OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 37 Packages Projects Releases Wiki Activity
76,532 Commits 8,304 Branches 150 Tags 1.5 GiB
5ea0f51187
Commit Graph

171 Commits

Author SHA1 Message Date
Yanbo Liang
5ea0f51187 [Dynamo] Support abc.MutableMapping.get (#132363) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/132363
Approved by: https://github.com/anijain2305, https://github.com/mlazos
 2024-08-02 04:17:35 +00:00
Oguz Ulgen
71e22e0959 Add basic mypy annotations to dynamo (#132415) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/132415
Approved by: https://github.com/XuehaiPan, https://github.com/jamesjwu
 2024-08-01 20:14:25 +00:00
Xuehai Pan
e74ba1b34a [BE][Easy][15/19] enforce style for empty lines in import segments in torch/_d*/ (#129767) 
See https://github.com/pytorch/pytorch/pull/129751#issue-2380881501. Most changes are auto-generated by linter.

You can review these PRs via:

```bash
git diff --ignore-all-space --ignore-blank-lines HEAD~1
```

Pull Request resolved: https://github.com/pytorch/pytorch/pull/129767
Approved by: https://github.com/anijain2305
 2024-07-31 21:18:11 +00:00
Luca Wehrstedt
f4f7aba75d Expose function to probe whether PyTorch was built with FlashAttention (#131894) 
This is needed by downstream projects (e.g., xFormers) to determine whether they can count on FlashAttention in PyTorch or whether they need to build it themselves.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/131894
Approved by: https://github.com/drisspg, https://github.com/eqy
 2024-07-31 11:33:09 +00:00
ekamiti
9e473fd868 Make adding Buffers more like adding Parameters (#125971) 
Add similar semantics for creating a buffer object similar to creating a parameter. This is done by introducing a new Buffer class that can be used for type disambiguation. The underlying functionality of registering a buffer remains the same as the register_buffer method has not been changed. The persistent parameter in the Buffer type is to indicate whether a buffer object should be persistent or not. Other non-test changes have to do with getting the new Buffer type recognized by inductor and dynamo. Remaining changes are test changes to make sure that the Buffer type can be used as a drop in replacement for register_buffer as it just leads to register_buffer being called. The addition of this new functionality still allows for normal tensors to be used as buffers so these changes are intended to be backwards compatible.

Fixes #35735

Co-authored-by: Mikayla Gawarecki <mikaylagawarecki@gmail.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125971
Approved by: https://github.com/albanD, https://github.com/anijain2305, https://github.com/mlazos
 2024-07-31 10:32:40 +00:00
Shuai Yang
58f76bc301 Revise skip torchrec logic (#130783) 
Summary:
The previous logic adds skipped files when the file was imported which happens at very early stage. However, we could set skip_torchrec at later stage (e.g, in APS, we set it during the trainer execution). In that case, the skip logic will still take effect since skipped files have been added.

So in this diff, we revise the logic so that it can adapt to changes of skip_torchrec at later stages.

Test Plan:
Tested on APS models:

  buck2 run mode/opt //aps_models/ads/icvr:icvr_launcher_live -- mode=local_ig_fm_uhm_mini model_name=ig_fm_one_sparse_benchmark features=ig_fm_one_sparse_benchmark model=ig_fm_one_sparse_benchmark training.pipeline_type=pt2

commit: 2fb485d9e

torchrec related paths were not skipped.

Differential Revision: D59779153

Pull Request resolved: https://github.com/pytorch/pytorch/pull/130783
Approved by: https://github.com/yanboliang
 2024-07-30 19:55:20 +00:00
Guilherme Leobas
a843178529 Let dynamo inline functional_call (#128646) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128646
Approved by: https://github.com/zou3519
 2024-07-30 14:22:23 +00:00
Animesh Jain
f806128619 [dynamo] Skip <frozen abc> to skip __isisintance__ check on abc objects (#131956) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/131956
Approved by: https://github.com/williamwen42, https://github.com/mlazos
ghstack dependencies: #131827
 2024-07-30 05:49:58 +00:00
PyTorch MergeBot
f72266ecea Revert "Let dynamo inline functional_call (#128646)" 
This reverts commit 5aab1acc84.

