Xuanlei Zhao
30dd13c450
[autochunk] support complete benchmark ( #3121 )
...
* refact memory code
* dont log free var memory
* add memory align
* update chunk target
* update setting for new memory
* finish test
* update tracer
* update typo
* update test
* add unet test
* add bench
* update bench
* update bench
* init
* support vit
* move to cpu
* add cpu benchmark
2 years ago
Xuanlei Zhao
10c61de2f7
[autochunk] support vit ( #3084 )
...
support vit for autochunk
* support some new ops for vit
* fix some bugs
* add test for vit
2 years ago
Xuanlei Zhao
2ca9728cbb
[autochunk] refactor chunk memory estimation ( #2762 )
...
* refact memory code
* dont log free var memory
* add memory align
* update chunk target
* update setting for new memory
* finish test
* update tracer
* update typo
* update test
2 years ago
oahzxl
6ba8364881
[autochunk] support diffusion for autochunk ( #2621 )
...
* add alphafold benchmark
* renae alphafold test
* rename tests
* rename diffuser
* renme
* rename
* update transformer
* update benchmark
* update benchmark
* update bench memory
* update transformer benchmark
* rename
* support diffuser
* support unet metainfo prop
* fix bug and simplify code
* update linear and support some op
* optimize max region search, support conv
* update unet test
* support some op
* support groupnorm and interpolate
* update flow search
* add fix dim in node flow
* fix utils
* rename
* support diffusion
* update diffuser
* update chunk search
* optimize imports
* import
* finish autochunk
2 years ago
oahzxl
c4b15661d7
[autochunk] add benchmark for transformer and alphafold ( #2543 )
2 years ago
oahzxl
05671fcb42
[autochunk] support multi outputs chunk search ( #2538 )
...
Support multi outputs chunk search. Previously we only support single output chunk search. It is more flexible and improve performance by a large margin. For transformer, we reduce memory by 40% than previous search strategy.
1. rewrite search strategy to support multi outputs chunk search
2. fix many, many bugs
3. update tests
2 years ago
oahzxl
63199c6687
[autochunk] support transformer ( #2526 )
2 years ago
oahzxl
c04f183237
[autochunk] support parsing blocks ( #2506 )
2 years ago
oahzxl
72341e65f4
[auto-chunk] support extramsa ( #3 ) ( #2504 )
2 years ago
oahzxl
ecccc91f21
[autochunk] support autochunk on evoformer ( #2497 )
2 years ago
oahzxl
4953b4ace1
[autochunk] support evoformer tracer ( #2485 )
...
support full evoformer tracer, which is a main module of alphafold. previously we just support a simplifed version of it.
1. support some evoformer's op in fx
2. support evoformer test
3. add repos for test code
2 years ago
oahzxl
61fdd3464a
update doc
2 years ago
oahzxl
36ab2cb783
change import
2 years ago
oahzxl
7ab2db206f
adapt new fx
2 years ago
oahzxl
c1492e5013
add test in import
2 years ago
oahzxl
212b5b1b5f
add comments
2 years ago
oahzxl
aafc3516a5
add available
2 years ago
oahzxl
d5c4f0bf95
code style
2 years ago
oahzxl
d106b271f8
add chunk search test
2 years ago
oahzxl
a005965d2d
update codegen test
2 years ago
oahzxl
3abbaf8bc6
update codegen test
2 years ago
oahzxl
74b81395a2
update codegen test
2 years ago
oahzxl
18a51c87fe
rename test
2 years ago
oahzxl
cb68ee864a
set benchmark
2 years ago
oahzxl
a6cdbf9161
seperate trace flow
2 years ago
oahzxl
da4076846d
rename
2 years ago
oahzxl
fd87d78a28
rename ambiguous variable
2 years ago
oahzxl
8a634af2f5
close mem and code print
2 years ago
oahzxl
1a6d2a740b
take apart chunk code gen
2 years ago
oahzxl
d1f0773182
rename
2 years ago
oahzxl
06a5355d98
update test
2 years ago
oahzxl
efb1c64c30
restruct dir
2 years ago