CHANGELOG.md 15.1 KB
Newer Older
Gilbert Lee's avatar
Gilbert Lee committed
1
2
# Changelog for TransferBench

3
Documentation for TransferBench is available at
Lisa's avatar
Lisa committed
4
[https://rocm.docs.amd.com/projects/TransferBench](https://rocm.docs.amd.com/projects/TransferBench).
Lisa Delaney's avatar
Lisa Delaney committed
5

6
7
8
9
10
11
12
13
14
## v1.50

### Added
- Adding new parallel copy preset benchmark (pcopy)
  - Usage: ./TransferBench pcopy <numBytes=64M> <#CUs=8> <srcGpu=0> <minGpus=1> <maxGpus=#GPU-1>
### Fixed
- Removed non-copies DMA Transfers (this had previously been using hipMemset)
- Fixed CPU executor when operating on null destination

15
16
17
18
19
## v1.49

### Fixes
* Enumerating previously missed DMA engines used only for CPU traffic in topology display

20
21
22
23
24
25
26
27
28
29
30
31
## v1.48

### Fixes
* Various fixes for TransferBenchCuda

### Additions
* Support for targeting specific DMA engines via executor subindex (e.g. D0.1)
* Printing warnings when exeuctors are overcommited

### Modifications
* USE_REMOTE_READ supported for rwrite preset benchmark

32
33
34
35
36
## v1.47

### Fixes
* Fixing CUDA support

37
38
39
40
41
42
43
44
45
## v1.46

### Fixes
* Fixing GFX_UNROLL set to 13 (past 8) on gfx906 cards

### Modifications
* GFX_SINGLE_TEAM=1 by default
* Adding field showing summation of individual Transfer bandwidths for Executors

gilbertlee-amd's avatar
gilbertlee-amd committed
46
47
48
49
50
51
52
53
54
55
## v1.45

### Additions
* Adding A2A_MODE to a2a preset (0 = copy, 1 = read-only, 2 = write-only)
* Adding GFX_UNROLL to modify GFX kernel's unroll factor
* Adding GFX_WAVE_ORDER to modify order in which wavefronts process data

### Modifications
* Rewrote the GFX reduction kernel to support new wave ordering

56
57
58
59
60
61
## v1.44

### Additions
* Adding rwrite preset to benchmark remote parallel writes
 * Usage: ./TransferBench rwrite <numBytes=64M> <#CUs=8> <srcGpu=0> <minGpus=1> <maxGpus=3>

62
63
64
65
66
## v1.43

### Changes
* Modifying a2a to show executor timing, as well as executor min/max bandwidth

67
68
69
70
71
72
## v1.42

### Fixes
* Fixing schmoo maxNumCus optional arg parsing
* Schmoo output modified to be easier to copy

73
74
75
76
77
78
79
80
81
## v1.41

### Additions
* Adding schmoo preset config benchmarks local/remote reads/writes/copies
  * Usage: ./TransferBench schmoo <numBytes=64M> <localIdx=0> <remoteIdx=1> <maxNumCUs=32>

### Fixes
* Fixing some misreported timings when running with non-fixed number of iterations

82
83
84
85
86
## v1.40

### Fixes
* Fixing XCC defaulting to 0 instead of random for preset configs, ignoring XCC_PREF_TABLE

87
88
89
90
91
92
93
## v1.39

### Additions
* (Experimental) Adding support for Executor sub-index
### Fixes
- Remove deprecated gcnArch code.  ROCm version must include support for hipDeviceMallocUncached

94
95
96
97
98
## v1.38

### Fixes
* Adding missing threadfence which could cause non-fine-grained Transfers to report higher speeds

99
100
101
102
103
104
105
106
## v1.37

### Changes
* USE_SINGLE_STREAM is enabled by default now.  (Disable via USE_SINGLE_STREAM=0)

### Fixes
* Fix unrecognized token error when XCC_PREF_TABLE is unspecified

107
108
109
110
111
112
113
## v1.36

### Additions

* (Experimental) Adding XCC filtering - combined with XCC_PREF_TABLE, this tries to select
  specific XCCs to use for specific (SRC->DST) Transfers

