE_LFC_THREADS >= 16) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 8].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 8].y;
}
if (REDUCE_LFC_THREADS >= 8) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 4].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 4].y;
}
if (REDUCE_LFC_THREADS >= 4) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 2].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 2].y;
}
if (REDUCE_LFC_THREADS >= 2) {
smem[tid].x = mySum.x = mySum.x + smem[tid + 1].x;
smem[tid].y = mySum.y = mySum.y + smem[tid + 1].y;
}
#else
if (REDUCE_LFC_THREADS >= 64)
smem[tid] = mySum = ADD(mySum, smem[tid + 32]);
if (REDUCE_LFC_THREADS >= 32)
smem[tid] = mySum = ADD(mySum, smem[tid + 16]);
if (REDUCE_LFC_THREADS >= 16)
smem[tid] = mySum = ADD(mySum, smem[tid + 8]);
if (REDUCE_LFC_THREADS >= 8)
smem[tid] = mySum = ADD(mySum, smem[tid + 4]);
if (REDUCE_LFC_THREADS >= 4)
smem[tid] = mySum = ADD(mySum, smem[tid + 2]);
if (REDUCE_LFC_THREADS >= 2)
smem[tid] = mySum = ADD(mySum, smem[tid + 1]);
#endif
}
// write result for this block to global mem
if (tid==0) {
if (vv==0)
out_t2[blockIdx.x] = Zgpu(sdata)[0];
else
IADD(out_t2[blockIdx.x], Zgpu(sdata)[0]);
}
A_gm2 += len_A_gm;
out_t2 += block_out;
__syncthreads();
}
out_t += nM * block_out;
}
}
if (gridDim.x==1) {
__shared__ bool amLast;
__threadfence();
if (tid == 0) {
unsigned int ticket = atomicInc(&INNAME(lfc_retirementCount),
gridDim.y);
amLast = (ticket == gridDim.y - 1);
}
__syncthreads();
if ((amLast)) {
for (int i=tid; i < nM * yy * nvec; i += blockDim.x) {
results[i] = out[i * block_out];
}
INNAME(lfc_retirementCount) = 0;
}
}
}
|
