bioem_cuda.cu 15.6 KB
Newer Older
Pilar Cossio's avatar
License    
Pilar Cossio committed
1
2
3
4
/* ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
        < BioEM software for Bayesian inference of Electron Microscopy images>
            Copyright (C) 2014 Pilar Cossio, David Rohr and Gerhard Hummer.
            Max Planck Institute of Biophysics, Frankfurt, Germany.
5

Pilar Cossio's avatar
License    
Pilar Cossio committed
6
7
8
9
                See license statement for terms of distribution.

   ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/

10
11
12
13
14
15
16
17
18
19
20
21
#define BIOEM_GPUCODE

#if defined(_WIN32)
#include <windows.h>
#endif

#include <iostream>
using namespace std;

#include "bioem_cuda_internal.h"
#include "bioem_algorithm.h"

22
23
24
25
26
27
28
#define checkCudaErrors(error) \
{ \
	if ((error) != cudaSuccess) \
	{ \
		printf("CUDA Error %d / %s (%s: %d)\n", error, cudaGetErrorString(error), __FILE__, __LINE__); \
		exit(1); \
	} \
29
30
}

David Rohr's avatar
David Rohr committed
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
static const char *cufftGetErrorStrung(cufftResult error)
{
    switch (error)
    {
        case CUFFT_SUCCESS:
            return "CUFFT_SUCCESS";

        case CUFFT_INVALID_PLAN:
            return "CUFFT_INVALID_PLAN";

        case CUFFT_ALLOC_FAILED:
            return "CUFFT_ALLOC_FAILED";

        case CUFFT_INVALID_TYPE:
            return "CUFFT_INVALID_TYPE";

        case CUFFT_INVALID_VALUE:
            return "CUFFT_INVALID_VALUE";

        case CUFFT_INTERNAL_ERROR:
            return "CUFFT_INTERNAL_ERROR";

        case CUFFT_EXEC_FAILED:
            return "CUFFT_EXEC_FAILED";

        case CUFFT_SETUP_FAILED:
            return "CUFFT_SETUP_FAILED";

        case CUFFT_INVALID_SIZE:
            return "CUFFT_INVALID_SIZE";

        case CUFFT_UNALIGNED_DATA:
            return "CUFFT_UNALIGNED_DATA";
    }
    return "UNKNOWN";
}

68
69
70
71
bioem_cuda::bioem_cuda()
{
	deviceInitialized = 0;
	GPUAlgo = getenv("GPUALGO") == NULL ? 2 : atoi(getenv("GPUALGO"));
72
73
	GPUAsync = getenv("GPUASYNC") == NULL ? 1 : atoi(getenv("GPUASYNC"));
	GPUWorkload = getenv("GPUWORKLOAD") == NULL ? 100 : atoi(getenv("GPUWORKLOAD"));
74
	GPUDualStream = getenv("GPUDUALSTREAM") == NULL ? 1 : atoi(getenv("GPUDUALSTREAM"));
75
76
77
78
79
80
81
}

bioem_cuda::~bioem_cuda()
{
	deviceExit();
}

82
__global__ void compareRefMap_kernel(const int iOrient, const int iConv, const myfloat_t* pMap, bioem_Probability pProb, const bioem_param_device param, const bioem_RefMap_Mod RefMap, const int cent_x, const int cent_y, const int maxRef)
83
84
{
	const int iRefMap = myBlockIdxX * myBlockDimX + myThreadIdxX;
85
	if (iRefMap < maxRef)
86
	{
87
		compareRefMap<0>(iRefMap, iOrient, iConv, pMap, pProb, param, RefMap, cent_x, cent_y);
88
89
90
	}
}

91
__global__ void compareRefMapShifted_kernel(const int iOrient, const int iConv, const myfloat_t* pMap, bioem_Probability pProb, const bioem_param_device param, const bioem_RefMap_Mod RefMap, const int maxRef)
92
93
{
	const int iRefMap = myBlockIdxX * myBlockDimX + myThreadIdxX;
94
	if (iRefMap < maxRef)
95
	{
96
		compareRefMapShifted<1>(iRefMap, iOrient, iConv, pMap, pProb, param, RefMap);
97
98
99
	}
}