114
115
116
117
118
## v1.35

### Additions

* USE_FINE_GRAIN also applies to a2a preset
119

120
## v1.34
121
122
123
124

### Additions

* Set `GPU_KERNEL=3` as default for gfx942
125

126
## v1.33
Lisa Delaney's avatar
Lisa Delaney committed
127
128
129
130
131

### Additions

* Added the `ALWAYS_VALIDATE` environment variable to allow for validation after every iteration, instead
  of only once at the end of all iterations
132

133
## v1.32
Lisa Delaney's avatar
Lisa Delaney committed
134
135
136
137

### Changes

* Increased the line limit from 2048 to 32768
138

gilbertlee-amd's avatar
gilbertlee-amd committed
139
## v1.31
Lisa Delaney's avatar
Lisa Delaney committed
140
141
142
143
144

### Changes

* `SHOW_ITERATIONS` now shows XCC:CU instead of just CU ID
* `SHOW_ITERATIONS` is printed when `USE_SINGLE_STREAM`=1
gilbertlee-amd's avatar
gilbertlee-amd committed
145

146
## v1.30
Lisa Delaney's avatar
Lisa Delaney committed
147
148
149
150
151
152
153
154
155

### Additions

* `BLOCK_SIZE` has been added to control the threadblock size (must be a multiple of 64, up to 512)
* `BLOCK_ORDER` has been added to control how work is ordered for GFX-executors running
  `USE_SINGLE_STREAM`=1
  * 0 - Threadblocks for transfers are ordered sequentially (default)
  * 1 - Threadblocks for transfers are interleaved
  * 2 - Threadblocks for transfers are ordered randomly
156

157
## v1.29
Lisa Delaney's avatar
Lisa Delaney committed
158
159
160
161
162
163
164
165
166
167
168
169
170
171

### Additions

* A2A preset config now responds to `USE_REMOTE_READ`

### Fixes

* Race-condition during wall-clock initialization caused "inf" during single-stream runs
* CU numbering output after CU masking

### Changes

* The default number of warmups has been reverted to 3
* The default unroll factor for gfx940/941 has been set to 6
172

173
## v1.28
Lisa Delaney's avatar
Lisa Delaney committed
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190

### Additions

* Added `A2A_DIRECT`, which only runs all-to-all on directly connected GPUs (now on by default)
* Added average statistics for P2P and A2A benchmarks
* Added `USE_FINE_GRAIN` for P2P benchmark
  * With older devices, P2P performance with default coarse-grain device memory stops timing as soon
    as a request is sent to data fabric, and not actually when it arrives remotely. This can artificially
    inflate bandwidth numbers, especially when sending small amounts of data.

### Changes

* Modified P2P output to help distinguish between CPU and GPU devices

### Fixes

* Fixed Makefile target to prevent unnecessary re-compilation
191

192
## v1.27
Lisa Delaney's avatar
Lisa Delaney committed
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211

### Additions

* Added cmdline preset to allow specification of  simple tests on command line (e.g.,
  `./TransferBench cmdline 64M "1 4 G0->G0->G1"`)
* Adding the `HIDE_ENV` environment variable, which stops environment variable values from printing
* Adding the `CU_MASK` environment variable, which allows you to select the CUs to run on
* `CU_MASK` is specified in CU indices (0-#CUs-1), where ' - ' can be used to denote ranges of values
  (e.g., `CU_MASK`=3-8,16 requests that transfer be run only on CUs 3,4,5,6,7,8,16)
  * Note that this is somewhat experimental and may not work on all hardware
* `SHOW_ITERATIONS` now shows CU usage for that iteration (experimental)

### Changes

* Added extra comments on commonly missing includes with details on how to install them

### Fixes

* CUDA compilation works again (the `wall_clock64` CUDA alias was not defined)
212

213
214
## v1.26

Lisa Delaney's avatar
Lisa Delaney committed
215
216
217
218
219
220
221
222
223
224
225
226
### Additions

* Setting SHOW_ITERATIONS=1 provides additional information about per-iteration timing for file and
  P2P configs
  * For file configs, iterations are sorted from min to max bandwidth and displayed with standard
    deviation
  * For P2P, min/max/standard deviation is shown for each direction

### Changes

* P2P benchmark formatting now reports bidirectional bandwidth in each direction (as well as sum) for
  clarity
227

