pytorch/xla

Pull Requests Commits

Liyang90 committed 1 year ago

Verified 84499f57

Truncate python stack when outputting frame that cause the graph executation (#5933)

JackCaoG committed 1 year ago

Verified a5e07389

Don't fallback for pow (#5919)

qihqi committed 1 year ago

Verified 402166be

Lower quant/dequant torch op to StableHLO (#5763)

lsy323 committed 1 year ago

Verified a3b0c6ec

Fix `as_strided` for inputs smaller than the arguments specification. (#5914)

ysiraichi committed 1 year ago

Verified 4ac255ae

Collect CUDA/CPU profiling info into result sheets. (#5921)

golechwierowicz committed 1 year ago

Verified c03afb14

Fix get_gpu_name returning empty name. (#5925)

golechwierowicz committed 1 year ago

Verified 16a40cc6

Revert "Configure PATH and LD_LIBRARY_PATH to include cuda (#5761)" (#5928)

vanbasten23 committed 1 year ago

Verified cfb36175

Copy input tensors before async transfer (#5830)

will-cromar committed 1 year ago

Verified b69fb93d

Refactor Dynamo (custom op) integration code (#5805)

yeounoh committed 1 year ago

Verified 3385bd64

update result analyzer script (#5920)

zpcore committed 1 year ago

Verified 037bbd66

Fix tier import failure for relative import (#5918)

frgossen committed 1 year ago

Verified fe7b0075

Merge benchmark scripts into the main branch (#5913)

frgossen committed 1 year ago

Verified 94ca4597

Create FIX_LOWERING_FOR_CORE_ATEN_OPS.md (#5909)

wonjoo-wj committed 1 year ago

Verified 3fa035e3

Re-implement `ComputationClient::WaitDeviceOps` (#5811)

will-cromar committed 1 year ago

Verified 175e86b8

Add python 3.9 and 3.11 build for nightly (#5840)

JackCaoG committed 1 year ago

Verified 57e60352

Add a unittest testing each core ATen ops. (#5796)

qihqi committed 1 year ago

Verified 2b04794a

Sync `xla_args` before computation. (#5823)

ysiraichi committed 1 year ago

Verified f80e3ba9

Use cuda 12.1 and build pytorch with cuda enabled. (#5831)

vanbasten23 committed 1 year ago

Verified 8c03ac96

Configure PATH and LD_LIBRARY_PATH to include cuda (#5761)

vanbasten23 committed 1 year ago

Verified 4fb54281

Extend HLO metadata to include class hierarchy information (#5715)

mrnikwaws committed 1 year ago

Verified 5e503794

[SPMD] fix bug with XLAShardedTensor.__repr__ (#5807)

yeounoh committed 1 year ago

Verified 6addbde3

Use upstream XLA concurrency utilities (#5799)

will-cromar committed 1 year ago

Verified 6e661304

Add python 3.11 trigger for 2.1 release build (#5810)

JackCaoG committed 1 year ago

Verified a3bbbb1e

Enable passing down dynamic dimensions from torch to XLA (#5790)

lsy323 committed 1 year ago

Verified a0a048e1

Don't set $TPU_LIBRARY_PATH during import

will-cromar committed 1 year ago

Verified d8c6a83f

Record the lazy tracing time(C++) in metrics (#5757)

JackCaoG committed 1 year ago

Verified 58ce7e17

Use TPU profiler plugin (#5793)

will-cromar committed 1 year ago

Verified 3b34cb27

Add GKE support and various usability improvements in CheckpointManager (#5770)

jonb377 committed 1 year ago

Verified 5710a83c

Lower AtenFull op (#5781)

danielvegamyhre committed 1 year ago

Verified b3ed82c9

Older