100
101
102
103
104
105
__global__ void cudaZeroMem(void* ptr, size_t size)
{
	int* myptr = (int*) ptr;
	int mysize = size / sizeof(int);
	int myid = myBlockDimX * myBlockIdxX + myThreadIdxX;
	int mygrid = myBlockDimX * myGridDimX;
106
	for (int i = myid; i < mysize; i += mygrid) myptr[i] = 0;
107
108
}

109
__global__ void compareRefMapLoopShifts_kernel(const int iOrient, const int iConv, const myfloat_t* pMap, bioem_Probability pProb, const bioem_param_device param, const bioem_RefMap RefMap, const int blockoffset, const int nShifts, const int nShiftBits, const int maxRef)
110
111
112
113
114
115
116
117
118
{
	const size_t myid = (myBlockIdxX + blockoffset) * myBlockDimX + myThreadIdxX;
	const int iRefMap = myid >> (nShiftBits << 1);
	const int myRef = myThreadIdxX >> (nShiftBits << 1);
	const int myShiftIdx = (myid >> nShiftBits) & (nShifts - 1);
	const int myShiftIdy = myid & (nShifts - 1);
	const int myShift = myid & (nShifts * nShifts - 1);
	const int cent_x = myShiftIdx * param.GridSpaceCenter - param.maxDisplaceCenter;
	const int cent_y = myShiftIdy * param.GridSpaceCenter - param.maxDisplaceCenter;
119

120
	const bool threadActive = myShiftIdx < nShifts && myShiftIdy < nShifts && iRefMap < maxRef;
121

122
	compareRefMap<2>(iRefMap, iOrient, iConv, pMap, pProb, param, RefMap, cent_x, cent_y, myShift, nShifts * nShifts, myRef, threadActive);
123
124
}

125
__global__ void multComplexMap(const mycomplex_t* convmap, const mycomplex_t* refmap, mycuComplex_t* out, const int NumberPixelsTotal, const int MapSize, const int NumberMaps, const int Offset)
126
127
{
	if (myBlockIdxX >= NumberMaps) return;
128
	const mycuComplex_t* myin = (mycuComplex_t*) &refmap[(myBlockIdxX + Offset) * MapSize];
129
	const mycuComplex_t* myconv = (mycuComplex_t*) convmap;
130
	mycuComplex_t* myout = &out[myBlockIdxX * MapSize];
131
	for(int i = myThreadIdxX; i < NumberPixelsTotal; i += myBlockDimX)
132
	{
133
134
135
136
137
138
139
		mycuComplex_t val;
		const mycuComplex_t conv = myconv[i];
		const mycuComplex_t in = myin[i];

		val.x = conv.x * in.x + conv.y * in.y;
		val.y = conv.y * in.x - conv.x * in.y;
		myout[i] = val;
140
141
142
	}
}

143
__global__ void cuDoRefMapsFFT(const int iOrient, const int iConv, const myfloat_t* lCC, const myfloat_t sumC, const myfloat_t sumsquareC, bioem_Probability pProb, const bioem_param_device param, const bioem_RefMap RefMap, const int maxRef, const int Offset)
144
{
145
	if (myBlockIdxX * myBlockDimX + myThreadIdxX >= maxRef) return;
146
147
	const int iRefMap = myBlockIdxX * myBlockDimX + myThreadIdxX + Offset;
	const myfloat_t* mylCC = &lCC[(myBlockIdxX * myBlockDimX + myThreadIdxX) * param.NumberPixels * param.NumberPixels];
148
	doRefMapFFT(iRefMap, iOrient, iConv, mylCC, sumC, sumsquareC, pProb, param, RefMap);
149
150
}

151
152
153
154
155
156
157
158
159
160
161
162
163
template <class T> static inline T divup(T num, T divider) {return((num + divider - 1) / divider);}
static inline bool IsPowerOf2(int x) {return ((x > 0) && ((x & (x - 1)) == 0));}
#if defined(_WIN32)
static inline int ilog2 (int value)
{
	DWORD index;
	_BitScanReverse (&index, value);
	return(value);
}
#else
static inline int ilog2(int value) {return 31 - __builtin_clz(value);}
#endif

