-
Notifications
You must be signed in to change notification settings - Fork 293
/
partitioning.py
1147 lines (1004 loc) · 45.2 KB
/
partitioning.py
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 2024 The T5X Authors.
#
# 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.
"""Utilities for partitioning."""
import abc
import collections
import dataclasses
import functools
import typing
from typing import Any, Callable, Optional, Sequence, Set, Tuple, Union
from absl import logging
import cached_property
from flax import traverse_util
from flax.linen import partitioning as flax_partitioning
import jax
from jax import numpy as jnp
from jax import random
from jax.experimental import multihost_utils
from jax.experimental.mesh_utils import create_hybrid_device_mesh
from jax.experimental.pjit import pjit
from jax.interpreters import pxla
from jax.sharding import Mesh
from jax.sharding import PartitionSpec
import numpy as np
from t5x import train_state as train_state_lib
JaxDevice = jax.Device
TpuMesh = Tuple[int, int, int, int] # (x, y, z, num_cores).
OtherMesh = Tuple[int, int]
HardwareMesh = Union[TpuMesh, OtherMesh]
TrainState = train_state_lib.TrainState
LogicalAxisRules = Sequence[Tuple[str, Optional[str]]]
if typing.TYPE_CHECKING: # See b/163639353
cached_property = property # pylint: disable=invalid-name
else:
cached_property = cached_property.cached_property
class AxisNames(tuple):
"""Tuple of strings specifying name for each axis.
We create a separate class for this so JAX's pytree utilities can distinguish
it from a tuple that should be treated as a pytree, instead treating it as a
leaf.
"""
def __new__(cls, *names):
return tuple.__new__(AxisNames, names)
def __repr__(self):
return 'AxisNames%s' % tuple.__repr__(self)
def with_sharding_constraint(x, axis_resources):
"""Wrapper for lax.with_sharding_constraint, no-op on cpu or outside pjit."""
if jax.devices()[0].platform == 'cpu' or not global_mesh_defined():
return x
else:
return jax.lax.with_sharding_constraint(x, axis_resources)
# pjit Mesh creation functions.
# -----------------------------------------------------------------------------
def bounds_from_last_device(last_device: jax.Device) -> HardwareMesh:
"""Get the bound from the given last device."""
# Must be passed the device at the highest-coordinate corner of the
# relevant mesh, which is a requirement we know is satisfied by the last
# device in jax.devices().
if hasattr(last_device, 'coords') and len(last_device.coords) == 3:
x, y, z = last_device.coords
return x + 1, y + 1, z + 1, last_device.core_on_chip + 1
else:
# On non-TPU platforms, the "mesh" is hosts x devices per host in order
# to take advantage of faster within-host interconnect.
return jax.process_count(), jax.local_device_count()
def get_coords(device: jax.Device) -> HardwareMesh:
"""Returns the coordinates of the given device."""
if hasattr(device, 'coords'):
return (*device.coords, device.core_on_chip)
return (device.process_index, device.id % jax.local_device_count())
def global_mesh_defined():
"""Checks if global xmap/pjit mesh resource environment is defined."""
maps_env = pxla.thread_resources.env
return maps_env.physical_mesh.devices.shape != () # pylint: disable=g-explicit-bool-comparison
def get_mesh(model_parallel_submesh: HardwareMesh,
input_devices: Sequence[JaxDevice] = (),
input_local_devices: Sequence[JaxDevice] = (),
tile_by_host_if_needed: bool = True,
backend: Optional[str] = None) -> Mesh:
"""Construct an xmap/pjit Mesh for the given model-parallel submesh.
The resulting mesh has two resource axes: 'model', with the provided submesh
shape, and 'data', which covers the rest of the mesh.
Args:
model_parallel_submesh: a HardwareMesh spec, namely (x,y,z,core) on TPU for
a single model-parallel replica's "tile" in the physical device mesh. The
first three elements (`x`, `y`, and `z`) should be factors of the pod
slice; e.g., if you are using df_4x8, then `x` should be a factor of 4
(one of 1, 2, 4), `y` should be a factor of 8 (one of 1, 2, 4, 8), and `z`
must be 1, because TPU v3 slices are only 2D. `z` can be >1 for TPU v4
(and maybe later TPUs) that allow 3D slices. `core` is the number of cores
to use from each TPU node. As communication is usually fastest inside the
same node, if you need a tile of more than 1 core, then
you should first increase `core`: e.g., for TPU v3, (1,1,1,2) is better
than (2,1,1,1). To pick a good spec, try a few possible values until you
get high TPU utilization.
input_devices: the devices to use, will use jax.devices() if this is not
set.
input_local_devices: the local devices to use, will use jax.local_devices()
if this is not set.
tile_by_host_if_needed: JAX currently requires that the parts of any sharded
array that are located on one host's local devices form a single
contiguous slice. A best effort will be made to achieve this without
"tiling" the device assignment over hosts (which can reduce XLA collective
performance). If this flag is True, then the device assignment will be
tiled over hosts if necessary to satisfy this constraint and create a
buildable mesh; if false, mesh construction will fail instead.
backend: get devices from the pinned backend, if specified. This is
useful for explicitly specifying the devices other than relying on
jax_platform_name.
