pytorch/xla

Pull Requests Commits

remove debug messages and fix tests

JackCaoG committed 2 years ago

866b2bc9

JackCaoG committed 2 years ago

fda1c411

remove debugging messages

JackCaoG committed 2 years ago

93af2910

JackCaoG committed 2 years ago

d12446bd

Add test for non-dynamo buffer donation

JackCaoG committed 2 years ago

0b60229d

Fix a bug where warm up cache might accidentlly execute the graph

JackCaoG committed 2 years ago

e1c3a447

only enable buffer donor aliasing in dynamo

JackCaoG committed 2 years ago

3561bdf6

make sure compilation hash tracks buffer donor index

JackCaoG committed 2 years ago

0b6c3a81

add test to testing script

JackCaoG committed 2 years ago

5671ee57

add python tests, currently they will fail if being run together due to hash conflict

JackCaoG committed 2 years ago

92480b49

Add get_buffer_donation, fix a bug where mark+step with devicedata will overwrite the buffer donation

JackCaoG committed 2 years ago

ccbdc532

add SetBufferDonors

JackCaoG committed 2 years ago

ebbafda2

Add api to buffer donation

JackCaoG committed 2 years ago

507ebc1f

Make sure torch_xla2 works with torch stable 2.2 (#6593)

qihqi committed 2 years ago

Verified a659d970

[benchmarks] Fix YAML file name. (#6597)

ysiraichi committed 2 years ago

Verified 1a2ea9de

[Core Aten Ops] Lower reflection_pad1d, reflection_pad1d_backward, reflection_pad3d and reflection_pad3d_backward (#6588)

ManfeiBai committed 2 years ago

Verified 13e8647c

Allow Inductor benchmarks for CPU backends. (#6555)

changm committed 2 years ago

Verified 263da4ee

Update OpenXLA pin to fix GPU build (#6569)

cota committed 2 years ago

Verified 2bbc9a49

[Fori_loop|While_loop] Placeholder lower torch.while_loop with python dispatch for simple addition test case (#6532)

ManfeiBai committed 2 years ago

Verified 5405c4d4

Enable graph input matcher test (#6591)

JackCaoG committed 2 years ago

Verified 69d453ea

Add extra APIs. (#6586)

qihqi committed 2 years ago

Verified 5145ea1c

Add XLA specific XPlane profiling. (#6552)

golechwierowicz committed 2 years ago

Verified adabb59a

Fix some more core aten ops (#6580)

wonjoo-wj committed 2 years ago

Verified 2cb31440

fix linter (#6585)

JackCaoG committed 2 years ago

Verified c524b4dc

Upgrade C++ formater to clang-format-11 (#6583)

JackCaoG committed 2 years ago

Verified 0c3fbc66

Register OpOverload when OpOverloadPacket is registered (#6579)

lsy323 committed 2 years ago

Verified d8ed46d5

Add basic device APIs to the top-level `torch_xla` module. (#6571)

will-cromar committed 2 years ago

Verified 0ec5b917

Fix typo in TROUBLESHOTTING (#6468)

JackCaoG committed 2 years ago

Verified e77a629c

Fix a incorrect assert on frame count for PT_XLA_DEBUG=1 (#6466)

JackCaoG committed 2 years ago

Verified c08ae21a

lower replication_pad3d and replication_pad3d_backward (#6566)

ManfeiBai committed 2 years ago

Verified 0192ff75

Older