forked from junhyungshin/mapd-core-fpd
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathResultSetTest.cpp
2885 lines (2694 loc) · 112 KB
/
ResultSetTest.cpp
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
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Copyright 2017 MapD Technologies, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/**
* @file ResultSetTest.cpp
* @author Alex Suhan <[email protected]>
* @brief Unit tests for the result set interface.
*
* Copyright (c) 2014 MapD Technologies, Inc. All rights reserved.
*/
#include "ResultSetTestUtils.h"
#include "../QueryEngine/ResultRows.h"
#include "../QueryEngine/ResultSet.h"
#include "../QueryEngine/RuntimeFunctions.h"
#include "../StringDictionary/StringDictionary.h"
#include <boost/make_unique.hpp>
#include <glog/logging.h>
#include <gtest/gtest.h>
#include <algorithm>
#include <queue>
#include <random>
TEST(Construct, Allocate) {
std::vector<TargetInfo> target_infos;
QueryMemoryDescriptor query_mem_desc{};
ResultSet result_set(
target_infos, ExecutorDeviceType::CPU, query_mem_desc, std::make_shared<RowSetMemoryOwner>(), nullptr);
result_set.allocateStorage();
}
namespace {
class NumberGenerator {
public:
virtual int64_t getNextValue() = 0;
virtual void reset() = 0;
};
class EvenNumberGenerator : public NumberGenerator {
public:
EvenNumberGenerator() : crt_(0) {}
int64_t getNextValue() override {
const auto crt = crt_;
crt_ += 2;
return crt;
}
void reset() override { crt_ = 0; }
private:
int64_t crt_;
};
class ReverseOddOrEvenNumberGenerator : public NumberGenerator {
public:
ReverseOddOrEvenNumberGenerator(const int64_t init) : crt_(init), init_(init) {}
int64_t getNextValue() override {
const auto crt = crt_;
crt_ -= 2;
return crt;
}
void reset() override { crt_ = init_; }
private:
int64_t crt_;
int64_t init_;
};
void write_int(int8_t* slot_ptr, const int64_t v, const size_t slot_bytes) {
switch (slot_bytes) {
case 4:
*reinterpret_cast<int32_t*>(slot_ptr) = v;
break;
case 8:
*reinterpret_cast<int64_t*>(slot_ptr) = v;
break;
default:
CHECK(false);
}
}
void write_fp(int8_t* slot_ptr, const int64_t v, const size_t slot_bytes) {
switch (slot_bytes) {
case 4: {
float fi = v;
*reinterpret_cast<int32_t*>(slot_ptr) = *reinterpret_cast<const int32_t*>(may_alias_ptr(&fi));
break;
}
case 8: {
double di = v;
*reinterpret_cast<int64_t*>(slot_ptr) = *reinterpret_cast<const int64_t*>(may_alias_ptr(&di));
break;
}
default:
CHECK(false);
}
}
int8_t* fill_one_entry_no_collisions(int8_t* buff,
const QueryMemoryDescriptor& query_mem_desc,
const int64_t v,
const std::vector<TargetInfo>& target_infos,
const bool empty,
const bool null_val = false) {
size_t target_idx = 0;
int8_t* slot_ptr = buff;
int64_t vv = 0;
for (const auto& target_info : target_infos) {
CHECK_LT(target_idx, query_mem_desc.agg_col_widths.size());
const auto slot_bytes = query_mem_desc.agg_col_widths[target_idx].actual;
CHECK_LE(target_info.sql_type.get_size(), slot_bytes);
bool isNullable = !target_info.sql_type.get_notnull();
if (target_info.agg_kind == kCOUNT) {
if (empty || null_val) {
vv = 0;
} else {
vv = v;
}
} else {
if (isNullable && target_info.skip_null_val && null_val) {
vv = inline_int_null_val(target_info.sql_type);
} else {
vv = v;
}
}
if (empty) {
write_int(slot_ptr, query_mem_desc.keyless_hash ? 0 : vv, slot_bytes);
} else {
if (target_info.sql_type.is_integer()) {
write_int(slot_ptr, vv, slot_bytes);
} else if (target_info.sql_type.is_string()) {
write_int(slot_ptr, -(vv + 2), slot_bytes);
} else {
CHECK(target_info.sql_type.is_fp());
write_fp(slot_ptr, vv, slot_bytes);
}
}
slot_ptr += slot_bytes;
if (target_info.agg_kind == kAVG) {
const auto count_slot_bytes = query_mem_desc.agg_col_widths[target_idx + 1].actual;
if (empty) {
write_int(slot_ptr, query_mem_desc.keyless_hash ? 0 : 0, count_slot_bytes);
} else {
if (isNullable && target_info.skip_null_val && null_val) {
write_int(slot_ptr, 0, count_slot_bytes); // count of elements should be set to 0 for elements with null_val
} else {
