Repository navigation
Expand file tree
/
Copy pathsetup.py
More file actions
1036 lines (930 loc) · 41.1 KB
/
Copy pathsetup.py
File metadata and controls
1036 lines (930 loc) · 41.1 KB
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 (c) 2024, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
import ast
import itertools
import os
import platform
import re
import shutil
import stat
import subprocess
import sys
import sysconfig
import tarfile
import urllib.error
import urllib.request
import warnings
from pathlib import Path
import torch
from packaging.version import Version, parse
from setuptools import find_packages, setup
from torch.utils.cpp_extension import (
CUDA_HOME,
BuildExtension,
CppExtension,
CUDAExtension,
)
from wheel.bdist_wheel import bdist_wheel as _bdist_wheel
# Source directory for the lite_attention package
SRC_DIR = "hopper"
# ninja build does not work unless include_dirs are abs path
this_dir = os.path.dirname(os.path.abspath(__file__))
PACKAGE_NAME = "lite_attention"
BASE_WHEEL_URL = "https://github.com/Dao-AILab/flash-attention/releases/download/{tag_name}/{wheel_name}"
# FORCE_BUILD: Force a fresh build locally, instead of attempting to find prebuilt wheels
# SKIP_CUDA_BUILD: Intended to allow CI to use a simple `python setup.py sdist` run to copy over raw files, without any cuda compilation
FORCE_BUILD = os.getenv("LITE_ATTENTION_FORCE_BUILD", "TRUE") == "TRUE"
SKIP_CUDA_BUILD = os.getenv("LITE_ATTENTION_SKIP_CUDA_BUILD", "FALSE") == "TRUE"
# For CI, we want the option to build with C++11 ABI since the nvcr images use C++11 ABI
FORCE_CXX11_ABI = os.getenv("LITE_ATTENTION_FORCE_CXX11_ABI", "FALSE") == "TRUE"
DISABLE_BACKWARD = os.getenv("LITE_ATTENTION_DISABLE_BACKWARD", "TRUE") == "TRUE"
DISABLE_SPLIT = os.getenv("LITE_ATTENTION_DISABLE_SPLIT", "TRUE") == "TRUE"
DISABLE_PAGEDKV = os.getenv("LITE_ATTENTION_DISABLE_PAGEDKV", "TRUE") == "TRUE"
DISABLE_APPENDKV = os.getenv("LITE_ATTENTION_DISABLE_APPENDKV", "TRUE") == "TRUE"
DISABLE_LOCAL = os.getenv("LITE_ATTENTION_DISABLE_LOCAL", "TRUE") == "TRUE"
DISABLE_SOFTCAP = os.getenv("LITE_ATTENTION_DISABLE_SOFTCAP", "TRUE") == "TRUE"
DISABLE_PACKGQA = os.getenv("LITE_ATTENTION_DISABLE_PACKGQA", "TRUE") == "TRUE"
DISABLE_FP16 = os.getenv("LITE_ATTENTION_DISABLE_FP16", "TRUE") == "TRUE"
DISABLE_FP8 = os.getenv("LITE_ATTENTION_DISABLE_FP8", "TRUE") == "TRUE"
DISABLE_INT8 = os.getenv("LITE_ATTENTION_DISABLE_INT8", "FALSE") == "TRUE"
DISABLE_VARLEN = os.getenv("LITE_ATTENTION_DISABLE_VARLEN", "TRUE") == "TRUE"
DISABLE_CLUSTER = os.getenv("LITE_ATTENTION_DISABLE_CLUSTER", "TRUE") == "TRUE"
DISABLE_HDIM64 = os.getenv("LITE_ATTENTION_DISABLE_HDIM64", "FALSE") == "TRUE"
DISABLE_HDIM96 = os.getenv("LITE_ATTENTION_DISABLE_HDIM96", "FALSE") == "TRUE"
DISABLE_HDIM128 = os.getenv("LITE_ATTENTION_DISABLE_HDIM128", "FALSE") == "TRUE"
DISABLE_HDIM192 = os.getenv("LITE_ATTENTION_DISABLE_HDIM192", "FALSE") == "TRUE"
DISABLE_HDIM256 = os.getenv("LITE_ATTENTION_DISABLE_HDIM256", "FALSE") == "TRUE"
DISABLE_SM8x = os.getenv("LITE_ATTENTION_DISABLE_SM80", "TRUE") == "TRUE"
ENABLE_VCOLMAJOR = os.getenv("LITE_ATTENTION_ENABLE_VCOLMAJOR", "FALSE") == "TRUE"
DISABLE_HDIMDIFF64 = os.getenv("LITE_ATTENTION_DISABLE_HDIMDIFF64", "TRUE") == "TRUE"
DISABLE_HDIMDIFF192 = os.getenv("LITE_ATTENTION_DISABLE_HDIMDIFF192", "TRUE") == "TRUE"
# HACK: we monkey patch pytorch's _write_ninja_file to pass
# "-gencode arch=compute_sm90a,code=sm_90a" to files ending in '_sm90.cu',
# and pass "-gencode arch=compute_sm80,code=sm_80" to files ending in '_sm80.cu'
from torch.utils.cpp_extension import (
COMMON_HIP_FLAGS,
IS_HIP_EXTENSION,
IS_WINDOWS,
SUBPROCESS_DECODE_ARGS,
_is_cuda_file,
_join_cuda_home,
_join_rocm_home,
_maybe_write,
get_cxx_compiler,
)
def _write_ninja_file(
path,
cflags,
post_cflags,
cuda_cflags,
cuda_post_cflags,
cuda_dlink_post_cflags,
sources,
objects,
ldflags,
library_target,
with_cuda,
**kwargs, # kwargs (ignored) to absorb new flags in torch.utils.cpp_extension
) -> None:
r"""Write a ninja file that does the desired compiling and linking.
