File: example_utils.hpp

package info (click to toggle)
hipcub 6.4.3-2
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid
  • size: 4,528 kB
  • sloc: cpp: 56,703; python: 564; sh: 365; makefile: 118; xml: 26
file content (658 lines) | stat: -rw-r--r-- 19,577 bytes parent folder | download
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
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
/******************************************************************************
 * Copyright (c) 2011, Duane Merrill.  All rights reserved.
 * Copyright (c) 2011-2018, NVIDIA CORPORATION.  All rights reserved.
 * Modifications Copyright (c) 2021-2024, Advanced Micro Devices, Inc.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Neither the name of the NVIDIA CORPORATION nor the
 *       names of its contributors may be used to endorse or promote products
 *       derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 ******************************************************************************/

#ifndef EXAMPLES_EXAMPLE_UTILS_HPP
#define EXAMPLES_EXAMPLE_UTILS_HPP
#include "mersenne.h"
#include <vector>
#include <sstream>
#include <iostream>

#include <hipcub/util_type.hpp>
#include <hipcub/util_allocator.hpp>
#include <hipcub/iterator/discard_output_iterator.hpp>

#define AssertEquals(a, b) if ((a) != (b)) { std::cerr << "\n(" << __FILE__ << ": " << __LINE__ << ")\n"; exit(1);}

#define HIP_CHECK(condition)                                                           \
    do                                                                                 \
    {                                                                                  \
        hipError_t error = condition;                                                  \
        if(error != hipSuccess)                                                        \
        {                                                                              \
            std::cout << "HIP error: " << error << " line: " << __LINE__ << std::endl; \
            exit(error);                                                               \
        }                                                                              \
    }                                                                                  \
    while(0);

template <typename T>
T CoutCast(T val) { return val; }

int CoutCast(char val) { return val; }

int CoutCast(unsigned char val) { return val; }

int CoutCast(signed char val) { return val; }
/******************************************************************************
 * Command-line parsing functionality
 ******************************************************************************/

/**
 * Utility for parsing command line arguments
 */
struct CommandLineArgs
{

    std::vector<std::string>    keys;
    std::vector<std::string>    values;
    std::vector<std::string>    args;
    hipDeviceProp_t             deviceProp;
    float                       device_giga_bandwidth;
    std::size_t                 device_free_physmem;
    std::size_t                 device_total_physmem;

    /**
     * Constructor
     */
    CommandLineArgs(int argc, char **argv) :
        keys(10),
        values(10)
    {
        using namespace std;

        // Initialize mersenne generator
        unsigned int mersenne_init[4]=  {0x123, 0x234, 0x345, 0x456};
        mersenne::init_by_array(mersenne_init, 4);

        for (int i = 1; i < argc; i++)
        {
            string arg = argv[i];

            if ((arg[0] != '-') || (arg[1] != '-'))
            {
                args.push_back(arg);
                continue;
            }

            string::size_type pos;
            string key, val;
            if ((pos = arg.find('=')) == string::npos) {
                key = string(arg, 2, arg.length() - 2);
                val = "";
            } else {
                key = string(arg, 2, pos - 2);
                val = string(arg, pos + 1, arg.length() - 1);
            }

            keys.push_back(key);
            values.push_back(val);
        }
    }


    /**
     * Checks whether a flag "--<flag>" is present in the commandline
     */
    bool CheckCmdLineFlag(const char* arg_name)
    {
        using namespace std;

        for (std::size_t i = 0; i < keys.size(); ++i)
        {
            if (keys[i] == string(arg_name))
                return true;
        }
        return false;
    }


    /**
     * Returns number of naked (non-flag and non-key-value) commandline parameters
     */
    template <typename T>
    int NumNakedArgs()
    {
        return args.size();
    }


    /**
     * Returns the commandline parameter for a given index (not including flags)
     */
    template <typename T>
    void GetCmdLineArgument(std::size_t index, T &val)
    {
        using namespace std;
        if (index < args.size()) {
            std::istringstream str_stream(args[index]);
            str_stream >> val;
        }
    }

    /**
     * Returns the value specified for a given commandline parameter --<flag>=<value>
     */
    template <typename T>
    void GetCmdLineArgument(const char *arg_name, T &val)
    {
        using namespace std;

        for (std::size_t i = 0; i < keys.size(); ++i)
        {
            if (keys[i] == string(arg_name))
            {
                std::istringstream str_stream(values[i]);
                str_stream >> val;
            }
        }
    }