Returns:
A xmap / pjit Mesh containing the virtual device mesh with data, model axes.
"""
input_devices = input_devices or jax.devices(backend)
input_local_devices = input_local_devices or jax.local_devices(0, backend)
# Sort input_devices based on coords, as backends might not return devices
# in order.
last_device = sorted(input_devices, key=get_coords)[-1]
last_input_local_devices = sorted(input_local_devices, key=get_coords)[-1]
logging.info('last device coords : %r\nlast local device coords: %r',
get_coords(last_device), get_coords(last_input_local_devices))
global_hardware_mesh = bounds_from_last_device(last_device)
mesh_ndim = len(global_hardware_mesh)
local_hardware_mesh = bounds_from_last_device(last_input_local_devices)
mesh_err = (
f'each dimension of the model parallel submesh {model_parallel_submesh} '
'must be a factor of the corresponding dimension of the global device '
f'mesh {global_hardware_mesh}')
assert not any(
g % m
for g, m in zip(global_hardware_mesh, model_parallel_submesh)), mesh_err
assert not any(
g % l for g, l in zip(global_hardware_mesh, local_hardware_mesh))
devices = np.empty(global_hardware_mesh, dtype=object)
for device in input_devices:
device_coords = get_coords(device)
devices[device_coords] = device
tile_by_host = tile_by_host_if_needed
if len(global_hardware_mesh) == 4:
# enable contiguous local chunks without host tiling by making Z major
global_hardware_mesh = typing.cast(Tuple[int, int, int, int],
global_hardware_mesh)
model_parallel_submesh = typing.cast(Tuple[int, int, int, int],
model_parallel_submesh)
gx, gy, gz, gc = global_hardware_mesh
mx, my, mz, mc = model_parallel_submesh
if (mx == gx > 1 and my == mz == 1) or (mx == 1 and my == gy > 1 and
mz == gz > 1):
logging.info('ensuring YZ plane has a Z-major device order')
# YZ should be ZY
assert mc == gc, (mc, gc)
global_hardware_mesh = gx, gz, gy, gc
model_parallel_submesh = mx, mz, my, mc
devices = devices.swapaxes(1, 2)
tile_by_host = False
if (my == gy > 1 and mx == mz == 1) or (my == 1 and mx == gx > 1 and
mz == gz > 1):
logging.info('ensuring XZ plane has a Z-major device order')
# XZ should be ZX
assert mc == gc, (mc, gc)
global_hardware_mesh = gz, gy, gx, gc
model_parallel_submesh = mz, my, mx, mc
devices = devices.swapaxes(0, 2)
tile_by_host = False
if tile_by_host:
logging.warning(
'Tiling device assignment mesh by hosts, which may lead to '
'reduced XLA collective performance. To avoid this, modify '
'the model parallel submesh or run with more tasks per host.')
tile_err = (
'to tile the mesh by hosts, each dimension of the model parallel '
'submesh must be either a factor or a multiple of the corresponding '
'dimension of the per-host submesh')
def dh_dd_mh_md(g: int, m: int, l: int) -> Tuple[int, int, int, int]:
"""Split a global mesh dimension into four tiling components.
Args:
g: global mesh bounds dimension size
m: model-parallel submesh bounds dimension size
l: local submesh bounds dimension size
Returns:
The resulting tuple divides the dimension into the hosts component of
the data-parallel submesh, the devices component of the data-parallel
submesh, the hosts component of the model-parallel submesh, and the
devices component of the model-parallel submesh.