write_int(slot_ptr, 1, count_slot_bytes); // count of elements in the group is 1 - good enough for testing
}
}
slot_ptr += count_slot_bytes;
}
target_idx = advance_slot(target_idx, target_info, false);
}
return slot_ptr;
}
void fill_one_entry_one_col(int8_t* ptr1,
const int8_t compact_sz1,
int8_t* ptr2,
const int8_t compact_sz2,
int64_t v,
const TargetInfo& target_info,
const bool empty_entry,
const bool null_val = false) {
int64_t vv = 0;
if (target_info.agg_kind == kCOUNT) {
if (empty_entry || null_val) {
vv = 0;
} else {
vv = v;
}
} else {
bool isNullable = !target_info.sql_type.get_notnull();
if (isNullable && target_info.skip_null_val && null_val) {
vv = inline_int_null_val(target_info.sql_type);
} else {
if (empty_entry && (target_info.agg_kind == kAVG)) {
vv = 0;
} else {
vv = v;
}
}
}
CHECK(ptr1);
switch (compact_sz1) {
case 8:
if (target_info.sql_type.is_fp()) {
double di = vv;
*reinterpret_cast<int64_t*>(ptr1) = *reinterpret_cast<const int64_t*>(may_alias_ptr(&di));
} else {
*reinterpret_cast<int64_t*>(ptr1) = vv;
}
break;
case 4:
CHECK(!target_info.sql_type.is_fp());
*reinterpret_cast<int32_t*>(ptr1) = vv;
break;
default:
CHECK(false);
}
if (target_info.is_agg && target_info.agg_kind == kAVG) {
CHECK(ptr2);
switch (compact_sz2) {
case 8:
*reinterpret_cast<int64_t*>(ptr2) = (empty_entry ? *ptr1 : 1);
break;
case 4:
*reinterpret_cast<int32_t*>(ptr2) = (empty_entry ? *ptr1 : 1);
break;
default:
CHECK(false);
}
}
}
void fill_one_entry_one_col(int64_t* value_slot,
const int64_t v,
const TargetInfo& target_info,
const size_t entry_count,
const bool empty_entry = false,
const bool null_val = false) {
auto ptr1 = reinterpret_cast<int8_t*>(value_slot);
int8_t* ptr2{nullptr};
if (target_info.agg_kind == kAVG) {
ptr2 = reinterpret_cast<int8_t*>(&value_slot[entry_count]);
}
fill_one_entry_one_col(ptr1, 8, ptr2, 8, v, target_info, empty_entry, null_val);
}
int8_t* advance_to_next_columnar_key_buff(int8_t* key_ptr,
const QueryMemoryDescriptor& query_mem_desc,
const size_t key_idx) {
CHECK(!query_mem_desc.keyless_hash);
CHECK_LT(key_idx, query_mem_desc.group_col_widths.size());
auto new_key_ptr = key_ptr + query_mem_desc.entry_count * query_mem_desc.group_col_widths[key_idx];
if (!query_mem_desc.key_column_pad_bytes.empty()) {
CHECK_LT(key_idx, query_mem_desc.key_column_pad_bytes.size());
new_key_ptr += query_mem_desc.key_column_pad_bytes[key_idx];
}
return new_key_ptr;
}
void write_key(const int64_t k, int8_t* ptr, const int8_t key_bytes) {
switch (key_bytes) {
case 8: {
*reinterpret_cast<int64_t*>(ptr) = k;
break;
}
case 4: {
*reinterpret_cast<int32_t*>(ptr) = k;
break;
}
default:
CHECK(false);
}
}
void fill_storage_buffer_perfect_hash_colwise(int8_t* buff,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& generator) {
const auto key_component_count = get_key_count_for_descriptor(query_mem_desc);
CHECK(query_mem_desc.output_columnar);
// initialize the key buffer(s)
auto col_ptr = buff;
for (size_t key_idx = 0; key_idx < key_component_count; ++key_idx) {
auto key_entry_ptr = col_ptr;
const auto key_bytes = query_mem_desc.group_col_widths[key_idx];
CHECK_EQ(8, key_bytes);
for (size_t i = 0; i < query_mem_desc.entry_count; ++i) {
if (i % 2 == 0) {
const auto v = generator.getNextValue();
write_key(v, key_entry_ptr, key_bytes);
} else {
write_key(EMPTY_KEY_64, key_entry_ptr, key_bytes);
}
key_entry_ptr += key_bytes;
}
col_ptr = advance_to_next_columnar_key_buff(col_ptr, query_mem_desc, key_idx);
generator.reset();
}
// initialize the value buffer(s)
size_t slot_idx = 0;
for (const auto& target_info : target_infos) {
auto col_entry_ptr = col_ptr;
const auto col_bytes = query_mem_desc.agg_col_widths[slot_idx].compact;
for (size_t i = 0; i < query_mem_desc.entry_count; ++i) {
int8_t* ptr2{nullptr};
if (target_info.agg_kind == kAVG) {
ptr2 = col_entry_ptr + query_mem_desc.entry_count * col_bytes;
}
if (i % 2 == 0) {
const auto gen_val = generator.getNextValue();
const auto val = target_info.sql_type.is_string() ? -(gen_val + 2) : gen_val;
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