164
int bioem_cuda::compareRefMaps(int iOrient, int iConv, const myfloat_t* conv_map, mycomplex_t* localmultFFT, myfloat_t sumC, myfloat_t sumsquareC, const int startMap)
165
{
166
167
168
169
170
	if (startMap)
	{
		cout << "Error startMap not implemented for GPU Code\n";
		exit(1);
	}
171
172
173
174
	if (GPUAsync)
	{
		checkCudaErrors(cudaEventSynchronize(cudaEvent[iConv & 1]));
	}
175

176
	if (FFTAlgo)
177
	{
178
		memcpy(&pConvMapFFT_Host[(iConv & 1) * param.FFTMapSize], localmultFFT, param.FFTMapSize * sizeof(mycomplex_t));
179
180
181
182
183
184
		checkCudaErrors(cudaMemcpyAsync(&pConvMapFFT[(iConv & 1) * param.FFTMapSize], &pConvMapFFT_Host[(iConv & 1) * param.FFTMapSize], param.FFTMapSize * sizeof(mycomplex_t), cudaMemcpyHostToDevice, cudaStream[GPUAsync ? 2 : 0]));
		if (GPUAsync)
		{
			checkCudaErrors(cudaEventRecord(cudaEvent[2], cudaStream[2]));
			checkCudaErrors(cudaStreamWaitEvent(cudaStream[0], cudaEvent[2], 0));
		}
185
		if (GPUDualStream)
186
		{
187
188
189
190
191
192
			checkCudaErrors(cudaEventRecord(cudaFFTEvent[0], cudaStream[0]));
			checkCudaErrors(cudaStreamWaitEvent(cudaStream[1], cudaFFTEvent[0], 0));
		}
		for (int i = 0, j = 0; i < maxRef; i += CUDA_FFTS_AT_ONCE, j++)
		{
			if (!GPUDualStream) j = 0;
193
			const int num = min(CUDA_FFTS_AT_ONCE, maxRef - i);
194
195
			multComplexMap<<<num, CUDA_THREAD_COUNT, 0, cudaStream[j & 1]>>>(&pConvMapFFT[(iConv & 1) * param.FFTMapSize], pRefMapsFFT, pFFTtmp2[j & 1], param.param_device.NumberPixels * param.param_device.NumberFFTPixels1D, param.FFTMapSize, num, i);
			cufftResult err = mycufftExecC2R(i + CUDA_FFTS_AT_ONCE > maxRef ? plan[1][j & 1] : plan[0][j & 1], pFFTtmp2[j & 1], pFFTtmp[j & 1]);
David Rohr's avatar
David Rohr committed
196
			if (err != CUFFT_SUCCESS)
197
			{
David Rohr's avatar
David Rohr committed
198
				cout << "Error running CUFFT " << cufftGetErrorStrung(err) << "\n";
199
200
				exit(1);
			}
201
			cuDoRefMapsFFT<<<divup(num, CUDA_THREAD_COUNT), CUDA_THREAD_COUNT, 0, cudaStream[j & 1]>>>(iOrient, iConv, pFFTtmp[j & 1], sumC, sumsquareC, pProb_device, param.param_device, *gpumap, num, i);
202
		}
203
		checkCudaErrors(cudaGetLastError());
204
205
206
207
208
		if (GPUDualStream)
		{
			checkCudaErrors(cudaEventRecord(cudaFFTEvent[1], cudaStream[1]));
			checkCudaErrors(cudaStreamWaitEvent(cudaStream[0], cudaFFTEvent[1], 0));
		}
209
210
211
	}
	else
	{
212
		checkCudaErrors(cudaMemcpyAsync(pConvMap_device[iConv & 1], conv_map, sizeof(myfloat_t) * RefMap.refMapSize, cudaMemcpyHostToDevice, cudaStream[0]));
213
214