"""
d = g // m
if m >= l:
assert not m % l, tile_err
return (d, 1, m // l, l)
else:
assert not l % m, tile_err
return (d // (l // m), l // m, 1, m)
# e.g. [(x_data_hosts, x_data_devs, x_model_hosts, x_model_devs), ...]
dh_dd_mh_md_tups = map(dh_dd_mh_md, global_hardware_mesh,
model_parallel_submesh, local_hardware_mesh)
# reshape to e.g. (x_dh, x_dd, x_mh, x_md, y_dh, ...)
devices = devices.reshape(*(s for t in dh_dd_mh_md_tups for s in t)) # pylint: disable=g-complex-comprehension
# TODO(jekbradbury): reorder local subgroups for ring locality
# Transpose to [data_host], [data_device], [model_host], [model_device]
# block ordering e.g. (x_dh, y_dh, ..., x_dd, y_dd, ...)
devices = devices.transpose(*(4 * i for i in range(mesh_ndim)),
*(4 * i + 1 for i in range(mesh_ndim)),
*(4 * i + 2 for i in range(mesh_ndim)),
*(4 * i + 3 for i in range(mesh_ndim)))
else:
# e.g. [(x_data, x_model), (y_data, y_model), ...]
model_data_tups = [
(g // m, m)
for g, m in zip(global_hardware_mesh, model_parallel_submesh)
]
# reshape to e.g. (x_data, x_model, y_data, y_model...)
devices = devices.reshape(*(s for t in model_data_tups for s in t)) # pylint: disable=g-complex-comprehension
# TODO(jekbradbury): reorder small subgroups for ring locality
# transpose to e.g. (x_data, y_data, ..., x_model, ...)
devices = devices.transpose(*(2 * i for i in range(mesh_ndim)),
*(2 * i + 1 for i in range(mesh_ndim)))
# reshape to (data, model)
devices = devices.reshape(-1, np.prod(model_parallel_submesh))
global_mesh = Mesh(devices, ['data', 'model'])
logging.info('global_mesh axis_names: %s', global_mesh.axis_names)
logging.info('global_mesh devices: %s', global_mesh.devices)
logging.info('global_mesh devices shape: %s', global_mesh.devices.shape)
return global_mesh
def get_cpu_mesh() -> Mesh:
"""Trivial mesh for CPU Testing."""
devices = np.empty(
(jax.process_count(), jax.local_device_count()), dtype=object
)
for device in jax.devices():
devices[device.process_index, device.id % jax.local_device_count()] = device
return Mesh(devices, ['data', 'model'])
def get_gpu_mesh(num_partitions: int) -> Mesh:
"""Mesh for GPUs that preferentially places 'model' on NVLink."""
nvlink_size = jax.local_device_count()
dcn_size = jax.process_count()
nvlink_mp = min(num_partitions, nvlink_size)
nvlink_dp, extra1 = divmod(nvlink_size, nvlink_mp)
dcn_mp, extra2 = divmod(num_partitions, nvlink_mp)
assert not (extra1 or extra2), ('number of partitions on GPU must be a factor'
' or multiple of the number of local devices')
dcn_dp = dcn_size // dcn_mp
devices = create_hybrid_device_mesh(
mesh_shape=[nvlink_dp, nvlink_mp],
dcn_mesh_shape=[dcn_dp, dcn_mp],
process_is_granule=True)
global_mesh = Mesh(devices, ['data', 'model'])
logging.info('global_mesh axis_names: %s', global_mesh.axis_names)
logging.info('global_mesh devices: %s', global_mesh.devices)
return global_mesh
def default_mesh(
num_partitions: int,
model_parallel_submesh: Optional[HardwareMesh] = None,
backend: Optional[str] = None,
ici_mesh_shape: Optional[HardwareMesh] = None,
dcn_mesh_shape: Optional[HardwareMesh] = None,
) -> Mesh:
"""Attempt to return a default mesh for simple cases.
Args:
num_partitions: number of partitions to use, will be ignored if
model_parallel_submesh is provided.
model_parallel_submesh: 4-tuple that specifies the x,y,z,c submesh to use as
the model-parallel device tile.
backend: get devices from the pinned backend, if specified. This is useful
for explicitly specifying the devices other than relying on
jax_platform_name.
ici_mesh_shape: Shape of the logical mesh used for SPMD parallelism in each
slice. The meaning of each mesh axis is defined by mesh_axis_names, so
these two params must be the same length. If dcn_mesh_shape is present,
the overall mesh is the product of ici_mesh_shape and dcn_mesh_shape. For
example, an ici_mesh_shape of [2, 3, 4] with mesh_axis_names ['replica',
'data', 'model'] indicates 2-way replica parallelism, 3-way data
parallelism, and 4-way model parallelism over 24 devices. None, the
default, is equivalent to a sequence of ones and means that the model is
placed on a single device.
dcn_mesh_shape: Shape of the logical mesh used for SPMD parallelism over
multiple slices. The overall mesh is the product of ici_mesh_shape and
dcn_mesh_shape, and the meaning of each mesh axis is defined by
mesh_axis_names, so these three params must be the same length.
Returns:
xmap/pjit 2D Mesh with 'data', 'model' mesh axes if single-slice, otherwise
3D Mesh with 'replica', 'data', and 'model' mesh axes.