`path`: Where to write this file
`cflags`: list of flags to pass to $cxx. Can be None.
`post_cflags`: list of flags to append to the $cxx invocation. Can be None.
`cuda_cflags`: list of flags to pass to $nvcc. Can be None.
`cuda_postflags`: list of flags to append to the $nvcc invocation. Can be None.
`sources`: list of paths to source files
`objects`: list of desired paths to objects, one per source.
`ldflags`: list of flags to pass to linker. Can be None.
`library_target`: Name of the output library. Can be None; in that case,
we do no linking.
`with_cuda`: If we should be compiling with CUDA.
"""
def sanitize_flags(flags):
if flags is None:
return []
else:
return [flag.strip() for flag in flags]
cflags = sanitize_flags(cflags)
post_cflags = sanitize_flags(post_cflags)
cuda_cflags = sanitize_flags(cuda_cflags)
cuda_post_cflags = sanitize_flags(cuda_post_cflags)
cuda_dlink_post_cflags = sanitize_flags(cuda_dlink_post_cflags)
ldflags = sanitize_flags(ldflags)
# Sanity checks...
assert len(sources) == len(objects)
assert len(sources) > 0
compiler = get_cxx_compiler()
# Version 1.3 is required for the `deps` directive.
config = ["ninja_required_version = 1.3"]
config.append(f"cxx = {compiler}")
if with_cuda or cuda_dlink_post_cflags:
if IS_HIP_EXTENSION:
nvcc = _join_rocm_home("bin", "hipcc")
else:
nvcc = _join_cuda_home("bin", "nvcc")
if "PYTORCH_NVCC" in os.environ:
nvcc_from_env = os.getenv(
"PYTORCH_NVCC"
) # user can set nvcc compiler with ccache using the environment variable here
else:
nvcc_from_env = nvcc
config.append(f"nvcc_from_env = {nvcc_from_env}")
config.append(f"nvcc = {nvcc}")
if IS_HIP_EXTENSION:
post_cflags = COMMON_HIP_FLAGS + post_cflags
flags = [f"cflags = {' '.join(cflags)}"]
flags.append(f"post_cflags = {' '.join(post_cflags)}")
if with_cuda:
flags.append(f"cuda_cflags = {' '.join(cuda_cflags)}")
flags.append(f"cuda_post_cflags = {' '.join(cuda_post_cflags)}")
cuda_post_cflags_sm80 = [
s if s != "arch=compute_90a,code=sm_90a" else "arch=compute_80,code=sm_80"
for s in cuda_post_cflags
]
flags.append(f"cuda_post_cflags_sm80 = {' '.join(cuda_post_cflags_sm80)}")
if IS_HIP_EXTENSION:
cuda_post_cflags_sm80_sm90 = cuda_post_cflags
else:
cuda_post_cflags_sm80_sm90 = cuda_post_cflags + [
"-gencode",
"arch=compute_80,code=sm_80",
]
flags.append(
f"cuda_post_cflags_sm80_sm90 = {' '.join(cuda_post_cflags_sm80_sm90)}"
)
cuda_post_cflags_sm100 = [
s
if s != "arch=compute_90a,code=sm_90a"
else "arch=compute_100a,code=sm_100a"
for s in cuda_post_cflags
]
flags.append(f"cuda_post_cflags_sm100 = {' '.join(cuda_post_cflags_sm100)}")
flags.append(f"cuda_dlink_post_cflags = {' '.join(cuda_dlink_post_cflags)}")
flags.append(f"ldflags = {' '.join(ldflags)}")
# Turn into absolute paths so we can emit them into the ninja build
# file wherever it is.
sources = [os.path.abspath(file) for file in sources]
# See https://ninja-build.org/build.ninja.html for reference.
compile_rule = ["rule compile"]
if IS_WINDOWS:
compile_rule.append(
" command = cl /showIncludes $cflags -c $in /Fo$out $post_cflags"
)
compile_rule.append(" deps = msvc")
else:
compile_rule.append(
" command = $cxx -MMD -MF $out.d $cflags -c $in -o $out $post_cflags"
)
compile_rule.append(" depfile = $out.d")
compile_rule.append(" deps = gcc")
if with_cuda:
cuda_compile_rule = ["rule cuda_compile"]
nvcc_gendeps = ""
# --generate-dependencies-with-compile is not supported by ROCm
# Nvcc flag `--generate-dependencies-with-compile` is not supported by sccache, which may increase build time.
if (
torch.version.cuda is not None
and os.getenv("TORCH_EXTENSION_SKIP_NVCC_GEN_DEPENDENCIES", "0") != "1"
):
cuda_compile_rule.append(" depfile = $out.d")
cuda_compile_rule.append(" deps = gcc")
# Note: non-system deps with nvcc are only supported
# on Linux so use --generate-dependencies-with-compile
# to make this work on Windows too.
nvcc_gendeps = (
"--generate-dependencies-with-compile --dependency-output $out.d"
)
cuda_compile_rule_sm80 = (
["rule cuda_compile_sm80"]
+ cuda_compile_rule[1:]
+ [
f" command = $nvcc_from_env {nvcc_gendeps} $cuda_cflags -c $in -o $out $cuda_post_cflags_sm80"
]
)
cuda_compile_rule_sm80_sm90 = (
["rule cuda_compile_sm80_sm90"]
+ cuda_compile_rule[1:]
+ [
f" command = $nvcc_from_env {nvcc_gendeps} $cuda_cflags -c $in -o $out $cuda_post_cflags_sm80_sm90"
]
)
cuda_compile_rule_sm100 = (
["rule cuda_compile_sm100"]
+ cuda_compile_rule[1:]
+ [
f" command = $nvcc_from_env {nvcc_gendeps} $cuda_cflags -c $in -o $out $cuda_post_cflags_sm100"
]
)
cuda_compile_rule.append(
f" command = $nvcc_from_env {nvcc_gendeps} $cuda_cflags -c $in -o $out $cuda_post_cflags"
)