		if (GPUAlgo == 2) //Loop over shifts
215
		{
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
			const int nShifts = 2 * param.param_device.maxDisplaceCenter / param.param_device.GridSpaceCenter + 1;
			if (!IsPowerOf2(nShifts))
			{
				cout << "Invalid number of displacements, no power of two\n";
				exit(1);
			}
			if (CUDA_THREAD_COUNT % (nShifts * nShifts))
			{
				cout << "CUDA Thread count (" << CUDA_THREAD_COUNT << ") is no multiple of number of shifts (" << (nShifts * nShifts) << ")\n";
				exit(1);
			}
			if (nShifts > CUDA_MAX_SHIFT_REDUCE)
			{
				cout << "Too many displacements for CUDA reduction\n";
				exit(1);
			}
			const int nShiftBits = ilog2(nShifts);
			size_t totalBlocks = divup((size_t) maxRef * (size_t) nShifts * (size_t) nShifts, (size_t) CUDA_THREAD_COUNT);
			size_t nBlocks = CUDA_BLOCK_COUNT;
235
			for (size_t i = 0; i < totalBlocks; i += nBlocks)
236
			{
237
				compareRefMapLoopShifts_kernel<<<min(nBlocks, totalBlocks - i), CUDA_THREAD_COUNT, (CUDA_THREAD_COUNT * 2 + CUDA_THREAD_COUNT / (nShifts * nShifts) * 4) * sizeof(myfloat_t), cudaStream[0] >>> (iOrient, iConv, pConvMap_device[iConv & 1], pProb_device, param.param_device, *gpumap, i, nShifts, nShiftBits, maxRef);
238
			}
239
		}
240
		else if (GPUAlgo == 1) //Split shifts in multiple kernels
241
		{
242
			for (int cent_x = -param.param_device.maxDisplaceCenter; cent_x <= param.param_device.maxDisplaceCenter; cent_x = cent_x + param.param_device.GridSpaceCenter)
243
			{
244
				for (int cent_y = -param.param_device.maxDisplaceCenter; cent_y <= param.param_device.maxDisplaceCenter; cent_y = cent_y + param.param_device.GridSpaceCenter)
245
				{
246
					compareRefMap_kernel<<<divup(maxRef, CUDA_THREAD_COUNT), CUDA_THREAD_COUNT, 0, cudaStream[0]>>> (iOrient, iConv, pConvMap_device[iConv & 1], pProb_device, param.param_device, *pRefMap_device_Mod, cent_x, cent_y, maxRef);
247
248
				}
			}
249
		}
250
		else if (GPUAlgo == 0) //All shifts in one kernel
251
		{
252
			compareRefMapShifted_kernel<<<divup(maxRef, CUDA_THREAD_COUNT), CUDA_THREAD_COUNT, 0, cudaStream[0]>>> (iOrient, iConv, pConvMap_device[iConv & 1], pProb_device, param.param_device, *pRefMap_device_Mod, maxRef);
253
		}
254
		else
255
		{
256
257
			cout << "Invalid GPU Algorithm selected\n";
			exit(1);
258
		}
259
	}
260
261
	if (GPUWorkload < 100)
	{
262
		bioem::compareRefMaps(iOrient, iConv, conv_map, localmultFFT, sumC, sumsquareC, maxRef);
263
	}
264
265
	if (GPUAsync)
	{
266
		checkCudaErrors(cudaEventRecord(cudaEvent[iConv & 1], cudaStream[0]));
267
	}
268
269
	else
	{
270
		checkCudaErrors(cudaStreamSynchronize(cudaStream[0]));
271
272
273
274
275
276
277
	}
	return(0);
}

int bioem_cuda::deviceInit()
{
	deviceExit();
278

279
280
	if (FFTAlgo) GPUAlgo = 2;

281
282
283
284
285
	gpumap = new bioem_RefMap;
	memcpy(gpumap, &RefMap, sizeof(bioem_RefMap));
	if (FFTAlgo == 0)
	{
		checkCudaErrors(cudaMalloc(&maps, sizeof(myfloat_t) * RefMap.ntotRefMap * RefMap.refMapSize));
286
287
288
289
290
291
292
293
294
295
296
297
298