Reverted https://github.com/pytorch/pytorch/pull/128646 on behalf of https://github.com/clee2000 due to the newly added test dynamo/test_higher_order_ops.py::FuncTorchHigherOrderOpTests::test_functional_call_sequential_params_and_buffers [GH job link](https://github.com/pytorch/pytorch/actions/runs/10147452270/job/28058682000) [HUD commit link](5aab1acc84) is broken, probably a landrace since it passed on PR ([comment](https://github.com/pytorch/pytorch/pull/128646#issuecomment-2256375501))
 2024-07-29 16:26:50 +00:00
Guilherme Leobas
5aab1acc84 Let dynamo inline functional_call (#128646) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128646
Approved by: https://github.com/zou3519
ghstack dependencies: #129091, #130490
 2024-07-29 15:41:03 +00:00
Aaron Gokaslan
d1c4e6b55f [BE]: Enable a few additional ruff rules (#130700) 
Enables a few extra ruff rules, most of which do not have any violations as I already cleaned them with earlier PRs, these just turns them on to enforce them. Adds 1 noqa as we want the suboptimal lambda generation + call kept as a test. Also enables the test in flake8

Pull Request resolved: https://github.com/pytorch/pytorch/pull/130700
Approved by: https://github.com/justinchuby, https://github.com/ezyang
 2024-07-17 02:06:04 +00:00
Xuehai Pan
973037be6a [BE][Easy] apply autofix for ruff rules unnecessary-collection-call (C408): list() / tuple() / dict() (#130199) 
This PR changes the empty collection factory call to Python literals:

- `list()` -> `[]`
- `tuple()` -> `()`
- `dict()` -> `{}`

The Python literals are more performant and safer. For example, the bytecode for building an empty dictionary:

```bash
$ python3 -m dis - <<EOS
import collections

d1 = {}
d2 = dict()

dict = collections.OrderedDict
d3 = dict()
EOS
```

```text
  0           0 RESUME                   0

  1           2 LOAD_CONST               0 (0)
              4 LOAD_CONST               1 (None)
              6 IMPORT_NAME              0 (collections)
              8 STORE_NAME               0 (collections)

  3          10 BUILD_MAP                0
             12 STORE_NAME               1 (d1)

  4          14 PUSH_NULL
             16 LOAD_NAME                2 (dict)
             18 CALL                     0
             26 STORE_NAME               3 (d2)

  6          28 LOAD_NAME                0 (collections)
             30 LOAD_ATTR                8 (OrderedDict)
             50 STORE_NAME               2 (dict)

  7          52 PUSH_NULL
             54 LOAD_NAME                2 (dict)
             56 CALL                     0
             64 STORE_NAME               5 (d3)
             66 RETURN_CONST             1 (None)
```

The dict literal `{}` only has one bytecode `BUILD_MAP`, while the factory call `dict()` has three `PUSH_NULL + LOAD_NAME + CALL`. Also, the factory call is not safe if users override the `dict` name in `locals` or `globals` (see the example of replacing with `OrderedDict` above).

Pull Request resolved: https://github.com/pytorch/pytorch/pull/130199
Approved by: https://github.com/malfet
 2024-07-11 17:30:28 +00:00
Yanbo Liang
111f9b5d44 [Dynamo] Add config to skip/inline torchrec (#129912) 
Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/129912
Approved by: https://github.com/anijain2305
 2024-07-03 00:14:51 +00:00
eqy
f845a7a91a [cuDNN][SDPA] Remove TORCH_CUDNN_SDPA_ENABLED=1, enable cuDNN SDPA by default on H100 and 2nd on other archs >= sm80 (#125343) 
Looks like one of the first failures seen is `test_causal_variants_compile_causal_variant_CausalVariant_LOWER_RIGHT_shape0_cuda` when `test_causal_variants_causal_variant_CausalVariant_LOWER_RIGHT_shape0_cuda` passes.

What seems interesting here is that the `torch.compile` version fails while the eager version passes. Not sure what the difference would be here...

Nevertheless, is there a recommended mechanism to skip cuDNN SDPA as a backend for this test? CC @drisspg

Pull Request resolved: https://github.com/pytorch/pytorch/pull/125343
Approved by: https://github.com/Skylion007
 2024-06-30 19:22:16 +00:00
PyTorch MergeBot
999eec8dea Revert "[cuDNN][SDPA] Remove TORCH_CUDNN_SDPA_ENABLED=1, enable cuDNN SDPA by default on H100 and 2nd on other archs >= sm80 (#125343)" 
This reverts commit b7e7a4cb01.

Reverted https://github.com/pytorch/pytorch/pull/125343 on behalf of https://github.com/huydhn due to Sorry for reverting your change but it seems to break some test_transformer running on internal A100 and V100 ([comment](https://github.com/pytorch/pytorch/pull/125343#issuecomment-2196202003))
 2024-06-28 06:03:54 +00:00
Eddie Yan
b7e7a4cb01 [cuDNN][SDPA] Remove TORCH_CUDNN_SDPA_ENABLED=1, enable cuDNN SDPA by default on H100 and 2nd on other archs >= sm80 (#125343) 
Looks like one of the first failures seen is `test_causal_variants_compile_causal_variant_CausalVariant_LOWER_RIGHT_shape0_cuda` when `test_causal_variants_causal_variant_CausalVariant_LOWER_RIGHT_shape0_cuda` passes.