# Emit one build rule per source to enable incremental build.
build = []
for source_file, object_file in zip(sources, objects):
is_cuda_source = _is_cuda_file(source_file) and with_cuda
if is_cuda_source:
if source_file.endswith("_sm90.cu"):
rule = "cuda_compile"
elif source_file.endswith("_sm80.cu"):
rule = "cuda_compile_sm80"
elif source_file.endswith("_sm100.cu"):
rule = "cuda_compile_sm100"
else:
rule = "cuda_compile_sm80_sm90"
else:
rule = "compile"
if IS_WINDOWS:
source_file = source_file.replace(":", "$:")
object_file = object_file.replace(":", "$:")
source_file = source_file.replace(" ", "$ ")
object_file = object_file.replace(" ", "$ ")
build.append(f"build {object_file}: {rule} {source_file}")
if cuda_dlink_post_cflags:
devlink_out = os.path.join(os.path.dirname(objects[0]), "dlink.o")
devlink_rule = ["rule cuda_devlink"]
devlink_rule.append(" command = $nvcc $in -o $out $cuda_dlink_post_cflags")
devlink = [f"build {devlink_out}: cuda_devlink {' '.join(objects)}"]
objects += [devlink_out]
else:
devlink_rule, devlink = [], []
if library_target is not None:
link_rule = ["rule link"]
if IS_WINDOWS:
cl_paths = (
subprocess.check_output(["where", "cl"])
.decode(*SUBPROCESS_DECODE_ARGS)
.split("\r\n")
)
if len(cl_paths) >= 1:
cl_path = os.path.dirname(cl_paths[0]).replace(":", "$:")
else:
raise RuntimeError("MSVC is required to load C++ extensions")
link_rule.append(
f' command = "{cl_path}/link.exe" $in /nologo $ldflags /out:$out'
)
else:
link_rule.append(" command = $cxx $in $ldflags -o $out")
link = [f"build {library_target}: link {' '.join(objects)}"]
default = [f"default {library_target}"]
else:
link_rule, link, default = [], [], []
# 'Blocks' should be separated by newlines, for visual benefit.
blocks = [config, flags, compile_rule]
if with_cuda:
blocks.append(cuda_compile_rule) # type: ignore[possibly-undefined]
blocks.append(cuda_compile_rule_sm80) # type: ignore[possibly-undefined]
blocks.append(cuda_compile_rule_sm80_sm90) # type: ignore[possibly-undefined]
blocks.append(cuda_compile_rule_sm100) # type: ignore[possibly-undefined]
blocks += [devlink_rule, link_rule, build, devlink, link, default]
content = "\n\n".join("\n".join(b) for b in blocks)
# Ninja requires a new lines at the end of the .ninja file
content += "\n"
_maybe_write(path, content)
# Monkey patching
torch.utils.cpp_extension._write_ninja_file = _write_ninja_file
def get_platform():
"""
Returns the platform name as used in wheel filenames.
"""
if sys.platform.startswith("linux"):
return "linux_x86_64"
elif sys.platform == "darwin":
mac_version = ".".join(platform.mac_ver()[0].split(".")[:2])
return f"macosx_{mac_version}_x86_64"
elif sys.platform == "win32":
return "win_amd64"
else:
raise ValueError("Unsupported platform: {}".format(sys.platform))
def get_cuda_bare_metal_version(cuda_dir):
raw_output = subprocess.check_output(
[cuda_dir + "/bin/nvcc", "-V"], universal_newlines=True
)
output = raw_output.split()
release_idx = output.index("release") + 1
bare_metal_version = parse(output[release_idx].split(",")[0])
return raw_output, bare_metal_version
def check_if_cuda_home_none(global_option: str) -> None:
if CUDA_HOME is not None:
return
# warn instead of error because user could be downloading prebuilt wheels, so nvcc won't be necessary
# in that case.
warnings.warn(
f"{global_option} was requested, but nvcc was not found. Are you sure your environment has nvcc available? "
"If you're installing within a container from https://hub.docker.com/r/pytorch/pytorch, "
"only images whose names contain 'devel' will provide nvcc."
)
# Taken from https://github.com/pytorch/pytorch/blob/master/tools/setup_helpers/env.py
def check_env_flag(name: str, default: str = "") -> bool:
return os.getenv(name, default).upper() in ["ON", "1", "YES", "TRUE", "Y"]
# Copied from https://github.com/triton-lang/triton/blob/main/python/setup.py
def is_offline_build() -> bool:
"""
Downstream projects and distributions which bootstrap their own dependencies from scratch
and run builds in offline sandboxes
may set `LITE_ATTENTION_OFFLINE_BUILD` in the build environment to prevent any attempts at downloading
pinned dependencies from the internet or at using dependencies vendored in-tree.
Dependencies must be defined using respective search paths (cf. `syspath_var_name` in `Package`).
Missing dependencies lead to an early abortion.
Dependencies' compatibility is not verified.
Note that this flag isn't tested by the CI and does not provide any guarantees.