query_mem_desc.agg_col_widths[slot_idx + 1].compact,
val,
target_info,
false);
} else {
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
query_mem_desc.agg_col_widths[slot_idx + 1].compact,
query_mem_desc.keyless_hash ? 0 : 0xdeadbeef,
target_info,
true);
}
col_entry_ptr += col_bytes;
}
col_ptr = advance_to_next_columnar_target_buff(col_ptr, query_mem_desc, slot_idx);
if (target_info.is_agg && target_info.agg_kind == kAVG) {
col_ptr = advance_to_next_columnar_target_buff(col_ptr, query_mem_desc, slot_idx + 1);
}
slot_idx = advance_slot(slot_idx, target_info, false);
generator.reset();
}
}
void fill_storage_buffer_perfect_hash_rowwise(int8_t* buff,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& generator) {
const auto key_component_count = get_key_count_for_descriptor(query_mem_desc);
CHECK(!query_mem_desc.output_columnar);
auto key_buff = buff;
for (size_t i = 0; i < query_mem_desc.entry_count; ++i) {
if (i % 2 == 0) {
const auto v = generator.getNextValue();
auto key_buff_i64 = reinterpret_cast<int64_t*>(key_buff);
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
*key_buff_i64++ = v;
}
auto entries_buff = reinterpret_cast<int8_t*>(key_buff_i64);
key_buff = fill_one_entry_no_collisions(entries_buff, query_mem_desc, v, target_infos, false);
} else {
auto key_buff_i64 = reinterpret_cast<int64_t*>(key_buff);
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
*key_buff_i64++ = EMPTY_KEY_64;
}
auto entries_buff = reinterpret_cast<int8_t*>(key_buff_i64);
key_buff = fill_one_entry_no_collisions(entries_buff, query_mem_desc, 0xdeadbeef, target_infos, true);
}
}
}
void fill_storage_buffer_baseline_colwise(int8_t* buff,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& generator,
const size_t step) {
CHECK(query_mem_desc.output_columnar);
const auto key_component_count = get_key_count_for_descriptor(query_mem_desc);
const auto i64_buff = reinterpret_cast<int64_t*>(buff);
const auto target_slot_count = get_slot_count(target_infos);
const auto slot_to_target = get_slot_to_target_mapping(target_infos);
for (size_t i = 0; i < query_mem_desc.entry_count; ++i) {
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
i64_buff[key_offset_colwise(i, key_comp_idx, query_mem_desc.entry_count)] = EMPTY_KEY_64;
}
for (size_t target_slot = 0; target_slot < target_slot_count; ++target_slot) {
auto target_it = slot_to_target.find(target_slot);
CHECK(target_it != slot_to_target.end());
const auto& target_info = target_infos[target_it->second];
i64_buff[slot_offset_colwise(i, target_slot, key_component_count, query_mem_desc.entry_count)] =
(target_info.agg_kind == kCOUNT ? 0 : 0xdeadbeef);
}
}
for (size_t i = 0; i < query_mem_desc.entry_count; i += step) {
const auto gen_val = generator.getNextValue();
std::vector<int64_t> key(key_component_count, gen_val);
auto value_slots = get_group_value_columnar(i64_buff, query_mem_desc.entry_count, &key[0], key.size());
CHECK(value_slots);
for (const auto& target_info : target_infos) {
const auto val = target_info.sql_type.is_string() ? -(gen_val + step) : gen_val;
fill_one_entry_one_col(value_slots, val, target_info, query_mem_desc.entry_count);
value_slots += query_mem_desc.entry_count;
if (target_info.agg_kind == kAVG) {
value_slots += query_mem_desc.entry_count;
}
}
}
}
void fill_storage_buffer_baseline_rowwise(int8_t* buff,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& generator,
const size_t step) {
const auto key_component_count = get_key_count_for_descriptor(query_mem_desc);
const auto i64_buff = reinterpret_cast<int64_t*>(buff);
const auto target_slot_count = get_slot_count(target_infos);
const auto slot_to_target = get_slot_to_target_mapping(target_infos);
for (size_t i = 0; i < query_mem_desc.entry_count; ++i) {
const auto first_key_comp_offset = key_offset_rowwise(i, key_component_count, target_slot_count);
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
i64_buff[first_key_comp_offset + key_comp_idx] = EMPTY_KEY_64;
}
for (size_t target_slot = 0; target_slot < target_slot_count; ++target_slot) {