    /**
     * Returns the values specified for a given commandline parameter --<flag>=<value>,<value>*
     */
    template <typename T>
    void GetCmdLineArguments(const char *arg_name, std::vector<T> &vals)
    {
        using namespace std;

        if (CheckCmdLineFlag(arg_name))
        {
            // Clear any default values
            vals.clear();

            // Recover from multi-value string
            for (std::size_t i = 0; i < keys.size(); ++i)
            {
                if (keys[i] == string(arg_name))
                {
                    string val_string(values[i]);
                    std::istringstream str_stream(val_string);
                    string::size_type old_pos = 0;
                    string::size_type new_pos = 0;

                    // Iterate comma-separated values
                    T val;
                    while ((new_pos = val_string.find(',', old_pos)) != string::npos)
                    {
                        if (new_pos != old_pos)
                        {
                            str_stream.width(new_pos - old_pos);
                            str_stream >> val;
                            vals.push_back(val);
                        }

                        // skip over comma
                        str_stream.ignore(1);
                        old_pos = new_pos + 1;
                    }

                    // Read last value
                    str_stream >> val;
                    vals.push_back(val);
                }
            }
        }
    }


    /**
     * The number of pairs parsed
     */
    int ParsedArgc()
    {
        return (int) keys.size();
    }

    /**
     * Initialize device
     */
    hipError_t DeviceInit(int dev = -1)
    {
        hipError_t error = hipSuccess;

        do
        {
            int deviceCount;
            error = hipGetDeviceCount(&deviceCount);
            if (error) break;

            if (deviceCount == 0) {
                fprintf(stderr, "No devices supporting CUDA.\n");
                exit(1);
            }
            if (dev < 0)
            {
                GetCmdLineArgument("device", dev);
            }
            if ((dev > deviceCount - 1) || (dev < 0))
            {
                dev = 0;
            }

            error = hipSetDevice(dev);
            if (error) break;

            HIP_CHECK(hipMemGetInfo(&device_free_physmem, &device_total_physmem));

            // int ptx_version = 0;
            // error = hipcub::PtxVersion(ptx_version);
            // if (error) break;

            error = hipGetDeviceProperties(&deviceProp, dev);
            if (error) break;

            if (deviceProp.major < 1) {
                fprintf(stderr, "Device does not support Hip.\n");
                exit(1);
            }

            device_giga_bandwidth = float(deviceProp.memoryBusWidth) * deviceProp.memoryClockRate * 2 / 8 / 1000 / 1000;

            if (!CheckCmdLineFlag("quiet"))
            {
                printf(
                        "Using device %d: %s ( SM%d, %d SMs, "
                        "%lld free / %lld total MB physmem, "
                        "%.3f GB/s @ %d kHz mem clock, ECC %s)\n",
                    dev,
                    deviceProp.name,
                    deviceProp.major * 100 + deviceProp.minor * 10,
                    deviceProp.multiProcessorCount,
                    (unsigned long long) device_free_physmem / 1024 / 1024,
                    (unsigned long long) device_total_physmem / 1024 / 1024,
                    device_giga_bandwidth,
                    deviceProp.memoryClockRate,
                    (deviceProp.ECCEnabled) ? "on" : "off");
                fflush(stdout);
            }

        } while (0);

        return error;
    }
};
/******************************************************************************
 * Helper routines for list comparison and display
 ******************************************************************************/


/**
 * Compares the equivalence of two arrays
 */
template <typename S, typename T, typename OffsetT>
int CompareResults(T* computed, S* reference, OffsetT len, bool verbose = true)
{
    for (OffsetT i = 0; i < len; i++)
    {
        if (computed[i] != reference[i])
        {
            if (verbose) std::cout << "INCORRECT: [" << i << "]: "
                << CoutCast(computed[i]) << " != "
                << CoutCast(reference[i]);
            return 1;
        }
    }
    return 0;
}


/**
 * Compares the equivalence of two arrays
 */
template <typename OffsetT>
int CompareResults(float* computed, float* reference, OffsetT len, bool verbose = true)
{
    for (OffsetT i = 0; i < len; i++)
    {
        if (computed[i] != reference[i])
        {
            float difference = std::abs(computed[i]-reference[i]);
            float fraction = difference / std::abs(reference[i]);

            if (fraction > 0.0001)
            {
                if (verbose) std::cout << "INCORRECT: [" << i << "]: "
                    << "(computed) " << CoutCast(computed[i]) << " != "
                    << CoutCast(reference[i]) << " (difference:" << difference << ", fraction: " << fraction << ")";
                return 1;
            }
        }
    }
    return 0;
}