"""
return check_env_flag("LITE_ATTENTION_OFFLINE_BUILD", "")
# Copied from https://github.com/triton-lang/triton/blob/main/python/setup.py
def get_lite_attention_cache_path():
user_home = os.getenv("LITE_ATTENTION_HOME")
if not user_home:
user_home = (
os.getenv("HOME")
or os.getenv("USERPROFILE")
or os.getenv("HOMEPATH")
or None
)
if not user_home:
raise RuntimeError("Could not find user home directory")
return os.path.join(user_home, ".lite_attention")
def open_url(url):
user_agent = (
"Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/119.0"
)
headers = {
"User-Agent": user_agent,
}
request = urllib.request.Request(url, None, headers)
# Set timeout to 300 seconds to prevent the request from hanging forever.
return urllib.request.urlopen(request, timeout=300)
def download_and_copy(name, src_func, dst_path, version, url_func):
if is_offline_build():
return
LITEATTn_cache_path = get_lite_attention_cache_path()
base_dir = os.path.dirname(__file__)
system = platform.system()
arch = platform.machine()
arch = {"arm64": "aarch64"}.get(arch, arch)
supported = {"Linux": "linux", "Darwin": "linux"}
url = url_func(supported[system], arch, version)
src_path = src_func(supported[system], arch, version)
tmp_path = os.path.join(
LITEATTn_cache_path, "nvidia", name
) # path to cache the download
dst_path = os.path.join(
base_dir, "third_party", "nvidia", "backend", dst_path
) # final binary path
src_path = os.path.join(tmp_path, src_path)
download = not os.path.exists(src_path)
if download:
print(f"downloading and extracting {url} ...")
file = tarfile.open(fileobj=open_url(url), mode="r|*")
file.extractall(path=tmp_path)
os.makedirs(os.path.split(dst_path)[0], exist_ok=True)
print(f"copy {src_path} to {dst_path} ...")
if os.path.isdir(src_path):
shutil.copytree(src_path, dst_path, dirs_exist_ok=True)
else:
shutil.copy(src_path, dst_path)
def nvcc_threads_args():
nvcc_threads = os.getenv("NVCC_THREADS") or "2"
return ["--threads", nvcc_threads]
# NVIDIA_TOOLCHAIN_VERSION = {"nvcc": "12.3.107"}
NVIDIA_TOOLCHAIN_VERSION = {"nvcc": "12.6.85", "ptxas": "12.8.93"}
exe_extension = sysconfig.get_config_var("EXE")
cmdclass = {}
ext_modules = []
# We want this even if SKIP_CUDA_BUILD because when we run python setup.py sdist we want the .hpp
# files included in the source distribution, in case the user compiles from source.
subprocess.run(["git", "submodule", "update", "--init", "csrc/cutlass"])
# Prefer ROCm path when torch has HIP, or when ROCm toolchain is present (e.g. on AMD machine
# where pip's build isolation may install CUDA torch)
def _rocm_toolchain_present():
if os.environ.get("ROCM_PATH") or os.environ.get("HIP_PATH"):
return True
return shutil.which("hipcc") is not None
is_rocm = (torch.version.hip is not None) or _rocm_toolchain_present()
if is_rocm:
print("\n\nBuilding for AMD ROCm...\n\n")
os.environ["PYTORCH_ROCM_ARCH"] = "gfx942"
os.environ["TORCH_ROCM_ARCH_LIST"] = "gfx942"
os.environ["ROCM_ARCH"] = "gfx942"
# Head dims to generate for ROCm (same set as typical NVIDIA build for parity)
ROCM_HDIMS = [64, 96, 128, 192, 256]
rocm_optdim = ",".join(str(h) for h in ROCM_HDIMS)
# 1. Generate Kernels
repo_root = Path(this_dir)
ck_dir = repo_root / "csrc" / "composable_kernel"
generate_script = ck_dir / "example" / "ck_tile" / "01_fmha" / "generate.py"
generated_dir = Path(this_dir) / "generated_ck"
# Always clean and recreate to avoid stale kernels from a different CK version.
# Use multiple approaches for robustness on various filesystems.
if generated_dir.exists():
# First try subprocess rm -rf (handles NFS stale handles and special files)
subprocess.run(["rm", "-rf", str(generated_dir)], check=False)
if generated_dir.exists():
# Fallback: delete files one by one using Python
for f in list(generated_dir.iterdir()):
try:
f.unlink(missing_ok=True)
except Exception:
pass
try:
generated_dir.rmdir()
except Exception:
shutil.rmtree(str(generated_dir), ignore_errors=True)
generated_dir.mkdir(parents=True, exist_ok=True)
# Build FWD filter: for each head dim, lite + baseline (qr nbias nmask) for fp16 and bf16
fwd_filter_parts = []
for h in ROCM_HDIMS:
fwd_filter_parts.append(
f"*d{h}_fp16_batch*lite*|*d{h}_fp16_batch*qr*nbias*nmask*"
f"|*d{h}_bf16_batch*lite*|*d{h}_bf16_batch*qr*nbias*nmask*"
)
fwd_filter = "|".join(fwd_filter_parts)
print(f"Generating FWD kernels to {generated_dir} (head_dims={ROCM_HDIMS})...")
subprocess.run(
[
sys.executable,
str(generate_script),
"--api",
"fwd",
"--optdim",
rocm_optdim,
"--output_dir",
str(generated_dir),
"--filter",
fwd_filter,
],
check=True,
)
# Build BWD filter: for each head dim, fp16 and bf16
bwd_filter_parts = [f"*d{h}_fp16_batch*|*d{h}_bf16_batch*" for h in ROCM_HDIMS]
bwd_filter = "|".join(bwd_filter_parts)
print(f"Generating BWD kernels to {generated_dir} (head_dims={ROCM_HDIMS})...")
subprocess.run(
[
sys.executable,
str(generate_script),
"--api",
"bwd",
"--optdim",
rocm_optdim,
"--output_dir",
str(generated_dir),
"--filter",
bwd_filter,
],
check=True,
)
# Generate APPENDKV kernels (receipt=2 = Flash attention integration: fp16/bf16 row-layout only)
print(
f"Generating FWD_APPENDKV kernels to {generated_dir} (head_dims={ROCM_HDIMS})..."