auto target_it = slot_to_target.find(target_slot);
CHECK(target_it != slot_to_target.end());
const auto& target_info = target_infos[target_it->second];
i64_buff[slot_offset_rowwise(i, target_slot, key_component_count, target_slot_count)] =
(target_info.agg_kind == kCOUNT ? 0 : 0xdeadbeef);
}
}
for (size_t i = 0; i < query_mem_desc.entry_count; i += step) {
const auto v = generator.getNextValue();
std::vector<int64_t> key(key_component_count, v);
auto value_slots = get_group_value(i64_buff,
query_mem_desc.entry_count,
&key[0],
key.size(),
sizeof(int64_t),
key_component_count + target_slot_count,
nullptr);
CHECK(value_slots);
fill_one_entry_baseline(value_slots, v, target_infos);
}
}
void fill_storage_buffer(int8_t* buff,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& generator,
const size_t step) {
switch (query_mem_desc.hash_type) {
case GroupByColRangeType::OneColKnownRange:
case GroupByColRangeType::MultiColPerfectHash: {
if (query_mem_desc.output_columnar) {
fill_storage_buffer_perfect_hash_colwise(buff, target_infos, query_mem_desc, generator);
} else {
fill_storage_buffer_perfect_hash_rowwise(buff, target_infos, query_mem_desc, generator);
}
break;
}
case GroupByColRangeType::MultiCol: {
if (query_mem_desc.output_columnar) {
fill_storage_buffer_baseline_colwise(buff, target_infos, query_mem_desc, generator, step);
} else {
fill_storage_buffer_baseline_rowwise(buff, target_infos, query_mem_desc, generator, step);
}
break;
}
default:
CHECK(false);
}
CHECK(buff);
}
// TODO(alex): allow 4 byte keys
/* descriptor with small entry_count to simplify testing and debugging */
QueryMemoryDescriptor perfect_hash_one_col_desc_small(const std::vector<TargetInfo>& target_infos,
const int8_t num_bytes) {
QueryMemoryDescriptor query_mem_desc{};
query_mem_desc.hash_type = GroupByColRangeType::OneColKnownRange;
query_mem_desc.min_val = 0;
query_mem_desc.max_val = 19;
query_mem_desc.has_nulls = false;
query_mem_desc.group_col_widths.emplace_back(8);
for (const auto& target_info : target_infos) {
const auto slot_bytes = std::max(num_bytes, static_cast<int8_t>(target_info.sql_type.get_size()));
if (target_info.agg_kind == kAVG) {
CHECK(target_info.is_agg);
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.entry_count = query_mem_desc.max_val - query_mem_desc.min_val + 1;
return query_mem_desc;
}
QueryMemoryDescriptor perfect_hash_one_col_desc(const std::vector<TargetInfo>& target_infos,
const int8_t num_bytes,
const size_t min_val,
const size_t max_val) {
QueryMemoryDescriptor query_mem_desc{};
query_mem_desc.hash_type = GroupByColRangeType::OneColKnownRange;
query_mem_desc.min_val = min_val;
query_mem_desc.max_val = max_val;
query_mem_desc.has_nulls = false;
query_mem_desc.group_col_widths.emplace_back(8);
for (const auto& target_info : target_infos) {
const auto slot_bytes = std::max(num_bytes, static_cast<int8_t>(target_info.sql_type.get_size()));
if (target_info.agg_kind == kAVG) {
CHECK(target_info.is_agg);
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.entry_count = query_mem_desc.max_val - query_mem_desc.min_val + 1;
return query_mem_desc;
}
QueryMemoryDescriptor perfect_hash_two_col_desc(const std::vector<TargetInfo>& target_infos, const int8_t num_bytes) {
QueryMemoryDescriptor query_mem_desc{};
query_mem_desc.hash_type = GroupByColRangeType::MultiColPerfectHash;
query_mem_desc.min_val = 0;
query_mem_desc.max_val = 36;
query_mem_desc.has_nulls = false;
query_mem_desc.group_col_widths.emplace_back(8);
query_mem_desc.group_col_widths.emplace_back(8);
for (const auto& target_info : target_infos) {
const auto slot_bytes = std::max(num_bytes, static_cast<int8_t>(target_info.sql_type.get_size()));
if (target_info.agg_kind == kAVG) {
CHECK(target_info.is_agg);
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.entry_count = query_mem_desc.max_val;
return query_mem_desc;
}
QueryMemoryDescriptor baseline_hash_two_col_desc_large(const std::vector<TargetInfo>& target_infos,
const int8_t num_bytes) {
QueryMemoryDescriptor query_mem_desc{};
query_mem_desc.hash_type = GroupByColRangeType::MultiCol;