"""
devices = jax.devices(backend)
last_device = devices[-1]
platform = last_device.platform
device_kind = last_device.device_kind
bounds = bounds_from_last_device(last_device)
if ici_mesh_shape is not None and dcn_mesh_shape is not None:
device_mesh = create_hybrid_device_mesh(
ici_mesh_shape,
dcn_mesh_shape,
devices=devices,
)
multi_slice_global_mesh = Mesh(device_mesh, ['replica', 'data', 'model'])
logging.info(
'multi_slice_global_mesh axis_names: %s',
multi_slice_global_mesh.axis_names,
)
logging.info(
'multi_slice_global_mesh devices: %s', multi_slice_global_mesh.devices
)
logging.info(
'multi_slice_global_mesh devices shape: %s',
multi_slice_global_mesh.devices.shape,
)
return multi_slice_global_mesh
if model_parallel_submesh:
return get_mesh(model_parallel_submesh, backend=backend)
if platform == 'cpu':
return get_cpu_mesh()
elif platform == 'gpu':
return get_gpu_mesh(num_partitions)
mps = None
if device_kind in ('TPU v2', 'TPU v3'):
if num_partitions == 1:
mps = (1, 1, 1, 1)
elif num_partitions == 2:
mps = (1, 1, 1, 2)
elif num_partitions == 4:
mps = (2, 1, 1, 2)
elif num_partitions == 8:
mps = (2, 2, 1, 2)
elif num_partitions == 16:
mps = (4, 2, 1, 2)
# assume the use of megacore on TPU v4
elif (device_kind == 'TPU v4' or
device_kind == 'TPU v4 lite') and bounds[3] == 1:
if num_partitions == 1:
mps = (1, 1, 1, 1)
elif num_partitions == 2:
mps = (1, 2, 1, 1)
elif num_partitions == 4:
if bounds[0] >= 4:
mps = (4, 1, 1, 1)
else:
mps = (2, 2, 1, 1)
elif num_partitions == 8:
if bounds[2] >= 8:
mps = (1, 1, 8, 1)
else:
mps = (4, 2, 1, 1)
elif num_partitions == 16:
if bounds[2] >= 16:
mps = (1, 1, 16, 1)
elif bounds[0] >= 8:
mps = (8, 2, 1, 1)
elif bounds[0] >= 4:
mps = (4, 4, 1, 1)
else:
mps = (2, 2, 4, 1)
if mps is None:
raise ValueError(
'No default mesh for this configuration: specify '
'config.model_parallel_submesh explicitly. \n'
f'Platform: {platform}\n'
f'Device kind: {device_kind}\n'
f'Num partitions: {num_partitions}\n'
f'Bounds: {bounds}'
)
return get_mesh(mps, backend=backend)
# Data chunking helper.
# -----------------------------------------------------------------------------
@dataclasses.dataclass
class LocalChunkInfo:
# The logical slice of an array located on this host's local devices.
slice: Tuple[slice, ...]
# A unique index for this host/local chunk among chunks with the same slice.
replica_id: int
class LocalChunker:
"""Utility class to aid chunking of sharded arrays in multihost settings."""
def __init__(self, global_mesh: Mesh):
self.global_mesh = global_mesh
local_mesh = global_mesh.local_mesh
first_local_device = local_mesh.devices.reshape(-1)[0]
host_location = collections.OrderedDict(
zip(
global_mesh.shape.keys(),
list(zip(*np.nonzero(
global_mesh.devices == first_local_device)))[0]))
self.num_chunks = collections.OrderedDict()
self.chunk_ids = collections.OrderedDict()
self.mesh_axes = list(global_mesh.shape.keys())
for mesh_axis in self.mesh_axes:
num_devices_per_chunk = local_mesh.shape[mesh_axis]
self.num_chunks[mesh_axis] = (
global_mesh.shape[mesh_axis] // num_devices_per_chunk)
self.chunk_ids[mesh_axis] = (
host_location[mesh_axis] // num_devices_per_chunk)
def get_local_chunk_info(
self, global_shape: Tuple[int, ...],
mesh_axes: Sequence[Optional[str]]) -> LocalChunkInfo:
"""Get the local chunk info for a given array shape and sharded axes.
Args:
global_shape: the global, unsharded shape of the array to chunk.
mesh_axes: a sequence of names (or None) of equal rank to `global_shape`
that specifies which mesh dimensions the array is sharded along.
Returns:
LocalChunkInfo containing the logical slices of the array found on this
host's local devices, as well as the replica index for this chunk among
chunks with the same slice. The latter is used to determine which
host should write this chunk during checkpointing.
