Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Develop upstream sync 241105 #2744

Merged
merged 437 commits into from
Dec 2, 2024
Merged

Develop upstream sync 241105 #2744

merged 437 commits into from
Dec 2, 2024

Conversation

mmakevic-amd
Copy link

No description provided.

changm and others added 30 commits October 30, 2024 10:01
@changm @tensorflower-gardener
Internal CI/CD change
563dd9c 
PiperOrigin-RevId: 691456562
@tensorflower-gardener
Add LoadSerializedExecutable to TfrtCpuClient.
79dd13c 
PiperOrigin-RevId: 691468873
@GleasonK @tensorflower-gardener
Pin stablehlo_current_version in PJRT C API to 1.7.0.
7e4ea13 
This can be reverted to getCurrentVersion once frameworks release with the fix in:
openxla/xla@2f99455

Until then, a plugin that is newer than its framework will error on serialization, this feature was added in 1.7.X, so pinning to 1.7.0 should be safe.

PiperOrigin-RevId: 691471627
@matthiaskramm @tensorflower-gardener
tf2xla: Verify the MLIR we're about to convert to a string.
6e93dbb 
PiperOrigin-RevId: 691480719
@tensorflower-gardener
Merge pull request tensorflow#77167 from cdesouza-chromium:fix-memcpy…
0775507 
…-with-null-data-for-simple-dynamic-buffer

PiperOrigin-RevId: 691481168
@bartchr808 @tensorflower-gardener
#sdy add shardy CPU config for all JAX tests, disabling any known fai…
ddb3441 
…ling test cases.

Only test cases breaking on CPU are related to:
- pure callbacks
- export
- shard alike

Note that `layout_test` is broken on TPU, leaving a comment saying to enable it.

Also fixed `shard_map_test` test that was broken when running Shardy on one TPU, and `aot_test` which was breaking due to calling a different C++ StableHLO compilation function.

PiperOrigin-RevId: 691496997
@ddunl @tensorflower-gardener
Create implicit_cuda_headers_dependency alias in non-hermetic CUDA …
6f7945f 
…to stay consistent with hermetic CUDA

PiperOrigin-RevId: 691506371
@tensorflower-gardener
Rename APIs consistently with new LiteRT naming convention
dea7fc1 
PiperOrigin-RevId: 691516394
@ZixuanJiang @tensorflower-gardener
Cleanup. Update include and remove unused code in third_party/tensorf…
7ed33d9 
…low/compiler/xla/service/spmd/shardy/mhlo_round_trip/export_shardings.cc.

PiperOrigin-RevId: 691525161
@frgossen @tensorflower-gardener
[XLA:GPU] Unify the use of $direction in collective select folder test
7891c98 
PiperOrigin-RevId: 691528885
@changm @tensorflower-gardener
Reverts 563dd9c
5c97317 
PiperOrigin-RevId: 691539371
@GleasonK @tensorflower-gardener
Crop landing images so there is less whitespace in XLA README.
5e7abae 
PiperOrigin-RevId: 691565219
@tensorflower-gardener
1. Don't propagate errors from recoverable tasks.
c6c9c59 
2. Set tasks to error (and don't disconnect tasks) during a failed shutdown to avoid silent reconnects)

PiperOrigin-RevId: 691586482
@slackito @tensorflower-gardener
Integrate LLVM at llvm/llvm-project@4ba623f24479
d3720fe 
Updates LLVM usage to match
[4ba623f24479](llvm/llvm-project@4ba623f24479)

PiperOrigin-RevId: 691589403
@LukeBoyer @tensorflower-gardener
Delete unused serialize functions. There were replaced by byte_code_u…
03f5749 
…til.h/cc

PiperOrigin-RevId: 691591073
@sizhit2 @tensorflower-gardener
Add IFRT se gpu client and tests accordingly.
4c20ab4 
PiperOrigin-RevId: 691602092
@tensorflower-gardener
Set custom_option_alignment when exporting tflite.
ba07927 
PiperOrigin-RevId: 691648381
@tensorflower-gardener
Add version check for Compiler Plugin API
ccdf2a0 
And fix version check for Dispatch API

PiperOrigin-RevId: 691663743
@tensorflower-gardener
Automated Code Change
735f149 
PiperOrigin-RevId: 691667184
@tensorflower-gardener
Automated Code Change
6c1fb94 
PiperOrigin-RevId: 691667192
@tensorflower-gardener
Automated Code Change
c9806b2 
PiperOrigin-RevId: 691675947
@tensorflower-gardener
Automated Code Change
8af0d89 
PiperOrigin-RevId: 691681901
@tensorflower-gardener
Automated Code Change
be5dbd9 
PiperOrigin-RevId: 691692633
@beckerhe @tensorflower-gardener
Remove unused variable from base_ops_test.h
5f6e3c4 
PiperOrigin-RevId: 691718921
@tensorflower-gardener
Update GraphDef version to 2032.
33cb370 
PiperOrigin-RevId: 691720760
@tensorflower-gardener
compat: Update forward compatibility horizon to 2024-10-31
fb1f654 
PiperOrigin-RevId: 691720775
@tensorflower-gardener
Automated Code Change
b460b37 
PiperOrigin-RevId: 691739227
@hsharsha @tensorflower-gardener
PR tensorflow#18860: [ROCm] Remove IsEmpty check for execution_order …
4e44658 
…introduced in `c3d5769`