query_mem_desc.min_val = 0;
query_mem_desc.max_val = 19;
query_mem_desc.has_nulls = false;
query_mem_desc.group_col_widths.emplace_back(8);
query_mem_desc.group_col_widths.emplace_back(8);
for (const auto& target_info : target_infos) {
const auto slot_bytes = std::max(num_bytes, static_cast<int8_t>(target_info.sql_type.get_size()));
if (target_info.agg_kind == kAVG) {
CHECK(target_info.is_agg);
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.entry_count = query_mem_desc.max_val - query_mem_desc.min_val + 1;
return query_mem_desc;
}
QueryMemoryDescriptor baseline_hash_two_col_desc(const std::vector<TargetInfo>& target_infos, const int8_t num_bytes) {
QueryMemoryDescriptor query_mem_desc{};
query_mem_desc.hash_type = GroupByColRangeType::MultiCol;
query_mem_desc.min_val = 0;
query_mem_desc.max_val = 3;
query_mem_desc.has_nulls = false;
query_mem_desc.group_col_widths.emplace_back(8);
query_mem_desc.group_col_widths.emplace_back(8);
for (const auto& target_info : target_infos) {
const auto slot_bytes = std::max(num_bytes, static_cast<int8_t>(target_info.sql_type.get_size()));
if (target_info.agg_kind == kAVG) {
CHECK(target_info.is_agg);
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.agg_col_widths.emplace_back(ColWidths{slot_bytes, slot_bytes});
}
query_mem_desc.entry_count = query_mem_desc.max_val - query_mem_desc.min_val + 1;
return query_mem_desc;
}
template <class T>
const T* vptr(const TargetValue& r) {
auto scalar_r = boost::get<ScalarTargetValue>(&r);
CHECK(scalar_r);
return boost::get<T>(scalar_r);
}
typedef std::vector<TargetValue> OneRow;
/* This class allows to emulate and evaluate ResultSet and it's reduce function.
* It creates two ResultSet equivalents, populates them with randomly generated data,
* merges them into one, and provides access to the data contained in the merged set.
* Comparing these data with the ones received from the ResultSet code reduce procedure
* run on the same pair of the ResultSet equivalents, allows to evaluate the functionality
* of the ResultSet code.
*/
class ResultSetEmulator {
public:
ResultSetEmulator(int8_t* buff1,
int8_t* buff2,
const std::vector<TargetInfo>& target_infos,
const QueryMemoryDescriptor& query_mem_desc,
NumberGenerator& gen1,
NumberGenerator& gen2,
const size_t perc1,
const size_t perc2,
const size_t flow,
const bool silent)
:
rs1_buff(buff1),
rs2_buff(buff2),
rs_target_infos(target_infos),
rs_query_mem_desc(query_mem_desc),
rs1_gen(gen1),
rs2_gen(gen2),
rs1_perc(perc1),
rs2_perc(perc2),
rs_flow(flow),
rs_entry_count(query_mem_desc.entry_count),
rs_silent(silent) {
rs_entry_count = query_mem_desc.entry_count; // it's set to 10 in "small" query_mem_descriptor
rs1_groups.resize(rs_entry_count);
std::fill(rs1_groups.begin(), rs1_groups.end(), false);
rs2_groups.resize(rs_entry_count);
std::fill(rs2_groups.begin(), rs2_groups.end(), false);
rs1_values.resize(rs_entry_count);
std::fill(rs1_values.begin(), rs1_values.end(), 0);
rs2_values.resize(rs_entry_count);
std::fill(rs2_values.begin(), rs2_values.end(), 0);
rseReducedGroups.resize(rs_entry_count);
std::fill(rseReducedGroups.begin(), rseReducedGroups.end(), false);
emulateResultSets();
}
~ResultSetEmulator(){};
std::queue<std::vector<int64_t>> getReferenceTable() const { return rseReducedTable; }
std::vector<bool> getReferenceGroupMap() const { return rseReducedGroups; }
bool getReferenceGroupMapElement(size_t idx) {
CHECK_LE(idx, rseReducedGroups.size() - 1);
return rseReducedGroups[idx];
}
std::vector<int64_t> getReferenceRow(bool keep_row = false) {
std::vector<int64_t> rse_reduced_row(rseReducedTable.front());
rseReducedTable.pop();
if (keep_row) {
rseReducedTable.push(rse_reduced_row);
}
return rse_reduced_row;
}
int64_t rse_get_null_val() {
int64_t null_val = 0;
for (const auto& target_info : rs_target_infos) {
null_val = inline_int_null_val(target_info.sql_type);
break; // currently all of TargetInfo's columns used in tests have same type, so the
// they all share same null_val, and that's why the first column is used here.