"""
local_slice = [slice(None) for dim in global_shape]
sharded_mesh_axes = set()
for i, (mesh_axis, size) in enumerate(zip(mesh_axes, global_shape)):
if not mesh_axis:
continue
sharded_mesh_axes.add(mesh_axis)
if not isinstance(mesh_axis, str):
raise NotImplementedError('TODO(jekbradbury)')
chunk_id = self.chunk_ids[mesh_axis]
chunk_size = size // self.num_chunks[mesh_axis]
local_slice[i] = slice(chunk_id * chunk_size, (chunk_id + 1) * chunk_size)
replica_id = self.get_replica_id(sharded_mesh_axes)
return LocalChunkInfo(tuple(local_slice), replica_id)
def get_shard_id(self, sharded_mesh_axes: str | Set[Optional[str]]) -> int:
"""Given mesh axes used for sharding, computes current host's shard id.
To give an example, let's say there are two axes globally: replica, data,
and model, the mesh axes for sharding is ('replica', 'data'), which means we
are going to partition an array along 'replica' and 'data' axes.
The shard_id is to show the index of the current local host along the
sharding axes (in this example, it's 'replica' and 'data' axes).
More concretely, let's say we have 4 local hosts, and we use 'replica' and
'data' axes for data parallel (2 hosts along the replica axis, and 2 host
along the data axis). The host located in ('replica': 0, 'data': 0), we
should assign data shard-0 to it. For host ('replica': 0, 'data': 1), we
assign shard-1. For host ('replica': 1, 'data': 0), we assign shard-2.
For host ('replica': 1, 'data': 1), we assign shard-3.
Note: the host location along 'replica' and 'data' axes, e.g.,
('replica': 0, 'data': 0) is named chunk_id and stored in
self._local_chunker.chunk_ids[axis].
Args:
sharded_mesh_axes: the mesh axes for sharding.
Returns:
the index of the current local host along the sharding axes.
"""
if isinstance(sharded_mesh_axes, str):
sharded_mesh_axes = (sharded_mesh_axes,)
shard_id = 0
for mesh_axis in sharded_mesh_axes:
chunk_id = self.chunk_ids[mesh_axis]
shard_id = shard_id * self.num_chunks[mesh_axis] + chunk_id
return shard_id
def get_replica_id(self, sharded_mesh_axes: str | Set[Optional[str]]) -> int:
"""Given mesh axes used for sharding, computes current host's replica id.
To give an example, let's say there are two axes globally: data, and model,
the mesh axes for sharding is ('data', ), which means we are going to
partition an array along 'data' axis and replicate it along 'model' axis.
The replica_id is to show the index of the current local host along the
'model' axis.
Args:
sharded_mesh_axes: the mesh axes for sharding.
Returns:
the index of the current local host along the non-sharding axes (i.e.,
replicating axes).
"""
if isinstance(sharded_mesh_axes, str):
sharded_mesh_axes = (sharded_mesh_axes,)
replicated_mesh_axes = [
mesh_axis for mesh_axis in self.mesh_axes
if mesh_axis not in sharded_mesh_axes
]
replica_id = 0
for mesh_axis in replicated_mesh_axes:
chunk_id = self.chunk_ids[mesh_axis]
replica_id = replica_id * self.num_chunks[mesh_axis] + chunk_id
return replica_id
def standard_logical_axis_rules(
activation_partitioning_dims: int = 1,
parameter_partitioning_dims: int = 1,
additional_rules: Optional[LogicalAxisRules] = None) -> LogicalAxisRules:
"""Default sharding rules for T5X model in terms of logical axis names.
Args:
activation_partitioning_dims: enables 2-D activation sharding when set to 2.
parameter_partitioning_dims: enables 2-D parameter sharding when set to 2.
additional_rules: additional rules (a sequence of tuples) that will be
appended to the standard rules.
Returns:
Sequence of logical axis rules
"""
logging.info(
'`activation_partitioning_dims` = %d, `parameter_partitioning_dims` = %d',
activation_partitioning_dims, parameter_partitioning_dims)
if activation_partitioning_dims == 1 and parameter_partitioning_dims == 1:
rules = [
('batch', 'data'),
('vocab', 'model'),
('embed', None),
('mlp', 'model'),
('heads', 'model'),
('kv', None),
('joined_kv', 'model'), # joined heads+kv dim in 2D attn param layouts
]
elif activation_partitioning_dims == 2 and parameter_partitioning_dims == 1:
rules = [
('batch', 'data'),
('vocab', 'model'),
('mlp', 'model'),
('heads', 'model'),
('kv', None),
('joined_kv', 'model'),
('embed', 'model'),
]
elif activation_partitioning_dims == 1 and parameter_partitioning_dims == 2:
rules = [
('batch', 'data'),
('vocab', 'model'),
('mlp', 'model'),
('heads', 'model'),
('kv', None),
('joined_kv', 'model'),
('embed', 'data'),
]
elif activation_partitioning_dims == 2 and parameter_partitioning_dims == 2:
rules = [
('batch', 'data'),
('vocab', 'model'),
('mlp', 'model'),
('heads', 'model'),
('kv', None),
('joined_kv', 'model'),
('embed', 'model'),
('embed', 'data'),
]
else:
raise ValueError(
f'`activation_partitioning_dims` = {activation_partitioning_dims} '
f'`parameter_partitioning_dims` = {parameter_partitioning_dims} '
'is not supported.')
