pytorch/benchmark

Pull Requests Commits

fix xval for rmn_norm

FindHao committed 2 years ago

1d45b393

fix bug for oom

FindHao committed 2 years ago

ba2d8a35

fix bug for rms_norm x_val

FindHao committed 2 years ago

5d1b1015

FindHao committed 2 years ago

f48e1f91

update x_val to input shapes

FindHao committed 2 years ago

72e54512

update mem_footprint to make it consistent with torchinductor dashboard

FindHao committed 2 years ago

5efdd6eb

add mem_foot_print

FindHao committed 2 years ago

ba80e9e6

FindHao committed 2 years ago

0abec03c

merge all liger kernels

FindHao committed 2 years ago

b871dbce

Add logger logging for remote fx graph cache get + put (#2512)

masnesral committed 2 years ago

2614ca98

fixing gemm for amd

Adam Mainz committed 2 years ago

65e5f686

Account for older numpy versions in #2514 (#2524)

mikaylagawarecki committed 2 years ago

4f30c497

Release 2.5.1.yaml perf test (#2525)

atalman committed 2 years ago

47ba1ed8

updating hardware and device columns

Adam Mainz committed 2 years ago

3245fde9

Support overridden __call__ on nn modules (#138619)

anijain2305 committed 2 years ago

b5b342ba

Support range_iterator as a function input (#138657)

anijain2305 committed 2 years ago

367b6ef3

adding fp32 strict and tf32x3 benchmarks for gemm

Adam Mainz committed 2 years ago

036012ff

tls access helpers (#138061)

xmfan committed 2 years ago

405ba75b

Compiled autograd configs in TLS (#137821)

xmfan committed 2 years ago

0e038319

Log is_forward field to dynamo_compile scuba table (#138505)

masnesral committed 2 years ago

e57bbe23

Add metadata to events in progress, new `dynamo` event

jamesjwu committed 2 years ago

8fce9c12

typing compile_fx.py (#138033)

aorenste committed 2 years ago

11543183

Fixes to prep for weights_only default flip (#2514)

mikaylagawarecki committed 2 years ago

fb590d99

Update requirements.txt (#2523)

wdvr committed 2 years ago

a45e0dbf

Replace __str__ with __repr__ in some places (#136316)

rec committed 2 years ago

173774d1

fixing key error in aggregate data

Adam Mainz committed 2 years ago

a21b30e4

Disable torch function compilation during guard execution and in compiled bytecode (#137669)

mlazos committed 2 years ago

05620407

replace uses of np.ndarray with npt.NDArray

igorsugak committed 2 years ago

06e35fc5

specifying logged benchmark name for tritonBench servicelab logging

Adam Mainz committed 2 years ago

e737b8fe

adding aggregates to servicelab

Adam Mainz committed 2 years ago

0a9cd8f0

Older