)
subprocess.run(
[
sys.executable,
str(generate_script),
"--api",
"fwd_appendkv",
"--optdim",
rocm_optdim,
"--output_dir",
str(generated_dir),
"--receipt",
"2",
],
check=True,
)
# Generate SPLITKV kernels with sync-only (qr_vr/qr_vc) pipeline variants.
# receipt=2 limits to fp16/bf16/row/no-bias, and filter "@*qr_vr*|*qr_vc*" excludes
# qr_async variants (which reference BlockFmhaFwdSplitKVPipelineQRKSVSAsync that
# doesn't exist in the current CK headers). The '@' splits combine vs main filters.
print(
f"Generating FWD_SPLITKV kernels to {generated_dir} (head_dims={ROCM_HDIMS})..."
)
subprocess.run(
[
sys.executable,
str(generate_script),
"--api",
"fwd_splitkv",
"--optdim",
rocm_optdim,
"--output_dir",
str(generated_dir),
"--receipt",
"2",
"--filter",
"@*qr_vr*",
],
check=True,
)
# Rename generated .cpp files to .hip so they are compiled with hipcc as device code
for p in list(generated_dir.glob("*.cpp")):
hip_path = p.with_suffix(".hip")
try:
if hip_path.exists():
hip_path.unlink(missing_ok=True)
if p.exists():
p.rename(hip_path)
elif not hip_path.exists():
# file was already renamed or missing; use shutil.copy as fallback
pass
except (FileNotFoundError, OSError):
# On some filesystems (NFS/CIFS) rename can fail transiently; try copy+delete
try:
import shutil as _shutil
_shutil.copy2(str(p), str(hip_path))
p.unlink(missing_ok=True)
except Exception:
pass
# Copy flash_api.cpp -> flash_api.hip so we only maintain the .cpp (single source of truth)
flash_attn_ck_dir = repo_root / "csrc" / "flash_attn_ck"
shutil.copy(
flash_attn_ck_dir / "flash_api.cpp", flash_attn_ck_dir / "flash_api.hip"
)
# 2. Collect Sources
def get_rocm_source(path_str):
path = Path(path_str)
if path.suffix == ".cpp":
hip_path = path.with_suffix(".hip")
if hip_path.exists():
return str(hip_path)
return str(path)
sources = [
get_rocm_source(str(flash_attn_ck_dir / "flash_api.cpp")),
get_rocm_source(str(flash_attn_ck_dir / "flash_common.cpp")),
get_rocm_source(str(flash_attn_ck_dir / "mha_fwd.cpp")),
get_rocm_source(str(flash_attn_ck_dir / "mha_varlen_fwd.cpp")),
get_rocm_source(str(flash_attn_ck_dir / "mha_fwd_kvcache.cpp")),
str(generated_dir / "fmha_fwd_api.hip"),
# mha_bwd is always registered in flash_api.hip without a DISABLE_BACKWARD guard
get_rocm_source(str(flash_attn_ck_dir / "mha_bwd.cpp")),
str(generated_dir / "fmha_bwd_api.hip"),
]
if not DISABLE_VARLEN:
sources.append(get_rocm_source(str(flash_attn_ck_dir / "mha_varlen_bwd.cpp")))
sources.extend(
str(p)
for p in generated_dir.glob("*.hip")
if p.name not in ["fmha_fwd_api.hip", "fmha_bwd_api.hip"]
and not p.name.endswith(
"_hip.hip"
) # exclude hipify output files (stale from prior builds)
)
# 3. Include Directories
rocm_include_dirs = [
str(flash_attn_ck_dir),
str(ck_dir / "include"),
str(ck_dir / "library" / "include"),
str(ck_dir / "example" / "ck_tile" / "01_fmha"),
str(generated_dir),
]
# 4. Compiler Flags (ROCm/HIP uses hipcc which is Clang-based; -Wno-c++11-narrowing avoids
# "float cannot be narrowed to index_t" when CK args use float threshold and compiler is strict)
rocm_feature_args = (
[]
+ (["-DLITEATTENTION_DISABLE_BACKWARD"] if DISABLE_BACKWARD else [])
+ (["-DLITEATTENTION_DISABLE_PAGEDKV"] if DISABLE_PAGEDKV else [])
+ (["-DLITEATTENTION_DISABLE_SPLIT"] if DISABLE_SPLIT else [])
+ (["-DLITEATTENTION_DISABLE_APPENDKV"] if DISABLE_APPENDKV else [])
+ (["-DLITEATTENTION_DISABLE_LOCAL"] if DISABLE_LOCAL else [])
+ (["-DLITEATTENTION_DISABLE_SOFTCAP"] if DISABLE_SOFTCAP else [])
+ (["-DLITEATTENTION_DISABLE_PACKGQA"] if DISABLE_PACKGQA else [])
+ (["-DLITEATTENTION_DISABLE_FP16"] if DISABLE_FP16 else [])
+ (["-DLITEATTENTION_DISABLE_FP8"] if DISABLE_FP8 else [])
+ (["-DLITEATTENTION_DISABLE_INT8"] if DISABLE_INT8 else [])
+ (["-DLITEATTENTION_DISABLE_VARLEN"] if DISABLE_VARLEN else [])
+ (["-DLITEATTENTION_DISABLE_CLUSTER"] if DISABLE_CLUSTER else [])
+ (["-DLITEATTENTION_DISABLE_HDIM64"] if DISABLE_HDIM64 else [])
+ (["-DLITEATTENTION_DISABLE_HDIM96"] if DISABLE_HDIM96 else [])
+ (["-DLITEATTENTION_DISABLE_HDIM128"] if DISABLE_HDIM128 else [])
