pytorch/xla

Pull Requests Commits

Add gpu doc to documentation page

JackCaoG committed 2 years ago

a3612b1c

Fix bug in ShardTensor to generate correct shards for each device (#4549)

jonb377 committed 2 years ago

Verified a6a77ef8

Update env-setup.py to drop "m" in wheel name (#4563)

wonjoo-wj committed 2 years ago

Verified 933dcc21

Remove unused docker tagging in .circleci/config.yml (#4561)

yeounoh committed 2 years ago

Verified 7b3a33b0

Modify tpu version regex to match new names (#4557)

steventk-g committed 2 years ago

Verified 96fd3ce2

Implement missing XLASymNodeImpl::Sub (#4551)

vanbasten23 committed 2 years ago

Verified 0830b585

Add initial support for `scatter_reduce` (#4517)

seanlatias committed 2 years ago

Verified 99bdfe8d

make empty_symint support dynamism. (#4550)

vanbasten23 committed 2 years ago

Verified 34e91153

Reenable test_view_copy_out after https://github.com/pytorch/pytorch/pull/93411 (#4547)

yeounoh committed 2 years ago

Verified 4f05e6bb

Add SPMD benchmarking test (#4141)

jonb377 committed 2 years ago

Verified f5232da4

Reset Nvidia cuda-keyring in the CI env setup (#4548)

yeounoh committed 2 years ago

Verified 889a1039

Lower _linalg_svd (#4537)

wonjoo-wj committed 2 years ago

Verified b205042b

Transfer shards from server using replication (#4508)

yeounoh committed 2 years ago

Verified 780a8231

Align the dynamo dump file with the lazy (#4530)

JackCaoG committed 2 years ago

Verified ffd73b0d

Fix race condition between execution and buffer tear down on GPU when using bfc_allocator (#4542)

ymwangg committed 2 years ago

Verified 33e87a2f

Disable test_view_copy_out_xla (#4540)

yeounoh committed 2 years ago

Verified 9cbcdb40

Remove Python 3.7 build trigger from Terraform (#4534)

will-cromar committed 2 years ago

Verified ce48e42a

Optimize Dynamo Training Execute (#4523)

JackCaoG committed 2 years ago

Verified f5494cd7

Basic test for `PjrtComputationClient`. (#4478)

stgpetrovic committed 2 years ago

Verified c80cf439

[PJRT] Experimental support for `torch.distributed` and DDP on TPU v2/v3 (#4520)

will-cromar committed 2 years ago

Verified 021a1cc2

Enable python 3.8 and resolve dependency issues (#4529)

yeounoh committed 2 years ago

Verified bfd20f37

Try bootstrapping tpu env from env vars (#4499)

steventk-g committed 2 years ago

Verified f2cb76f9

Move XLA testing to Python-3.8 (#4527)

malfet committed 2 years ago

Verified 4f4ab8a3

Increase the timeout to make cuda docker build more stable (#4522)

JackCaoG committed 2 years ago

Verified b6b4da0d

[PJRT] Release the GIL during `TransferFromServer` (#4504)

will-cromar committed 2 years ago

Verified d41c5ba2

Always use `ExecuteSharded` future with PJRT C API. (#4521)

will-cromar committed 2 years ago

Verified efb53533

Disable the Sandbox build in Docker (#4515)

JackCaoG committed 2 years ago

Verified 661ab6d2

Bump up the xla_base image ver to v0.8 in upstream repo (#4516)

yeounoh committed 2 years ago

Verified 6acc0c9c

Switch to pytorch sccache (#4489)

huydhn committed 2 years ago

Verified 6bf36302

Remove deprecated torch.symeig (#4498)

IvanYashchuk committed 2 years ago

Verified 903f2ac5

Older