Imported from GitHub PR openxla/xla#18860

Copybara import of the project:

--
1b61efe8270e67140199bbbb70665955fbaa6656 by Harsha HS <Harsha.HavanurShamsundara@amd.com>:

[ROCm] Remove IsEmpty check for execution_order introduced in c3d5769

Merging this change closes tensorflow#18860

PiperOrigin-RevId: 691744865
@beckerhe @tensorflower-gardener
Wrap private functions in anonymous namespace in gpu_command_buffer.cc
1b88a6b 
These functions are not declared in any header file, hence they should have internal linkage.

PiperOrigin-RevId: 691746918
@akuegel @tensorflower-gardener
Reverts 9d5bb83
1492c2c 
PiperOrigin-RevId: 691750937
pifon2a and others added 14 commits November 5, 2024 02:43
@pifon2a @tensorflower-gardener
[XLA:GPU][Emitters] Port the complex.expm1 approximation used in the …
e0ccb4b 
…legacy emitters.

The default lowering in the MLIR repo is not stable for small imag(arg).

PiperOrigin-RevId: 693262812
@tensorflower-gardener
Automated Code Change
6026ad4 
PiperOrigin-RevId: 693264535
@alankelly @tensorflower-gardener
Update XNNPack doc to reflect that XNNPack can handle dynamic tensors…
f9d6ddd 
… efficently

PiperOrigin-RevId: 693265606
@loislo @tensorflower-gardener
[XLA:GPU] Fix test failures on Hopper for CUDA 12.6.2
c4bf15b 
The test failures are due to the fact that the names of kernels from CUDA are not deterministic.

PiperOrigin-RevId: 693272080
@chsigg @tensorflower-gardener
[XLA:GPU] Nest gemm fusions: hoist bitcasts up and down, but don't ba…
2798123 
…il out if it fails.

Add a large number of tests extracted from triton_fusion_emitter_device_legacy_test.

PiperOrigin-RevId: 693278160
@ddunl @tensorflower-gardener
Make filegroups in third_party/triton public
23f69eb 
PiperOrigin-RevId: 693279863
@tensorflower-gardener
Automated Code Change
d4d59e6 
PiperOrigin-RevId: 693283395
@tensorflower-gardener
Fix a dangling llvm::function_ref reference in tensorflow/compiler/ml…
09b8b49 
…ir/tensorflow/transforms/executor_island_coarsening.cc

PiperOrigin-RevId: 693288112
@akuegel @tensorflower-gardener
[XLA:GPU] Update test names to reflect what is now being tested.
2f66af0 
PiperOrigin-RevId: 693289401
@sergachev @tensorflower-gardener
PR tensorflow#18948: [GPU][NFC] Improve error messages.
6859c3b 
Imported from GitHub PR openxla/xla#18948

Copybara import of the project:

--
80e717c39e8a120cca974dca9f473d817d3a3457 by Ilia Sergachev <isergachev@nvidia.com>:

[GPU][NFC] Improve error messages.

Merging this change closes tensorflow#18948

PiperOrigin-RevId: 693291127
@tensorflower-gardener
Fix a dangling llvm::function_ref reference in third_party/tensorflow…
7b5dab1 
…/compiler/mlir/tosa/transforms/convert_tfl_uint8.cc.

PiperOrigin-RevId: 693300221
@golechwierowicz @tensorflower-gardener
[XLA:GPU] Extend AllReduceCombiner to combine pipelined collectives a…
a7e1d1f 
…s much as possible.

This is particularly useful in FSDP/HSDP where gradient propagation can be done fully in the i+1th iteration. It takes the responsibility of the user to set the `xla_gpu_all_reduce_combine_threshold_bytes` by themselves.

PiperOrigin-RevId: 693304915
@mmakevic-amd
Merge remote-tracking branch 'upstream/master' into develop-upstream-…
68c43a9 
…sync-241105
@mmakevic-amd
Fix merge conflicts
2baacd6
@mmakevic-amd
Copy link
Author

retest gpu-pycpp please

@mmakevic-amd mmakevic-amd merged commit 66ca76e into develop-upstream Dec 2, 2024
3 of 5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@hsharsha hsharsha hsharsha approved these changes

@i-chaochen i-chaochen Awaiting requested review from i-chaochen

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.