}
return null_val;
}
void print_rse_generated_result_sets() const;
void print_merged_result_sets(const std::vector<OneRow>& result);
private:
void emulateResultSets();
void createResultSet(size_t rs_perc, std::vector<bool>& rs_groups);
void mergeResultSets();
int8_t *rs1_buff, *rs2_buff;
const std::vector<TargetInfo> rs_target_infos;
const QueryMemoryDescriptor rs_query_mem_desc;
NumberGenerator &rs1_gen, &rs2_gen;
size_t rs1_perc, rs2_perc, rs_flow;
size_t rs_entry_count;
bool rs_silent;
std::vector<bool> rs1_groups; // true if group is in ResultSet #1
std::vector<bool> rs2_groups; // true if group is in ResultSet #2
std::vector<bool> rseReducedGroups; // true if group is in either ResultSet #1 or ResultSet #2
std::vector<int64_t> rs1_values; // generated values for ResultSet #1
std::vector<int64_t> rs2_values; // generated values for ResultSet #2
std::queue<std::vector<int64_t>> rseReducedTable; // combined/reduced values of ResultSet #1 and ResultSet #2
void rse_fill_storage_buffer_perfect_hash_colwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values);
void rse_fill_storage_buffer_perfect_hash_rowwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values);
void rse_fill_storage_buffer_baseline_colwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values);
void rse_fill_storage_buffer_baseline_rowwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values);
void rse_fill_storage_buffer(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values);
void print_emulator_diag();
int64_t rseAggregateKMIN(size_t i);
int64_t rseAggregateKMAX(size_t i);
int64_t rseAggregateKAVG(size_t i);
int64_t rseAggregateKSUM(size_t i);
int64_t rseAggregateKCOUNT(size_t i);
};
/* top level module to create and fill up ResultSets as well as to generate golden values */
void ResultSetEmulator::emulateResultSets() {
/* generate topology of ResultSet #1 */
if (!rs_silent) {
printf("\nResultSetEmulator (ResultSet #1): ");
}
createResultSet(rs1_perc, rs1_groups);
if (!rs_silent) {
printf("\n");
for (size_t i = 0; i < rs1_groups.size(); i++) {
if (rs1_groups[i])
printf("1");
else
printf("0");
}
}
/* generate topology of ResultSet #2 */
if (!rs_silent) {
printf("\nResultSetEmulator (ResultSet #2): ");
}
createResultSet(rs2_perc, rs2_groups);
if (!rs_silent) {
printf("\n");
for (size_t i = 0; i < rs2_groups.size(); i++) {
if (rs2_groups[i])
printf("1");
else
printf("0");
}
printf("\n");
}
/* populate both ResultSet's buffers with real data */
// print_emulator_diag();
rse_fill_storage_buffer(rs1_buff, rs1_gen, rs1_groups, rs1_values);
// print_emulator_diag();
rse_fill_storage_buffer(rs2_buff, rs2_gen, rs2_groups, rs2_values);
// print_emulator_diag();
/* merge/reduce data contained in both ResultSets and generate golden values */
mergeResultSets();
}
/* generate ResultSet topology (create rs_groups and rs_groups_idx vectors) */
void ResultSetEmulator::createResultSet(size_t rs_perc, std::vector<bool>& rs_groups) {
std::vector<int> rs_groups_idx;
rs_groups_idx.resize(rs_entry_count);
std::iota(rs_groups_idx.begin(), rs_groups_idx.end(), 0);
std::random_device rs_rd;
std::mt19937 rs_rand_gen(rs_rd());
std::shuffle(rs_groups_idx.begin(), rs_groups_idx.end(), rs_rand_gen);
for (size_t i = 0; i < (rs_entry_count * rs_perc / 100); i++) {
if (!rs_silent) {
printf(" %i", rs_groups_idx[i]);
}
rs_groups[rs_groups_idx[i]] = true;
}
}
/* merge/reduce data contained in both ResultSets and generate golden values */
void ResultSetEmulator::mergeResultSets() {
