pytorch/xla

Pull Requests Commits

alanwaketan committed 2 years ago

22279011

Support scalar tensors

alanwaketan committed 2 years ago

d5b9662a

[SPMD] Add a test case for mark_shard scalar tensors

alanwaketan committed 2 years ago

8e935bcf

[SPMD] Add apply_backward_optimization_barrier (#6097)

alanwaketan committed 2 years ago

Verified 07540f21

Fix bug for low-precision ranges (#5636)

jonb377 committed 2 years ago

Verified eb3c446f

Add Pattern Boundary Marking API (#5930)

lsy323 committed 2 years ago

Verified 8243b502

Add out-of-place all-gather coalesced (#6059)

jeffhataws committed 2 years ago

Verified e3012bf9

Revert prs to get heads green again (#6107)

JackCaoG committed 2 years ago

Verified 6ae66f78

Update benchmark readme (#6100)

frgossen committed 2 years ago

Verified 694047eb

Add out-of-place reduce-scatter coalescing (#6058)

jeffhataws committed 2 years ago

Verified ba7c3478

Add cuda bm dry run (#6093)

frgossen committed 2 years ago

Verified aa959af6

Add benchmark tests (#6082)

frgossen committed 2 years ago

Verified bbe1a8de

Add benchmark experiment test (#6081)

frgossen committed 2 years ago

Verified dbd3e33c

Allow scalar all gather (#5797)

bfontain committed 2 years ago

Verified b2243507

Increase delay in trace_detached test (#6075)

jonb377 committed 2 years ago

Verified 9be1e94c

Update Troubleshotting doc with new PT_XLA_DEBUG (#6039)

JackCaoG committed 2 years ago

Verified 45d4750b

Gather more summary data when running `result_analyzer.py`. (#6067)

ysiraichi committed 2 years ago

Verified 64239127

Move 0-dimensional tensors to CPU before copying to XLA. (#6071)

ysiraichi committed 2 years ago

Verified a80c1e7f

Reduce benchmark experiment filename string. (#6069)

ysiraichi committed 2 years ago

Verified 0857f2a0

Use `subprocess` for checking XLA supported devices. (#6068)

ysiraichi committed 2 years ago

Verified faae458d

Piz/release (#6049)

zpcore committed 2 years ago

Verified a923bfc6

ci: add benchmarks/run_tests.sh (#6043)

cota committed 2 years ago

Verified 0313c6ae

[OpenXLA-pin] Update OpenXLA-pin to Nov29 (#6053)

ManfeiBai committed 2 years ago

Verified 599b79ff

Move CUDA tensors to CPU before moving to XLA. (#6060)

ysiraichi committed 2 years ago

Verified d5df845b

Add TPU CI infrastructure (#5817)

mbzomowski committed 2 years ago

Verified a7d58f72

Remove redundant log message (#6056)

mrnikwaws committed 2 years ago

Verified a16e0114

benchmarks/aggregate: round floats before printing them (#6062)

cota committed 2 years ago

Verified b0e7f1b3

benchmarks/aggregate: output datetime in UTC (#6051)

cota committed 2 years ago

Verified a746f556

Not creating a coordinator service for the single processing job. (#6023)

vanbasten23 committed 2 years ago

Verified 467c18f0

[dynamo] Move CPU tensor constructor nodes to XLA. (#5922)

ysiraichi committed 2 years ago

Verified 9cccbe10

Older