228
## v1.25
Lisa Delaney's avatar
Lisa Delaney committed
229
230
231
232
233

### Fixes

* Fixed a bug in the P2P bidirectional benchmark that used the incorrect number of `subExecutors` for
  CPU<->GPU tests
234

gilbertlee-amd's avatar
gilbertlee-amd committed
235
## v1.24
Lisa Delaney's avatar
Lisa Delaney committed
236
237
238
239
240

### Additions

* New All-To-All GPU benchmark accessed by preset "A2A"
* Added gfx941 wall clock frequency
gilbertlee-amd's avatar
gilbertlee-amd committed
241

242
## v1.23
Lisa Delaney's avatar
Lisa Delaney committed
243
244
245
246
247

### Additions

* New GPU subexec scaling benchmark accessed by preset "scaling"
  * Tests GPU-GFX copy performance based on # of CUs used
248

249
## v1.22
Lisa Delaney's avatar
Lisa Delaney committed
250
251
252
253

### Changes

* Switched the kernel timing function to `wall_clock64`
254

gilbertlee-amd's avatar
gilbertlee-amd committed
255
## v1.21
Lisa Delaney's avatar
Lisa Delaney committed
256
257
258
259

### Fixes

* Fixed a bug with `SAMPLING_FACTOR`
gilbertlee-amd's avatar
gilbertlee-amd committed
260

261
## v1.20
Lisa Delaney's avatar
Lisa Delaney committed
262
263
264
265
266

### Fixes

* `VALIDATE_DIRECT` can now be used with `USE_PREP_KERNEL`
* Switched to local GPU for validating GPU memory
267

gilbertlee-amd's avatar
gilbertlee-amd committed
268
## v1.19
Lisa Delaney's avatar
Lisa Delaney committed
269
270
271
272
273

### Additions

* `VALIDATE_DIRECT` now also applies to source memory array checking
* Added null memory pointer check prior to deallocation
gilbertlee-amd's avatar
gilbertlee-amd committed
274

275
## v1.18
Lisa Delaney's avatar
Lisa Delaney committed
276
277
278
279
280
281
282
283
284
285
286
287
288

### Additions

* Adding the ability to validate GPU destination memory directly without going through the CPU
  staging buffer (`VALIDATE_DIRECT`)
  * Note that this only works on AMD devices with large-bar access enabled, and may slow things down
    considerably

### Changes

* Refactored how environment variables are displayed
* Mismatch stops after the first detected error within an array instead of listing all mismatched
  elements
289

290
## v1.17
Lisa Delaney's avatar
Lisa Delaney committed
291
292
293
294
295
296
297
298
299
300
301
302
303
304

### Additions

* Allowed switch to GFX kernel for source array initialization (`USE_PREP_KERNEL`)
  * Note that `USE_PREP_KERNEL` can't be used with `FILL_PATTERN`
* Added the ability to compile with nvcc only (`TransferBenchCuda`)

### Changes

* The default pattern was set to [Element i = ((i * 517) modulo 383 + 31) * (srcBufferIdx + 1)]

### Fixes

* Added the `example.cfg` file
305

306
## v1.16
Lisa Delaney's avatar
Lisa Delaney committed
307
308
309
310
311
312
313

### Additions

* Additional src array validation during preparation
* Added a new environment variable (`CONTINUE_ON_ERROR`) to resume tests after a mis-match
  detection
* Initialized GPU memory to 0 during allocation
314

315
## v1.15
Lisa Delaney's avatar
Lisa Delaney committed
316
317
318
319
320
321
322
323
324

### Fixes

* Fixed a bug that prevented single transfers greater than 8 GB

### Changes

* Removed "check for latest ROCm" warning when allocating too much memory
* Off-source memory value is now printed when a mis-match is detected
325

PedramAlizadeh's avatar
PedramAlizadeh committed
326
## v1.14
Lisa Delaney's avatar
Lisa Delaney committed
327
328
329
330
331
332

### Additions

* Added documentation
* Added pthread linking in src/Makefile and CMakeLists.txt
* Added printing off the hex value of the floats for output and reference
333

PedramAlizadeh's avatar
PedramAlizadeh committed
334
335
## v1.13

Lisa Delaney's avatar
Lisa Delaney committed
336
337
338
339
340
341
342
### Additions

* Added support for cmake

### Changes

* Converted to the Pitchfork layout standard
PedramAlizadeh's avatar
PedramAlizadeh committed
343

