pytorch/benchmark

Pull Requests Commits

[not4land] Local torchao benchmark

jerryzh168 committed 327 days ago

4b8e7a94

add debug logging for graph region expansion (#141382)

mlazos committed 1 year ago

675fb8f5

Fix bug involving missing compilation_metrics fields in tlparse logs (#142423)

masnesral committed 1 year ago

2ffd75a4

Add triton_compile_time_us column to dynamo_compile (#142068)

masnesral committed 1 year ago

34038d56

Add convnext_base to higher tolerance (#142159)

ezyang committed 1 year ago

692bf64e

Fix pynvml (#2553)

FindHao committed 1 year ago

364420ae

Fix pyproject (#2554)

FindHao committed 1 year ago

7340ede9

Add `--skip` arg for `test_bench` (#2551)

shink committed 1 year ago

40f68c54

Allow to continue on error when installing `test_bench` (#2550)

shink committed 1 year ago

38dd1bb0

Convert directory fbcode/pytorch to use the Ruff Formatter

Thomas Polasek committed 1 year ago

e5986f88

Add `RANGE_ITERATOR_MATCH` to properly guard on range iterators (#141902)

StrongerXi committed 1 year ago

3d686915

Update mypy to 1.13.0 (#140808)

Skylion007 committed 1 year ago

63904899

dynamo_timed: Add a log_waitcounter option. (#141402)

c00w committed 1 year ago

505e74ec

Set remote cache version and backend type once in compilation metrics

jamesjwu committed 1 year ago

a86ec7f1

Revert D66508492: Set remote cache version and backend type once in compilation metrics

malfet committed 1 year ago

711de8cf

Set remote cache version and backend type once in compilation metrics

jamesjwu committed 1 year ago

c07f33b5

fix generic namedtuple support when the class is

XuehaiPan committed 1 year ago

82a8a602

Enable autograd cache on inductor tests (#140890)

jamesjwu committed 1 year ago

9f8813e2

Fix `VariableBuilder._wrap` on frozenset and enforce invariants on `ConstantVariable` (#141504)

StrongerXi committed 1 year ago

72176267

expose option to collect sizes as dynamic (#141153)

xmfan committed 1 year ago

5c7b91e2

Move population of common MetricsContext fields to record_compilation_metrics (#141291)

masnesral committed 1 year ago

c57be069

Add install.py for test_bench (#2546)

shink committed 1 year ago

9ff23aba

Migrate to AWS A100 runner (#2549)

xuzhao9 committed 1 year ago

341ad14f

Add more devices to `torchbenchmark.util.experiment.instantiator.list_devices()` (#2545)

shink committed 1 year ago

820f2136

pytorch/feature: Record if inductor fx cache is enabled (#141059)

c00w committed 1 year ago

0fd86c06

Added cuda and triton versions to dynamo_compile (#141290)

jovianjaison committed 1 year ago

03e12ceb

pytorch/features: Make a feature logger and record triton bundling (#141056)

c00w committed 1 year ago

930f14b5

Add inductor_fx_graph_cache stats to dynamo_utils

ppanchalia committed 1 year ago

6c438b7c

Apply PERF401 autofixes from ruff (#140980)

Skylion007 committed 1 year ago

1de25b59

Skip uploading benchmark records when there is no model name (#141145)

huydhn committed 1 year ago

598f5f9c

Older