+ (["-DLITEATTENTION_DISABLE_HDIM192"] if DISABLE_HDIM192 else [])
+ (["-DLITEATTENTION_DISABLE_HDIM256"] if DISABLE_HDIM256 else [])
+ (["-DCK_TILE_FMHA_FWD_FAST_EXP2=1"])
)
extra_compile_args = {
"cxx": [
"-O3",
"-std=c++17",
"-Wno-c++11-narrowing",
"-D__HIP_PLATFORM_AMD__=1",
"-DTORCH_EXTENSION_NAME=lite_attention._C",
]
+ rocm_feature_args,
"nvcc": [
"-O3",
"-std=c++17",
"-Wno-c++11-narrowing",
"-D__HIP_PLATFORM_AMD__=1",
"--offload-arch=gfx942",
]
+ rocm_feature_args, # nvcc maps to hipcc in CUDAExtension when IS_HIP_EXTENSION is true
}
ext_modules.append(
CUDAExtension(
name="lite_attention._C",
sources=sources,
include_dirs=rocm_include_dirs,
extra_compile_args=extra_compile_args,
)
)
if not SKIP_CUDA_BUILD and not is_rocm:
print("\n\ntorch.__version__ = {}\n\n".format(torch.__version__))
TORCH_MAJOR = int(torch.__version__.split(".")[0])
TORCH_MINOR = int(torch.__version__.split(".")[1])
check_if_cuda_home_none(PACKAGE_NAME)
_, bare_metal_version = get_cuda_bare_metal_version(CUDA_HOME)
if bare_metal_version < Version("12.3"):
raise RuntimeError("Lite Attention is only supported on CUDA 12.3 and above")
# ptxas 12.8 gives the best perf currently
# We want to use the nvcc front end from 12.6 however, since if we use nvcc 12.8
# Cutlass 3.8 will expect the new data types in cuda.h from CTK 12.8, which we don't have.
# For CUDA 13.0+, use the system compiler directly
if bare_metal_version != Version("12.8") and bare_metal_version < Version("13.0"):
download_and_copy(
name="nvcc",
src_func=lambda system, arch, version: (
f"cuda_nvcc-{system}-{arch}-{version}-archive/bin"
),
dst_path="bin",
version=NVIDIA_TOOLCHAIN_VERSION["nvcc"],
url_func=lambda system, arch, version: (
f"https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/{system}-{arch}/cuda_nvcc-{system}-{arch}-{version}-archive.tar.xz"
),
)
download_and_copy(
name="ptxas",
src_func=lambda system, arch, version: (
f"cuda_nvcc-{system}-{arch}-{version}-archive/bin/ptxas"
),
dst_path="bin",
version=NVIDIA_TOOLCHAIN_VERSION["ptxas"],
url_func=lambda system, arch, version: (
f"https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/{system}-{arch}/cuda_nvcc-{system}-{arch}-{version}-archive.tar.xz"
),
)
download_and_copy(
name="ptxas",
src_func=lambda system, arch, version: (
f"cuda_nvcc-{system}-{arch}-{version}-archive/nvvm/bin"
),
dst_path="nvvm/bin",
version=NVIDIA_TOOLCHAIN_VERSION["ptxas"],
url_func=lambda system, arch, version: (
f"https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/{system}-{arch}/cuda_nvcc-{system}-{arch}-{version}-archive.tar.xz"
),
)
base_dir = os.path.dirname(__file__)
ctk_path_new = os.path.abspath(
os.path.join(base_dir, "third_party", "nvidia", "backend", "bin")
)
nvcc_path_new = os.path.join(ctk_path_new, f"nvcc{exe_extension}")
# Need to append to path otherwise nvcc can't find cicc in nvvm/bin/cicc
# nvcc 12.8 seems to hard-code looking for cicc in ../nvvm/bin/cicc
os.environ["PATH"] = ctk_path_new + os.pathsep + os.environ["PATH"]
os.environ["PYTORCH_NVCC"] = nvcc_path_new
# Make nvcc executable, sometimes after the copy it loses its permissions
os.chmod(nvcc_path_new, os.stat(nvcc_path_new).st_mode | stat.S_IEXEC)
cc_flag = []
cc_flag.append("-gencode")
cc_flag.append("arch=compute_90a,code=sm_90a")
# HACK: The compiler flag -D_GLIBCXX_USE_CXX11_ABI is set to be the same as
# torch._C._GLIBCXX_USE_CXX11_ABI
# https://github.com/pytorch/pytorch/blob/8472c24e3b5b60150096486616d98b7bea01500b/torch/utils/cpp_extension.py#L920
if FORCE_CXX11_ABI:
torch._C._GLIBCXX_USE_CXX11_ABI = True
repo_dir = Path(this_dir)
cutlass_dir = repo_dir / "csrc" / "cutlass"
feature_args = (
[]
+ (["-DFLASHATTENTION_DISABLE_BACKWARD"] if DISABLE_BACKWARD else [])
+ (["-DFLASHATTENTION_DISABLE_PAGEDKV"] if DISABLE_PAGEDKV else [])
+ (["-DFLASHATTENTION_DISABLE_SPLIT"] if DISABLE_SPLIT else [])
+ (["-DFLASHATTENTION_DISABLE_APPENDKV"] if DISABLE_APPENDKV else [])
+ (["-DFLASHATTENTION_DISABLE_LOCAL"] if DISABLE_LOCAL else [])
+ (["-DFLASHATTENTION_DISABLE_SOFTCAP"] if DISABLE_SOFTCAP else [])