std::vector<int64_t> rse_reduced_row;
rse_reduced_row.resize(rs_target_infos.size());
for (size_t j = 0; j < rs_entry_count; j++) { // iterates through rows
if (rs1_groups[j] || rs2_groups[j]) {
rseReducedGroups[j] = true;
for (size_t i = 0; i < rs_target_infos.size(); i++) { // iterates through columns
switch (rs_target_infos[i].agg_kind) {
case kMIN: {
rse_reduced_row[i] = rseAggregateKMIN(j);
break;
}
case kMAX: {
rse_reduced_row[i] = rseAggregateKMAX(j);
break;
}
case kAVG: {
rse_reduced_row[i] = rseAggregateKAVG(j);
break;
}
case kSUM: {
rse_reduced_row[i] = rseAggregateKSUM(j);
break;
}
case kCOUNT: {
rse_reduced_row[i] = rseAggregateKCOUNT(j);
break;
}
default:
CHECK(false);
}
}
rseReducedTable.push(rse_reduced_row);
}
}
}
void ResultSetEmulator::rse_fill_storage_buffer_perfect_hash_colwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values) {
const auto key_component_count = get_key_count_for_descriptor(rs_query_mem_desc);
CHECK(rs_query_mem_desc.output_columnar);
// initialize the key buffer(s)
auto col_ptr = buff;
for (size_t key_idx = 0; key_idx < key_component_count; ++key_idx) {
auto key_entry_ptr = col_ptr;
const auto key_bytes = rs_query_mem_desc.group_col_widths[key_idx];
CHECK_EQ(8, key_bytes);
for (size_t i = 0; i < rs_entry_count; i++) {
const auto v = generator.getNextValue();
if (rs_groups[i]) {
// const auto v = generator.getNextValue();
write_key(v, key_entry_ptr, key_bytes);
} else {
write_key(EMPTY_KEY_64, key_entry_ptr, key_bytes);
}
key_entry_ptr += key_bytes;
}
col_ptr = advance_to_next_columnar_key_buff(col_ptr, rs_query_mem_desc, key_idx);
generator.reset();
}
// initialize the value buffer(s)
size_t slot_idx = 0;
for (const auto& target_info : rs_target_infos) {
auto col_entry_ptr = col_ptr;
const auto col_bytes = rs_query_mem_desc.agg_col_widths[slot_idx].compact;
for (size_t i = 0; i < rs_entry_count; i++) {
int8_t* ptr2{nullptr};
if (target_info.agg_kind == kAVG) {
ptr2 = col_entry_ptr + rs_query_mem_desc.entry_count * col_bytes;
}
const auto v = generator.getNextValue();
if (rs_groups[i]) {
// const auto v = generator.getNextValue();
rs_values[i] = v;
if (rs_flow == 2) { // null_val test-cases
if (i >= rs_entry_count - 4) { // only the last four rows of RS #1 and RS #2 exersized for null_val test
rs_values[i] = -1;
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
rs_query_mem_desc.agg_col_widths[slot_idx + 1].compact,
v,
target_info,
false,
true);
} else {
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
rs_query_mem_desc.agg_col_widths[slot_idx + 1].compact,
v,
target_info,
false,
false);
}
} else {
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
rs_query_mem_desc.agg_col_widths[slot_idx + 1].compact,
v,
target_info,
false,
false);
}
} else {
if (rs_flow == 2) { // null_val test-cases
if (i >= rs_entry_count - 4) { // only the last four rows of RS #1 and RS #2 exersized for null_val test
rs_values[i] = -1;
}
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
rs_query_mem_desc.agg_col_widths[slot_idx + 1].compact,
rs_query_mem_desc.keyless_hash ? 0 : 0xdeadbeef,
target_info,
true,
true);
} else {
fill_one_entry_one_col(col_entry_ptr,
col_bytes,
ptr2,
rs_query_mem_desc.agg_col_widths[slot_idx + 1].compact,
rs_query_mem_desc.keyless_hash ? 0 : 0xdeadbeef,
target_info,
true,
false);
}
}
col_entry_ptr += col_bytes;
}
col_ptr = advance_to_next_columnar_target_buff(col_ptr, rs_query_mem_desc, slot_idx);
if (target_info.is_agg && target_info.agg_kind == kAVG) {