344
## v1.12
Lisa Delaney's avatar
Lisa Delaney committed
345
346
347
348
349

### Additions

* Added support for TransferBench on NVIDIA platforms (via `HIP_PLATFORM`=nvidia)
  * Note that CPU executors on NVIDIA platform cannot access GPU memory (no large-bar access)
350

gilbertlee-amd's avatar
gilbertlee-amd committed
351
## v1.11
Lisa Delaney's avatar
Lisa Delaney committed
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

### Additions

* Added multi-input/multi-output (MIMO) support: transfers now can reduce (element-wise summation)
  multiple input memory arrays and write sums to multiple outputs
* Added GPU-DMA executor 'D', which uses `hipMemcpy` for SDMA copies
  * Previously, this was done using `USE_HIP_CALL`, but now GPU-GFX kernel can run in parallel with
    GPU-DMA, instead of applying to all GPU executors globally
  * GPU-DMA executor can only be used for single-input/single-output transfers
  * GPU-DMA executor can only be associated with one SubExecutor
* Added new "Null" memory type 'N', which represents empty memory. This allows for read-only or
  write-only transfers
* Added new `GPU_KERNEL` environment variable that allows switching between various GPU-GFX
  reduction kernels

### Optimizations

* Improved GPU-GFX kernel performance based on hardware architecture when running with
  fewer CUs

### Changes

* Updated the `example.cfg` file to cover new features
* Updated output to support MIMO
* Changed CU and CPU thread naming to SubExecutors for consistency
* Sweep Preset: default sweep preset executors now includes DMA
* P2P benchmarks:
  * Removed `p2p_rr`, `g2g` and `g2g_rr` (now only works via P2P)
    * Setting `NUM_CPU_DEVICES`=0 can only be used to benchmark GPU devices (like `g2g`)
    * The new `USE_REMOTE_READ` environment variable replaces `_rr` presets
  * New environment variable `USE_GPU_DMA`=1 replaces `USE_HIP_CALL`=1 for benchmarking with
    GPU-DMA Executor
  * Number of GPU SubExecutors for benchmark can be specified via `NUM_GPU_SE`
    * Defaults to all CUs for GPU-GFX, 1 for GPU-DMA
  * Number of CPU SubExecutors for benchmark can be specified via `NUM_CPU_SE`
* Psuedo-random input pattern has been slightly adjusted to have different patterns for each input
  array within same transfer

### Removals

* `USE_HIP_CALL`: use `GPU-DMA` executor 'D' or set `USE_GPU_DMA`=1 for P2P
  benchmark presets
  * Currently, a warning will be issued if `USE_HIP_CALL` is set to 1 and the program will stop
* `NUM_CPU_PER_TRANSFER`: the number of CPU SubExecutors will be whatever is specified for the
  transfer
* `USE_MEMSET`: this function can now be done via a transfer using the null memory type
gilbertlee-amd's avatar
gilbertlee-amd committed
398

399
## v1.10
Lisa Delaney's avatar
Lisa Delaney committed
400
401
402
403

### Fixes

* Fixed incorrect bandwidth calculation when using single stream mode and per-transfer data sizes
404

405
## v1.09
Lisa Delaney's avatar
Lisa Delaney committed
406
407
408
409
410
411
412
413

### Additions

* Printing off src/dst memory addresses during interactive mode

### Changes

* Switching to `numa_set_preferred` instead of `set_mempolicy`
414

415
## v1.08
Lisa Delaney's avatar
Lisa Delaney committed
416
417
418
419
420
421

### Changes

* Fixed handling of non-configured NUMA nodes
* Topology detection now shows actual NUMA node indices
* Fixed 'for' issue with `NUM_GPU_DEVICES`
422

423
## v1.07
Lisa Delaney's avatar
Lisa Delaney committed
424
425
426
427

### Fixes

* Fixed bug with allocations involving non-default CPU memory types
428

gilbertlee-amd's avatar
gilbertlee-amd committed
429
## v1.06
Lisa Delaney's avatar
Lisa Delaney committed
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445

### Additions

* Unpinned CPU memory type ('U'), which may require `HSA_XNACK`=1 in order to access via
  GPU executors
* Added sweep configuration logging to `lastSweep.cfg`
* Ability to specify the number of CUs to use for sweep-based presets