+ (["-DFLASHATTENTION_DISABLE_PACKGQA"] if DISABLE_PACKGQA else [])
+ (["-DFLASHATTENTION_DISABLE_FP16"] if DISABLE_FP16 else [])
+ (["-DFLASHATTENTION_DISABLE_FP8"] if DISABLE_FP8 else [])
+ (["-DFLASHATTENTION_DISABLE_INT8"] if DISABLE_INT8 else [])
+ (["-DFLASHATTENTION_DISABLE_VARLEN"] if DISABLE_VARLEN else [])
+ (["-DFLASHATTENTION_DISABLE_CLUSTER"] if DISABLE_CLUSTER else [])
+ (["-DFLASHATTENTION_DISABLE_HDIM64"] if DISABLE_HDIM64 else [])
+ (["-DFLASHATTENTION_DISABLE_HDIM96"] if DISABLE_HDIM96 else [])
+ (["-DFLASHATTENTION_DISABLE_HDIM128"] if DISABLE_HDIM128 else [])
+ (["-DFLASHATTENTION_DISABLE_HDIM192"] if DISABLE_HDIM192 else [])
+ (["-DFLASHATTENTION_DISABLE_HDIM256"] if DISABLE_HDIM256 else [])
+ (["-DFLASHATTENTION_DISABLE_SM8x"] if DISABLE_SM8x else [])
+ (["-DFLASHATTENTION_ENABLE_VCOLMAJOR"] if ENABLE_VCOLMAJOR else [])
+ (["-DFLASHATTENTION_DISABLE_HDIMDIFF64"] if DISABLE_HDIMDIFF64 else [])
+ (["-DFLASHATTENTION_DISABLE_HDIMDIFF192"] if DISABLE_HDIMDIFF192 else [])
)
DTYPE_FWD_SM80 = ["bf16"] + (["fp16"] if not DISABLE_FP16 else [])
DTYPE_FWD_SM90 = (
["bf16"]
+ (["fp16"] if not DISABLE_FP16 else [])
+ (["e4m3"] if not DISABLE_FP8 else [])
+ (["int8"] if not DISABLE_INT8 else [])
)
HALF_DTYPE_FWD_SM90 = ["bf16"] + (["fp16"] if not DISABLE_FP16 else [])
DTYPE_BWD = ["bf16"] + (["fp16"] if not DISABLE_FP16 else [])
HEAD_DIMENSIONS_BWD = (
[]
+ ([64] if not DISABLE_HDIM64 else [])
+ ([96] if not DISABLE_HDIM96 else [])
+ ([128] if not DISABLE_HDIM128 else [])
+ ([192] if not DISABLE_HDIM192 else [])
+ ([256] if not DISABLE_HDIM256 else [])
)
# build will now explode with this compilation grouping given all our templating
# HEAD_DIMENSIONS_FWD = ["all", "diff"]
HEAD_DIMENSIONS_FWD = HEAD_DIMENSIONS_BWD
HEAD_DIMENSIONS_DIFF64_FWD = (
[]
+ (["64_256"] if not DISABLE_HDIMDIFF64 else [])
+ (["64_512"] if not DISABLE_HDIMDIFF64 else [])
)
HEAD_DIMENSIONS_DIFF192_FWD = [] + (["192_128"] if not DISABLE_HDIMDIFF192 else [])
HEAD_DIMENSIONS_FWD_SM80 = HEAD_DIMENSIONS_BWD
SPLIT = [""] + (["_split"] if not DISABLE_SPLIT else [])
PAGEDKV = [""] + (["_paged"] if not DISABLE_PAGEDKV else [])
SOFTCAP = [""] + (["_softcap"] if not DISABLE_SOFTCAP else [])
SOFTCAP_ALL = [""] if DISABLE_SOFTCAP else ["_softcapall"]
PACKGQA = [""] + (["_packgqa"] if not DISABLE_PACKGQA else [])
# We already always hard-code PackGQA=true for Sm8x
sources_fwd_sm80 = [
f"{SRC_DIR}/instantiations/flash_fwd_hdim{hdim}_{dtype}{paged}{split}{softcap}_sm80.cu"
for hdim, dtype, split, paged, softcap in itertools.product(
HEAD_DIMENSIONS_FWD_SM80, DTYPE_FWD_SM80, SPLIT, PAGEDKV, SOFTCAP_ALL
)
]
# We already always hard-code PackGQA=true for Sm9x if PagedKV or Split
sources_fwd_sm90 = [
f"{SRC_DIR}/instantiations/flash_fwd_hdim{hdim}_{dtype}{paged}{split}{softcap}{packgqa}_sm90.cu"
for hdim, dtype, split, paged, softcap, packgqa in itertools.product(
HEAD_DIMENSIONS_FWD, DTYPE_FWD_SM90, SPLIT, PAGEDKV, SOFTCAP, PACKGQA
)
if not (packgqa and (paged or split))
]
if not DISABLE_HDIMDIFF64:
sources_fwd_sm90 += [
f"{SRC_DIR}/instantiations/flash_fwd_hdim{hdim}_{dtype}{paged}{split}{softcap}{packgqa}_sm90.cu"
for hdim, dtype, split, paged, softcap, packgqa in itertools.product(
HEAD_DIMENSIONS_DIFF64_FWD,
HALF_DTYPE_FWD_SM90,
SPLIT,
PAGEDKV,
SOFTCAP,
PACKGQA,
)
if not (packgqa and (paged or split))
]
if not DISABLE_HDIMDIFF192:
sources_fwd_sm90 += [
f"{SRC_DIR}/instantiations/flash_fwd_hdim{hdim}_{dtype}{paged}{split}{softcap}{packgqa}_sm90.cu"
for hdim, dtype, split, paged, softcap, packgqa in itertools.product(
HEAD_DIMENSIONS_DIFF192_FWD,
DTYPE_FWD_SM90,
SPLIT,
PAGEDKV,
SOFTCAP,
PACKGQA,
)
if not (packgqa and (paged or split))
]
sources_bwd_sm80 = [
f"{SRC_DIR}/instantiations/flash_bwd_hdim{hdim}_{dtype}{softcap}_sm80.cu"
for hdim, dtype, softcap in itertools.product(
HEAD_DIMENSIONS_BWD, DTYPE_BWD, SOFTCAP
)
]
sources_bwd_sm90 = [
f"{SRC_DIR}/instantiations/flash_bwd_hdim{hdim}_{dtype}{softcap}_sm90.cu"
for hdim, dtype, softcap in itertools.product(