/**
 * Compares the equivalence of two arrays
 */
// template <typename OffsetT>
// int CompareResults(hipcub::NullType* computed, hipcub::NullType* reference, OffsetT len, bool verbose = true)
// {
//     return 0;
// }

/**
 * Compares the equivalence of two arrays
 */
template <typename OffsetT>
int CompareResults(double* computed, double* reference, OffsetT len, bool verbose = true)
{
    for (OffsetT i = 0; i < len; i++)
    {
        if (computed[i] != reference[i])
        {
            double difference = std::abs(computed[i]-reference[i]);
            double fraction = difference / std::abs(reference[i]);

            if (fraction > 0.0001)
            {
                if (verbose) std::cout << "INCORRECT: [" << i << "]: "
                    << CoutCast(computed[i]) << " != "
                    << CoutCast(reference[i]) << " (difference:" << difference << ", fraction: " << fraction << ")";
                return 1;
            }
        }
    }
    return 0;
}


// /**
//  * Verify the contents of a device array match those
//  * of a host array
//  */
// int CompareDeviceResults(
//     hipcub::NullType */* h_reference */,
//     hipcub::NullType */* d_data */,
//     std::size_t /* num_items */,
//     bool /* verbose */ = true,
//     bool /* display_data */ = false)
// {
//     return 0;
// }

/**
 * Verify the contents of a device array match those
 * of a host array
 */
// template <typename S, typename OffsetT>
// int CompareDeviceResults(
//     S *h_reference,
//     hipcub::DiscardOutputIterator<OffsetT> d_data,
//     std::size_t num_items,
//     bool verbose = true,
//     bool display_data = false)
// {
//     return 0;
// }

/**
 * Verify the contents of a device array match those
 * of a host array
 */
template <typename S, typename T>
int CompareDeviceResults(
    S *h_reference,
    T *d_data,
    std::size_t num_items,
    bool verbose = true,
    bool display_data = false)
{
    // Allocate array on host
    T *h_data = (T*) malloc(num_items * sizeof(T));

    // Copy data back
    HIP_CHECK(hipMemcpy(h_data, d_data, sizeof(T) * num_items, hipMemcpyDeviceToHost));

    // Display data
    if (display_data)
    {
        printf("Reference:\n");
        for (std::size_t i = 0; i < num_items; i++)
        {
            std::cout << CoutCast(h_reference[i]) << ", ";
        }
        printf("\n\nComputed:\n");
        for (std::size_t i = 0; i < num_items; i++)
        {
            std::cout << CoutCast(h_data[i]) << ", ";
        }
        printf("\n\n");
    }

    // Check
    int retval = CompareResults(h_data, h_reference, num_items, verbose);

    // Cleanup
    if (h_data) free(h_data);

    return retval;
}


/**
 * Verify the contents of a device array match those
 * of a device array
 */
template <typename T>
int CompareDeviceDeviceResults(
    T *d_reference,
    T *d_data,
    std::size_t num_items,
    bool verbose = true,
    bool display_data = false)
{
    // Allocate array on host
    T *h_reference = (T*) malloc(num_items * sizeof(T));
    T *h_data = (T*) malloc(num_items * sizeof(T));

    // Copy data back
    HIP_CHECK(hipMemcpy(h_reference, d_reference, sizeof(T) * num_items, hipMemcpyDeviceToHost));
    HIP_CHECK(hipMemcpy(h_data, d_data, sizeof(T) * num_items, hipMemcpyDeviceToHost));

    // Display data
    if (display_data) {
        printf("Reference:\n");
        for (std::size_t i = 0; i < num_items; i++)
        {
            std::cout << CoutCast(h_reference[i]) << ", ";
        }
        printf("\n\nComputed:\n");
        for (std::size_t i = 0; i < num_items; i++)
        {
            std::cout << CoutCast(h_data[i]) << ", ";
        }
        printf("\n\n");
    }

    // Check
    int retval = CompareResults(h_data, h_reference, num_items, verbose);

