pytorch/benchmark

Pull Requests Commits

Replace runners prefix amz2023.

jeanschmidt committed 2 years ago

dc0d5e63

Set dump_exported_program to True in bench (#131670)

justinchuby committed 2 years ago

06cf234e

Updating Types in torch/_dynamo/utils.py (#131001)

adriaorenstein committed 2 years ago

fe362296

Run only a subset for optim benchmarks, attempt to fix (#2392)

janeyx99 committed 2 years ago

9ad6639a

Fix dimension order for FlashAttention-3 (#2391)

bertmaher committed 2 years ago

ff65cb67

Read from dataclass fields

xuzhao9 committed 2 years ago

6af4bd93

Triton Partition K gemm to TritonBench

sijiac committed 2 years ago

53d98e37

Always obtain `model_flops` from the eager mode (#2390)

xuzhao9 committed 2 years ago

43d8a999

Run accuracy check at Diff time (#2388)

xuzhao9 committed 2 years ago

0032c0b4

Align xpu models batch size with A100 (#2378)

chuanqi129 committed 2 years ago

03cde49e

Add packaging to docker (#2389)

xuzhao9 committed 2 years ago

11cf319f

Add fp8 flash attention (#2382)

manman-ren committed 2 years ago

1a1f5411

Add oss support to ncu rep (#2387)

xuzhao9 committed 2 years ago

5efd1dc0

Add guard override capability for tensor subclass metadata (#130780)

mlazos committed 2 years ago

366d6a14

Add all operators to H100 RE unit tests

xuzhao9 committed 2 years ago

e8e5db7d

Add variable-length loop Triton kernel benchmark for jagged_softmax

jananisriram committed 2 years ago

11ebabb2

Add simple fused Triton kernel benchmark for jagged_softmax

jananisriram committed 2 years ago

c75d1a51

Add torch.sum benchmark for jagged_softmax operator

jananisriram committed 2 years ago

f50c632f

Add kernels submodule (#2380)

xuzhao9 committed 2 years ago

594a051c

Add flash attention v3 (#2381)

xuzhao9 committed 2 years ago

6dfede78

enforce style for empty lines in import segments in `benchmarks/` (#129754)

XuehaiPan committed 2 years ago

429137dd

typing: convert_frame (#130670)

aorenste committed 2 years ago

97770e96

Run performance test non-alternately (#130136)

xuzhao9 committed 2 years ago

736bd91d

Mark nn_module params and buffers as static in dynamo (#130391)

mlazos committed 2 years ago

fbfbe989

dynamo add support for `defaultdict(set)` (#130745)

alexcdennis committed 2 years ago

40ee53b1

Add TMA+persistent bf16 gemm (#2373)

bertmaher committed 2 years ago

0ec9f60e

Add persistent+TMA matmul to fp8 gemm benchmark (#2377)

bertmaher committed 2 years ago

4f245a87

Fix servicelab test failures

xuzhao9 committed 2 years ago

8a3bf01a

Add torch.sum benchmark for jagged_layernorm operator (#2376)

jananisriram committed 2 years ago

e26cd75d

Update FBGEMM commit (#2375)

xuzhao9 committed 2 years ago

c70ccb37

Older