		if (GPUAlgo == 0 || GPUAlgo == 1)
		{
			pRefMap_device_Mod = (bioem_RefMap_Mod*) gpumap;
			bioem_RefMap_Mod* RefMapGPU = new bioem_RefMap_Mod;
			RefMapGPU->init(RefMap);
			checkCudaErrors(cudaMemcpy(maps, RefMapGPU->maps, sizeof(myfloat_t) * RefMap.ntotRefMap * RefMap.refMapSize, cudaMemcpyHostToDevice));
			delete RefMapGPU;
		}
		else
		{
			checkCudaErrors(cudaMemcpy(maps, RefMap.maps, sizeof(myfloat_t) * RefMap.ntotRefMap * RefMap.refMapSize, cudaMemcpyHostToDevice));
		}
299
300
301
302
303
304
305
306
307
	}
	checkCudaErrors(cudaMalloc(&sum, sizeof(myfloat_t) * RefMap.ntotRefMap));
	checkCudaErrors(cudaMemcpy(sum, RefMap.sum_RefMap, sizeof(myfloat_t) * RefMap.ntotRefMap, cudaMemcpyHostToDevice));
	checkCudaErrors(cudaMalloc(&sumsquare, sizeof(myfloat_t) * RefMap.ntotRefMap));
	checkCudaErrors(cudaMemcpy(sumsquare, RefMap.sumsquare_RefMap, sizeof(myfloat_t) * RefMap.ntotRefMap, cudaMemcpyHostToDevice));
	gpumap->maps = maps;
	gpumap->sum_RefMap = sum;
	gpumap->sumsquare_RefMap = sumsquare;

308
309
310
	pProb_device = pProb;
	checkCudaErrors(cudaMalloc(&pProb_device.ptr, pProb_device.get_size(RefMap.ntotRefMap, param.nTotGridAngles)));
	pProb_device.set_pointers();
311
	for (int i = 0; i < 2; i++)
312
	{
313
		checkCudaErrors(cudaStreamCreate(&cudaStream[i]));
314
		checkCudaErrors(cudaEventCreate(&cudaEvent[i]));
315
		checkCudaErrors(cudaEventCreate(&cudaFFTEvent[i]));
316
		checkCudaErrors(cudaMalloc(&pConvMap_device[i], sizeof(myfloat_t) * RefMap.refMapSize));
317
	}
318
319
320
321
322
	if (GPUAsync)
	{
		checkCudaErrors(cudaStreamCreate(&cudaStream[2]));
		checkCudaErrors(cudaEventCreate(&cudaEvent[2]));
	}
323

324
325
	if (FFTAlgo)
	{
326
		checkCudaErrors(cudaMalloc(&pRefMapsFFT, RefMap.ntotRefMap * param.FFTMapSize * sizeof(mycomplex_t)));
327
328
329
330
		checkCudaErrors(cudaMalloc(&pFFTtmp2[0], CUDA_FFTS_AT_ONCE * param.FFTMapSize * 2 * sizeof(mycomplex_t)));
		checkCudaErrors(cudaMalloc(&pFFTtmp[0], CUDA_FFTS_AT_ONCE * param.param_device.NumberPixels * param.param_device.NumberPixels * 2 * sizeof(myfloat_t)));
		pFFTtmp2[1] = pFFTtmp2[0] + CUDA_FFTS_AT_ONCE * param.FFTMapSize;
		pFFTtmp[1] = pFFTtmp[0] + CUDA_FFTS_AT_ONCE * param.param_device.NumberPixels * param.param_device.NumberPixels;
331
		checkCudaErrors(cudaMalloc(&pConvMapFFT, param.FFTMapSize * sizeof(mycomplex_t) * 2));
332
		checkCudaErrors(cudaHostAlloc(&pConvMapFFT_Host, param.FFTMapSize * sizeof(mycomplex_t) * 2, 0));
333
		checkCudaErrors(cudaMemcpy(pRefMapsFFT, RefMap.RefMapsFFT, RefMap.ntotRefMap * param.FFTMapSize * sizeof(mycomplex_t), cudaMemcpyHostToDevice));
334
335
	}