# Add the common rules for the replicated logical axes names.
replicated_rules = [
('relpos_buckets', None),
('abspos_buckets', None),
('length', None),
('layers', None),
('stack', None),
('mlp_activations', None),
]
rules.extend(replicated_rules)
if additional_rules:
rules.extend(additional_rules)
return rules
# NB: This needs to be top-level for the jax compilation cache.
def _id_fn(x, ix):
"""Identity function for copying parameters to the devices, sharded."""
# A pure identity such as `lambda x, *: x` can get optimized away, so we
# include a random.split as a cheap function that cannot be optimized away.
y = random.split(random.PRNGKey(jnp.array(ix, dtype=jnp.uint32)))
return x, y
@dataclasses.dataclass
class DataLayout:
"""Represents data layout for the partitioned model."""
batch_size: int
shard_id: int
num_shards: int
is_first_host_in_replica_set: bool
PartitionedCallable = Callable[..., Any]
CompiledPartitionedCallable = Callable[..., Any]
class BasePartitioner(metaclass=abc.ABCMeta):
"""Interface for partitioning computations across hardware devices."""
def __init__(
self,
num_partitions: Optional[int] = None,
model_parallel_submesh: Optional[HardwareMesh] = None,
params_on_devices: bool = True,
backend: Optional[str] = None,
ici_mesh_shape: Optional[HardwareMesh] = None,
dcn_mesh_shape: Optional[HardwareMesh] = None,
):
"""Configures the partitioner.
Args:
num_partitions: the number of partitions to use. Ignored if
`model_parallel_submesh` is provided.
model_parallel_submesh: 4-tuple that specifies the x,y,z,c submesh to use
as the model-parallel device tile. This submesh is used for the larger
of the two parameter dimensions, and, if 2-D activation sharding is
enabled, for the model dimension of activations. The rest of the mesh is
used for data parallelism and, if 2-D parameter sharding is enabled, the
other parameter dimension.
params_on_devices: whether to keep the params on devices, if False -
params stay in the host memory. Note that some partitioners might ignore
this setting, for example if they don't support storing all params on
device memory.
backend: get devices from the pinned backend, if specified. This is useful
for explicitly specifying the devices other than relying on
jax_platform_name.
ici_mesh_shape: Shape of the logical mesh used for SPMD parallelism in
each slice. The meaning of each mesh axis is defined by mesh_axis_names,
so these two params must be the same length. If dcn_mesh_shape is
present, the overall mesh is the product of ici_mesh_shape and
dcn_mesh_shape. For example, an ici_mesh_shape of [2, 3, 4] with
mesh_axis_names ['replica', 'data', 'mdl'] indicates 2-way replica
parallelism, 3-way data parallelism, and 4-way model parallelism over 24
devices. None, the default, is equivalent to a sequence of ones and
means that the model is placed on a single device.
dcn_mesh_shape: Shape of the logical mesh used for SPMD parallelism over
multiple slices. The overall mesh is the product of ici_mesh_shape and
dcn_mesh_shape, and the meaning of each mesh axis is defined by
mesh_axis_names, so these three params must be the same length.
"""
if not num_partitions and not model_parallel_submesh:
raise ValueError('At least one of `num_partitions` or '
'`model_parallel_submesh` must be set.')
if model_parallel_submesh is not None and len(model_parallel_submesh) != 4:
logging.error(
(
'`model_parallel_submesh` must be either None or a 4-tuple. Got'
' `model_parallel_submesh`=%r. A ValueError will be raised'
' beginning March 1, 2022.'
),
model_parallel_submesh,
)
if bool(num_partitions) and bool(model_parallel_submesh):
logging.error(
'At most one of `num_partitions` or `model_parallel_submesh` can be '
'set. Got `num_partitions=%r` and `model_parallel_submesh`=%r. A '
'ValueError will be raised beginning March 21, 2022.',
num_partitions,
model_parallel_submesh,
)
self._num_partitions = num_partitions
self._model_parallel_submesh = model_parallel_submesh
self._params_on_devices = params_on_devices
if ici_mesh_shape is None or dcn_mesh_shape is None:
self._data_axis = 'data'
else:
self._data_axis = ('replica', 'data')
self._backend = backend
self._ici_mesh_shape = ici_mesh_shape
self._dcn_mesh_shape = dcn_mesh_shape
@property
def mesh(self) -> Mesh:
raise NotImplementedError
@property
def data_partition_spec(self) -> PartitionSpec:
return PartitionSpec(self._data_axis)
@property
def data_mesh_size(self) -> int:
"""Data mesh size.