col_ptr = advance_to_next_columnar_target_buff(col_ptr, rs_query_mem_desc, slot_idx + 1);
}
slot_idx = advance_slot(slot_idx, target_info, false);
generator.reset();
}
}
void ResultSetEmulator::rse_fill_storage_buffer_perfect_hash_rowwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values) {
const auto key_component_count = get_key_count_for_descriptor(rs_query_mem_desc);
CHECK(!rs_query_mem_desc.output_columnar);
auto key_buff = buff;
CHECK_EQ(rs_groups.size(), rs_query_mem_desc.entry_count);
for (size_t i = 0; i < rs_groups.size(); i++) {
const auto v = generator.getNextValue();
if (rs_groups[i]) {
// const auto v = generator.getNextValue();
rs_values[i] = v;
auto key_buff_i64 = reinterpret_cast<int64_t*>(key_buff);
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
*key_buff_i64++ = v;
}
auto entries_buff = reinterpret_cast<int8_t*>(key_buff_i64);
if (rs_flow == 2) { // null_vall test-cases
if (i >= rs_entry_count - 4) { // only the last four rows of RS #1 and RS #2 exersized for null_val test
rs_values[i] = -1;
key_buff = fill_one_entry_no_collisions(entries_buff, rs_query_mem_desc, v, rs_target_infos, false, true);
} else {
key_buff = fill_one_entry_no_collisions(entries_buff, rs_query_mem_desc, v, rs_target_infos, false, false);
}
} else {
key_buff = fill_one_entry_no_collisions(entries_buff, rs_query_mem_desc, v, rs_target_infos, false);
}
} else {
auto key_buff_i64 = reinterpret_cast<int64_t*>(key_buff);
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
*key_buff_i64++ = EMPTY_KEY_64;
}
auto entries_buff = reinterpret_cast<int8_t*>(key_buff_i64);
if (rs_flow == 2) { // null_val test-cases
if (i >= rs_entry_count - 4) { // only the last four rows of RS #1 and RS #2 exersized for null_val test
rs_values[i] = -1;
}
key_buff =
fill_one_entry_no_collisions(entries_buff, rs_query_mem_desc, 0xdeadbeef, rs_target_infos, true, true);
} else {
key_buff = fill_one_entry_no_collisions(entries_buff, rs_query_mem_desc, 0xdeadbeef, rs_target_infos, true);
}
}
}
}
void ResultSetEmulator::rse_fill_storage_buffer_baseline_colwise(int8_t* buff,
NumberGenerator& generator,
const std::vector<bool>& rs_groups,
std::vector<int64_t>& rs_values) {
CHECK(rs_query_mem_desc.output_columnar);
const auto key_component_count = get_key_count_for_descriptor(rs_query_mem_desc);
const auto i64_buff = reinterpret_cast<int64_t*>(buff);
for (size_t i = 0; i < rs_entry_count; i++) {
for (size_t key_comp_idx = 0; key_comp_idx < key_component_count; ++key_comp_idx) {
i64_buff[key_offset_colwise(i, key_comp_idx, rs_entry_count)] = EMPTY_KEY_64;
}
size_t target_slot = 0;
int64_t init_val = 0;
for (const auto& target_info : rs_target_infos) {
if (target_info.agg_kind == kCOUNT) {
init_val = 0;
} else if (!target_info.sql_type.get_notnull() && target_info.skip_null_val &&
(rs_flow == 2)) { // null_val support
init_val = inline_int_null_val(target_info.sql_type);
} else {
init_val = 0xdeadbeef;
}
if (target_info.agg_kind != kAVG) {
i64_buff[slot_offset_colwise(i, target_slot, key_component_count, rs_entry_count)] = init_val;
} else {
i64_buff[slot_offset_colwise(i, target_slot, key_component_count, rs_entry_count)] = 0;
}
target_slot++;
}
}
for (size_t i = 0; i < rs_entry_count; i++) {
const auto v = generator.getNextValue();
if (rs_groups[i]) {
bool null_val = false;
if ((rs_flow == 2) && (i >= rs_entry_count - 4)) { // null_val test-cases: last four rows
rs_values[i] = -1;
null_val = true;
} else {
rs_values[i] = v;