336
337
338
339
340
341
342
	deviceInitialized = 1;
	return(0);
}

int bioem_cuda::deviceExit()
{
	if (deviceInitialized == 0) return(0);
343

344

345
	cudaFree(pProb_device.ptr);
346
347
	cudaFree(sum);
	cudaFree(sumsquare);
348
	for (int i = 0; i < 2; i++)
349
	{
350
		cudaStreamDestroy(cudaStream[i]);
351
		cudaEventDestroy(cudaEvent[i]);
352
		cudaEventDestroy(cudaFFTEvent[i]);
353
		cudaFree(pConvMap_device[i]);
354
	}
355
356
357
358
	if (FFTAlgo)
	{
		cudaFree(pRefMapsFFT);
		cudaFree(pConvMapFFT);
359
		cudaFreeHost(pConvMapFFT_Host);
360
361
		cudaFree(pFFTtmp[0]);
		cudaFree(pFFTtmp2[0]);
362
	}
363
364
365
366
367
368
369
370
	else
	{
		cudaFree(maps);
	}
	if (GPUAlgo == 0 || GPUAlgo == 1)
	{
		cudaFree(pRefMap_device_Mod);
	}
371
372
373
374
375
376
	if (GPUAsync)
	{
		cudaStreamDestroy(cudaStream[2]);
		cudaEventDestroy(cudaEvent[2]);
	}

377
	delete gpumap;
378
	cudaThreadExit();
379

380
381
382
383
384
385
	deviceInitialized = 0;
	return(0);
}

int bioem_cuda::deviceStartRun()
{
386
	maxRef = GPUWorkload >= 100 ? RefMap.ntotRefMap : ((size_t) RefMap.ntotRefMap * (size_t) GPUWorkload / 100);
387

388
	cudaMemcpyAsync(pProb_device.ptr, pProb.ptr, pProb.get_size(RefMap.ntotRefMap, param.nTotGridAngles), cudaMemcpyHostToDevice, cudaStream[0]);
389
390
391

	if (FFTAlgo)
	{
392
		for (int j = 0;j < 2;j++)
393
		{
394
			for (int i = 0; i < 2; i++)
395
			{
396
				if (i && maxRef % CUDA_FFTS_AT_ONCE == 0) continue;
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
				int n[2] = {param.param_device.NumberPixels, param.param_device.NumberPixels};
				if (cufftPlanMany(&plan[i][j], 2, n, NULL, 1, param.FFTMapSize, NULL, 1, 0, MY_CUFFT_C2R, i ? (maxRef % CUDA_FFTS_AT_ONCE) : CUDA_FFTS_AT_ONCE) != CUFFT_SUCCESS)
				{
					cout << "Error planning CUFFT\n";
					exit(1);
				}
				if (cufftSetCompatibilityMode(plan[i][j], CUFFT_COMPATIBILITY_NATIVE) != CUFFT_SUCCESS)
				{
					cout << "Error planning CUFFT compatibility\n";
					exit(1);
				}
				if (cufftSetStream(plan[i][j], cudaStream[j]) != CUFFT_SUCCESS)
				{
					cout << "Error setting CUFFT stream\n";
					exit(1);
				}
413
			}
414
			if (!GPUDualStream) break;
415
416
		}
	}
417
418
419
420
421
	return(0);
}

int bioem_cuda::deviceFinishRun()
{
422
	if (GPUAsync) cudaStreamSynchronize(cudaStream[0]);
423
	cudaMemcpyAsync(pProb.ptr, pProb_device.ptr, pProb.get_size(RefMap.ntotRefMap, param.nTotGridAngles), cudaMemcpyDeviceToHost, cudaStream[0]);
424

425
426
	if (FFTAlgo)
	{
427
428
		for (int j = 0;j < 2;j++)
		{
429
430
431
432
433
			for (int i = 0; i < 2; i++)
			{
				if (i && maxRef % CUDA_FFTS_AT_ONCE == 0) continue;
				cufftDestroy(plan[i][j]);
			}
434
435
			if (!GPUDualStream) break;
		}
436
437
	}

438
439
440
	return(0);
}

441
442
443
444
445
446
447
448
449
450
451
452
void* bioem_cuda::malloc_device_host(size_t size)
{
	void* ptr;
	checkCudaErrors(cudaHostAlloc(&ptr, size, 0));
	return(ptr);
}

void bioem_cuda::free_device_host(void* ptr)
{
	cudaFreeHost(ptr);
}

453
454
455
456
bioem* bioem_cuda_create()
{
	return new bioem_cuda;
}