HEAD_DIMENSIONS_BWD, DTYPE_BWD, SOFTCAP_ALL
)
]
if DISABLE_BACKWARD:
sources_bwd_sm90 = []
sources_bwd_sm80 = []
sources = (
[f"{SRC_DIR}/_internal/cpp/flash_api.cpp"]
+ (sources_fwd_sm80 if not DISABLE_SM8x else [])
+ sources_fwd_sm90
+ (sources_bwd_sm80 if not DISABLE_SM8x else [])
+ sources_bwd_sm90
)
if not DISABLE_SPLIT:
sources += [f"{SRC_DIR}/_internal/cpp/flash_fwd_combine.cu"]
sources += [f"{SRC_DIR}/_internal/cpp/flash_prepare_scheduler.cu"]
# Add quantization kernels for INT8 support
if not DISABLE_INT8:
sources += [f"{SRC_DIR}/_internal/cpp/quant.cu"]
nvcc_flags = [
"-O3",
# "-g",
# "-std=c++17",
"-std=c++20",
"--ftemplate-backtrace-limit=0", # To debug template code
"--use_fast_math",
# "--keep",
# "--ptxas-options=--verbose,--register-usage-level=5,--warn-on-local-memory-usage", # printing out number of registers
"--resource-usage", # printing out number of registers
# f"--split-compile={os.getenv('NVCC_THREADS', '4')}", # split-compile is faster
"-lineinfo", # TODO: disable this for release to reduce binary size
"-DCUTE_SM90_EXTENDED_MMA_SHAPES_ENABLED", # Necessary for the WGMMA shapes that we use
# "-DCUTLASS_ENABLE_GDC_FOR_SM90", # For PDL
"-DCUTLASS_DEBUG_TRACE_LEVEL=0", # Can toggle for debugging
"-DNDEBUG", # Important, otherwise performance is severely impacted
]
if get_platform() == "win_amd64":
nvcc_flags.extend(
[
"-D_USE_MATH_DEFINES", # for M_LN2
"-Xcompiler=/Zc:__cplusplus", # sets __cplusplus correctly, CUTLASS_CONSTEXPR_IF_CXX17 needed for cutlass::gcd
]
)
include_dirs = [
Path(this_dir) / SRC_DIR,
Path(this_dir) / SRC_DIR / "_internal" / "cpp",
cutlass_dir / "include",
]
ext_modules.append(
CUDAExtension(
name=f"lite_attention._C",
sources=sources,
extra_compile_args={
"cxx": ["-O3", "-std=c++17", "-DPy_LIMITED_API=0x03090000"]
+ feature_args,
"nvcc": nvcc_threads_args() + nvcc_flags + cc_flag + feature_args,
},
include_dirs=include_dirs,
py_limited_api=True,
)
)
def get_package_version():
with open(Path(this_dir) / SRC_DIR / "__init__.py", "r") as f:
version_match = re.search(r"^__version__\s*=\s*(.*)$", f.read(), re.MULTILINE)
public_version = ast.literal_eval(version_match.group(1))
local_version = os.environ.get("LITE_ATTN_LOCAL_VERSION")
if local_version:
return f"{public_version}+{local_version}"
else:
return str(public_version)
def get_wheel_url():
# Determine the version numbers that will be used to determine the correct wheel
# We're using the CUDA version used to build torch, not the one currently installed
# _, cuda_version_raw = get_cuda_bare_metal_version(CUDA_HOME)
torch_cuda_version = parse(torch.version.cuda)
torch_version_raw = parse(torch.__version__)
# For CUDA 11, we only compile for CUDA 11.8, and for CUDA 12 we only compile for CUDA 12.2
# to save CI time. Minor versions should be compatible.
torch_cuda_version = (
parse("11.8") if torch_cuda_version.major == 11 else parse("12.2")
)
python_version = f"cp{sys.version_info.major}{sys.version_info.minor}"
platform_name = get_platform()
package_version = get_package_version()
# cuda_version = f"{cuda_version_raw.major}{cuda_version_raw.minor}"
cuda_version = f"{torch_cuda_version.major}{torch_cuda_version.minor}"
torch_version = f"{torch_version_raw.major}.{torch_version_raw.minor}"
cxx11_abi = str(torch._C._GLIBCXX_USE_CXX11_ABI).upper()
# Determine wheel URL based on CUDA version, torch version, python version and OS
wheel_filename = f"{PACKAGE_NAME}-{package_version}+cu{cuda_version}torch{torch_version}cxx11abi{cxx11_abi}-{python_version}-{python_version}-{platform_name}.whl"
wheel_url = BASE_WHEEL_URL.format(
tag_name=f"v{package_version}", wheel_name=wheel_filename
)
return wheel_url, wheel_filename
class CachedWheelsCommand(_bdist_wheel):
"""
The CachedWheelsCommand plugs into the default bdist wheel, which is ran by pip when it cannot
find an existing wheel (which is currently the case for all installs). We use
the environment parameters to detect whether there is already a pre-built version of a compatible
wheel available and short-circuits the standard full build pipeline.
"""