### Changes

* Modified advanced configuration file format to accept bytes-per-transfer

### Fixes

* Fixed random sweep repeatability
* Fixed bug with CPU NUMA node memory allocation
gilbertlee-amd's avatar
gilbertlee-amd committed
446

447
## v1.05
Lisa Delaney's avatar
Lisa Delaney committed
448
449
450
451
452
453
454
455
456

### Additions

* Topology output now includes NUMA node information
* Support for NUMA nodes with no CPU cores (e.g., CXL memory)

### Removals

* The `SWEEP_SRC_IS_EXE` environment variable was removed
457

458
## v1.04
Lisa Delaney's avatar
Lisa Delaney committed
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473

### Additions

* There are new environment variables for sweep based presets:
  * `SWEEP_XGMI_MIN`: The minumum number of XGMI hops for transfers
  * `SWEEP_XGMI_MAX`: The maximum number of XGMI hops for transfers
  * `SWEEP_SEED`: Uses a random seed
  * `SWEEP_RAND_BYTES`: Uses a random amount of bytes (up to pre-specified N) for each transfer

### Changes

* CSV output for sweep now includes an environment variables section followed by output
* CSV output no longer lists environment variable parameters in columns
* We changed the default number of warmup iterations from 3 to 1
* Split CSV output of link type to `ExeToSrcLinkType` and `ExeToDstLinkType`
474

Gilbert Lee's avatar
Gilbert Lee committed
475
## v1.03
Lisa Delaney's avatar
Lisa Delaney committed
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498

### Additions

* There are new preset modes stress-test benchmarks: `sweep` and `randomsweep`
  * `sweep` iterates over all possible sets of transfers to test
  * `randomsweep` iterates over random sets of transfers
  * New sweep-only environment variables can modify `sweep`
    * `SWEEP_SRC`: String containing only "B","C","F", or "G" that defines possible source memory types
    * `SWEEP_EXE`: String containing only "C" or "G" that defines possible executors
    * `SWEEP_DST`: String containing only "B","C","F", or "G" that defines possible destination memory types
    * `SWEEP_SRC_IS_EXE`: Restrict the executor to be the same as the source, if non-zero
    * `SWEEP_MIN`: Minimum number of parallel transfers to test
    * `SWEEP_MAX`: Maximum number of parallel transfers to test
    * `SWEEP_COUNT`: Maximum number of tests to run
    * `SWEEP_TIME_LIMIT`: Maximum number of seconds to run tests
* New environment variables to restrict number of available devices to test on (primarily for sweep
  runs)
  * `NUM_CPU_DEVICES`: Number of CPU devices
  * `NUM_GPU_DEVICES`: Number of GPU devices

### Fixes

* Fixed timing display for CPU executors when using single-stream mode
Gilbert Lee's avatar
Gilbert Lee committed
499

Gilbert Lee's avatar
Gilbert Lee committed
500
## v1.02
Lisa Delaney's avatar
Lisa Delaney committed
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517

### Additions

* Setting `NUM_ITERATIONS` to a negative number indicates a run of -`NUM_ITERATIONS` seconds per
  test

### Changes

* Copies are now referred to as 'transfers' instead of 'links'
* Reordered how environment variables are displayed (alphabetically now)

### Removals

* Combined timing is now always on for kernel-based GPU copies; the `COMBINED_TIMING`
  environment variable has been removed
* Single sync is no longer supported for facility variable iterations; the `USE_SINGLE_SYNC`
  environmental variable has been removed
Gilbert Lee's avatar
Gilbert Lee committed
518

Gilbert Lee's avatar
Gilbert Lee committed
519
## v1.01
Lisa Delaney's avatar
Lisa Delaney committed
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538

### Additions

* Added the `USE_SINGLE_STREAM` feature
  * All Links that run on the same GPU device are run with a single kernel launch on a single stream
  * This doesn't work with `USE_HIP_CALL`, and it forces `USE_SINGLE_SYNC` to collect timings
  * Added the ability to request coherent or fine-grained host memory ('B')

### Changes

* Separated the TransferBench repository from the RCCL repository
* Peer-to-peer benchmark mode now works with `OUTPUT_TO_CSV`
* Toplogy display now works with `OUTPUT_TO_CSV`
* Moved the documentation about the config file into `example.cfg`

### Removals

* Removed config file generation
* Removed the 'show pointer address' (`SHOW_ADDR`) environment variable