Data mesh size is defined as the number of global devices involved to
carry out data parallel. Let's say we have a global mesh: ('replica': 2,
'data': 4, 'model': 2), and axes 'replica' and 'data' are responsible for
the data parallel, that means we have 2*4 = 8 devices involved - i.e., data
mesh size is 8.
Returns:
the id of the shard for the axes being replicated among the devices used
to shard the sharded_mesh_axes.
"""
data_submesh_sizes = (
[self.mesh.shape[self._data_axis]]
if isinstance(self._data_axis, str)
else [self.mesh.shape[axis] for axis in self._data_axis]
)
data_mesh_size = functools.reduce(lambda x, y: x * y, data_submesh_sizes)
return data_mesh_size
@property
def data_shards(self) -> int:
"""Number of data shards.
Let's say we are dealing with 2 slices of df4x2 TPUs. In data pipeline
we need prepare / send one data shard to each local host. This means, we
need 4 shards since we have 4 local hosts. How to infer the number of hosts
from mesh information? In this case, we have a global mesh: ('replica': 2,
'data': 8, 'model': 2). Each local host (i.e., df2x2) has this local mesh:
('replica': 1, 'data': 4, 'model': 2). By dividing global mesh with local
mesh, we can get the count of hosts.
Returns:
Number of data shards. Each shard will be sent to one local host.
"""
data_chunks = (
[self._local_chunker.num_chunks[self._data_axis]]
if isinstance(self._data_axis, str)
else [self._local_chunker.num_chunks[axis] for axis in self._data_axis]
)
data_shards = functools.reduce(lambda x, y: x * y, data_chunks)
return data_shards
@property
def data_shard_id(self) -> int:
"""Data shard id for the current host.
Returns:
Index of data shard that will be sent to the current local host.
"""
return self._local_chunker.get_shard_id(self._data_axis)
def get_data_layout(
self, batch_size: Optional[int] = None, host_index: Optional[int] = None
) -> DataLayout:
"""Returns filled `DataLayout` based on the partitioned model layout.
Args:
batch_size: if set, indicates the requested batch size. The exception will
be raised if this batch size is not compatible with the layout. If not
set, the batch size is inferred from the layout.
host_index: indicates the host index to use for the calculations, if not
set - use JAX-provided one. Should be in [0, num_hosts) interval and the
order should match the order of corresponding CPU devices in
`jax.devices()`.
Returns:
Filled `DataLayout` structure.
"""
if host_index is not None:
raise NotImplementedError('Explicit host_index is not yet implemented.')
if self._data_axis is None:
return DataLayout(
batch_size=batch_size,
shard_id=0,
num_shards=1,
is_first_host_in_replica_set=(jax.process_index() == 0))
batch_size = batch_size or self.data_mesh_size
if batch_size % self.data_mesh_size:
raise ValueError(
f'Batch size ({batch_size}) must be divisible by corresponding '
f'data mesh size ({self.data_mesh_size}).'
)
if batch_size % self.data_shards:
raise ValueError(
f'Batch size ({batch_size}) must be divisible by number of '
f'data shards ({self.data_shards}).'
)
replica_id = self._local_chunker.get_replica_id(self._data_axis)
return DataLayout(
batch_size=int(batch_size),
shard_id=int(self.data_shard_id),
num_shards=int(self.data_shards),
is_first_host_in_replica_set=(replica_id == 0),
)
def get_local_chunk_info(
self, global_shape: Tuple[int, ...],
mesh_axes: Sequence[Optional[str]]) -> LocalChunkInfo:
"""Returns the local chunk info for a given array shape and sharded axes."""
return self._local_chunker.get_local_chunk_info(global_shape, mesh_axes)
@property
def params_on_devices(self):
return self._params_on_devices
@params_on_devices.setter
def params_on_devices(self, value):
self._params_on_devices = value
def move_params_to_devices(self, train_state: TrainState,
train_state_axes: TrainState) -> TrainState:
"""Moves the optimizer parameters to devices."""
p_id_fn = self.partition(
_id_fn,
in_axis_resources=(train_state_axes, None),
out_axis_resources=(train_state_axes, None),
donate_argnums=(0,))
if jax.process_count() > 1:
train_state = host_local_array_to_global_array(
train_state, self.mesh, train_state_axes
)
train_state, _ = p_id_fn(train_state, jnp.ones((), dtype=jnp.uint32))
return train_state
@property
@abc.abstractmethod
def _local_chunker(self):
"""Returns the chunker that matches the parameters of this partitioner."""
raise NotImplementedError
def get_logical_axes(self, train_state: TrainState) -> TrainState:
"""Returns a copy of TrainState with Optional[AxisNames] as leaves."""