    // Cleanup
    if (h_reference) free(h_reference);
    if (h_data) free(h_data);

    return retval;
}

/**
 * Print the contents of a host array
 */
template <typename InputIteratorT>
void DisplayResults(
    InputIteratorT h_data,
    std::size_t num_items)
{
    // Display data
    for (std::size_t i = 0; i < num_items; i++)
    {
        std::cout << CoutCast(h_data[i]) << ", ";
    }
    printf("\n");
}


int g_num_rand_samples = 0;
/**
 * Generates random keys.
 *
 * We always take the second-order byte from rand() because the higher-order
 * bits returned by rand() are commonly considered more uniformly distributed
 * than the lower-order bits.
 *
 * We can decrease the entropy level of keys by adopting the technique
 * of Thearling and Smith in which keys are computed from the bitwise AND of
 * multiple random samples:
 *
 * entropy_reduction    | Effectively-unique bits per key
 * -----------------------------------------------------
 * -1                   | 0
 * 0                    | 32
 * 1                    | 25.95 (81%)
 * 2                    | 17.41 (54%)
 * 3                    | 10.78 (34%)
 * 4                    | 6.42 (20%)
 * ...                  | ...
 *
 */
template <typename K>
void RandomBits(
    K &key,
    int entropy_reduction = 0,
    int begin_bit = 0,
    int end_bit = sizeof(K) * 8)
{
    const int NUM_BYTES = sizeof(K);
    const int WORD_BYTES = sizeof(unsigned int);
    const int NUM_WORDS = (NUM_BYTES + WORD_BYTES - 1) / WORD_BYTES;

    unsigned int word_buff[NUM_WORDS];

    if (entropy_reduction == -1)
    {
        memset((void *) &key, 0, sizeof(key));
        return;
    }

    if (end_bit < 0)
        end_bit = sizeof(K) * 8;

    while (true)
    {
        // Generate random word_buff
        for (int j = 0; j < NUM_WORDS; j++)
        {
            int current_bit = j * WORD_BYTES * 8;

            unsigned int word = 0xffffffff;
            word &= 0xffffffff << std::max(0, begin_bit - current_bit);
            word &= 0xffffffff >> std::max(0, (current_bit + (WORD_BYTES * 8)) - end_bit);

            for (int i = 0; i <= entropy_reduction; i++)
            {
                // Grab some of the higher bits from rand (better entropy, supposedly)
                word &= mersenne::genrand_int32();
                g_num_rand_samples++;
            }

            word_buff[j] = word;
        }

        memcpy(&key, word_buff, sizeof(K));

        K copy = key;
        if HIPCUB_IF_CONSTEXPR(std::is_floating_point<K>::value)
#ifndef _WIN32
            if(!std::isnan(copy))
#else
            // MSVC STL is missing the integral overload of std::isnan
            // https://github.com/microsoft/STL/issues/3400
            // They're right, the integral overloads were removed from the standard
            // after national body comments. std::isnan is supposed to behave _identically_
            // to the isnan CRT macro, where ISO C 7.12.3 states
            // > In the synopses in this subclause, real-floating indicates that the
            //   argument shall be an expression of real floating type.
            if(!std::isnan(static_cast<double>(copy)))
#endif
                break; // avoids NaNs when generating random floating point numbers
    }
}

/// Randomly select number between [0:max)
template <typename T>
T RandomValue(T max)
{
    unsigned int bits;
    unsigned int max_int = (unsigned int) -1;
    do {
        RandomBits(bits);
    } while (bits == max_int);

    return (T) ((double(bits) / double(max_int)) * double(max));
}

struct GpuTimer
{
    hipEvent_t start;
    hipEvent_t stop;

    GpuTimer()
    {
        HIP_CHECK(hipEventCreate(&start));
        HIP_CHECK(hipEventCreate(&stop));
    }

    ~GpuTimer()
    {
        HIP_CHECK(hipEventDestroy(start));
        HIP_CHECK(hipEventDestroy(stop));
    }

    void Start()
    {
        HIP_CHECK(hipEventRecord(start, 0));
    }

    void Stop()
    {
        HIP_CHECK(hipEventRecord(stop, 0));
    }

    float ElapsedMillis()
    {
        float elapsed;
        HIP_CHECK(hipEventSynchronize(stop));
        HIP_CHECK(hipEventElapsedTime(&elapsed, start, stop));
        return elapsed;
    }
};

#endif