# By default, return None for the logical axes.
return train_state.restore_state(
jax.tree.map(lambda x: None, train_state.state_dict())
)
def get_mesh_axes(self, train_state: TrainState) -> TrainState:
"""Returns a copy of TrainState with Optional[PartitionSpecs] as leaves."""
raise NotImplementedError
@abc.abstractmethod
def partition(
self,
fn: Callable, # pylint: disable=g-bare-generic
in_axis_resources,
out_axis_resources,
static_argnums: Union[int, Sequence[int]] = (),
donate_argnums: Union[int, Sequence[int]] = ()
) -> PartitionedCallable:
"""Partitions the computation using partitioner-specific implementation.
Args:
fn: the function to partition.
in_axis_resources: Pytree of structure matching that of arguments to `fn`,
with all actual arguments replaced by resource assignment
specifications. It is also valid to specify a pytree prefix (e.g. one
value in place of a whole subtree), in which case the leaves get
broadcast to all values in that subtree.
The valid resource assignment specifications are:
`None`: in which case the value will be replicated on all devices
`PartitionSpec`: a tuple of length at most equal to the rank of the
partitioned value. Each element can be a `None`, a mesh axis or a
tuple of mesh axes, and specifies the set of resources assigned to
partition the value's dimension matching its position in the spec.
out_axis_resources: Like `in_axis_resources`, but specifies resource
assignment for function outputs.
static_argnums: an optional int or collection of ints that specify which
positional arguments to treat as static (compile-time constant) in the
partitioned function.
donate_argnums: an optional int or collection of ints that specify which
argument buffers are "donated" to the computation. It is safe to donate
argument buffers if you no longer need them once the computation has
finished.
Returns:
A partitioned version of the input function.
"""
raise NotImplementedError
@abc.abstractmethod
def compile(self, partitioned_fn: PartitionedCallable,
*args) -> CompiledPartitionedCallable:
"""Compiles and returns the partitioned function, or the original.
Args:
partitioned_fn: The partitioned function.
*args: Sample arguments to the partitioned function matching the input
shapes that will be passed to the compiled function.
Returns:
The compiled function, or the original if this partitioner does not
support compilation.
"""
raise NotImplementedError
class PjittedFnWithContext(PartitionedCallable):
"""Wraps pjitted function to apply the appropriate contexts."""
def __init__(self,
pjitted_fn,
partition_mesh: Mesh,
logical_axis_rules: flax_partitioning.LogicalRules = ()):
self._pjitted_fn = pjitted_fn
self._mesh = partition_mesh
self._logical_axis_rules = logical_axis_rules
def __call__(self, *args, **kwargs):
with Mesh(self._mesh.devices,
self._mesh.axis_names), flax_partitioning.axis_rules(
self._logical_axis_rules):
return self._pjitted_fn(*args, **kwargs)
def lower(self, *args, **kwargs):
with Mesh(self._mesh.devices,
self._mesh.axis_names), flax_partitioning.axis_rules(
self._logical_axis_rules):
return self._pjitted_fn.lower(*args, **kwargs)
class BasePjitPartitioner(BasePartitioner):
"""Partitioner that uses T5X version of jax.pjit."""
@cached_property
def _local_chunker(self) -> LocalChunker:
return LocalChunker(self.mesh)
@cached_property
def mesh(self) -> Mesh:
return default_mesh(
self._num_partitions,
self._model_parallel_submesh,
self._backend,
self._ici_mesh_shape,
self._dcn_mesh_shape,
)
def partition(
self,
fn: Callable, # pylint: disable=g-bare-generic
in_axis_resources,
out_axis_resources,
static_argnums: Union[int, Sequence[int]] = (),
donate_argnums: Union[int, Sequence[int]] = (),
) -> PjittedFnWithContext:
pjitted = pjit(
fn,
in_shardings=in_axis_resources,
out_shardings=out_axis_resources,
static_argnums=static_argnums,
donate_argnums=donate_argnums,
)
return PjittedFnWithContext(pjitted, self.mesh)
def compile(self, partitioned_fn: PjittedFnWithContext,
*args) -> CompiledPartitionedCallable:
return partitioned_fn.lower(*args).compile()