What seems interesting here is that the `torch.compile` version fails while the eager version passes. Not sure what the difference would be here...

Nevertheless, is there a recommended mechanism to skip cuDNN SDPA as a backend for this test? CC @drisspg

Pull Request resolved: https://github.com/pytorch/pytorch/pull/125343
Approved by: https://github.com/Skylion007
 2024-06-26 00:49:18 +00:00
Isuru Fernando
e6bfa2958b Add aten._unsafe_masked_index (#116491) 
To generate masked indexing operations that would generate
masked loads in triton code

Pull Request resolved: https://github.com/pytorch/pytorch/pull/116491
Approved by: https://github.com/lezcano, https://github.com/peterbell10
 2024-06-25 02:45:02 +00:00
Animesh Jain
6b5fbc544e [dynamo] Use polyfill to trace through the attributes of torch.jit.* and lru_cache_wrapper (#128336) 
Earlier we were taking the vt for `obj` and then monkeypatching that `vt.source` to be `obj._torchdynamo_inline`. If one accesses `obj.attr_a`, this would cause problems because Dynamo would then search it in `obj._torchdynamo_inline.attr_a`. This PR makes it more functional, so that we have different vts for obj and `ob._torchdynamo_inline`.

Fixes https://github.com/pytorch/pytorch/issues/93698

Pull Request resolved: https://github.com/pytorch/pytorch/pull/128336
Approved by: https://github.com/jansel, https://github.com/yanboliang
ghstack dependencies: #129117
 2024-06-21 07:44:44 +00:00
Jiong Gong
914d3ca2ba [inductor][cpp] BF16 AMX micro-gemm support (#127195) 
This PR adds the intrinsics based micro-gemm for BF16 using Advanced Matrix eXtension (AMX) instructions available in Intel 4th and 5th Xeon processors. A compilation check is added to `codecache.py` to check the validity of the compiler support. Also, since AMX requires an initialization in the Linux kernel to extra register states, an initialization function is added to do that and triggered via `codecache.py`.

Performance speedups with >=10% on BF16 AMP, max_autotune vs. no autotune, measured on Intel(R) Xeon(R) Platinum 8488C:
Static shapes
Single-threaded
| Model Family | Model Name | Speedup |
|--------------|------------|---------|
| timm_models | mixer_b16_224 | 1.54 |
| timm_models | convit_base | 1.53 |
| huggingface | MobileBertForQuestionAnswering | 1.52 |
| torchbench | fastNLP_Bert | 1.44 |
| torchbench | llama | 1.33 |
| timm_models | swin_base_patch4_window7_224 | 1.31 |
| torchbench | dlrm | 1.28 |
| torchbench | timm_vision_transformer_large | 1.28 |
| huggingface | MobileBertForMaskedLM | 1.27 |
| timm_models | vit_base_patch16_224 | 1.26 |
| timm_models | beit_base_patch16_224 | 1.23 |
| timm_models | jx_nest_base | 1.21 |
| torchbench | pyhpc_equation_of_state | 1.18 |
| huggingface | Speech2Text2ForCausalLM | 1.15 |
| timm_models | pit_b_224 | 1.14 |
| timm_models | twins_pcpvt_base | 1.14 |
| torchbench | maml_omniglot | 1.1 |
| timm_models | botnet26t_256 | 1.1 |

Multi-threaded
| Model Family | Model Name | Speedup |
|--------------|------------|---------|
| torchbench | BERT_pytorch | 1.35 |
| torchbench | lennard_jones | 2.43 |
| torchbench | hf_Albert | 1.35 |
| torchbench | hf_T5 | 1.34 |
| torchbench | soft_actor_critic | 1.34 |
| torchbench | fastNLP_Bert | 1.28 |
| huggingface | LayoutLMForSequenceClassification | 1.26 |
| torchbench | llama | 1.24 |
| huggingface | GPT2ForSequenceClassification | 1.19 |
| torchbench | hf_Bart | 1.17 |
| torchbench | hf_Bert_large | 1.16 |
| torchbench | hf_GPT2 | 1.16 |
| timm_models | gmixer_24_224 | 1.16 |
| torchbench | hf_GPT2_large | 1.15 |
| torchbench | maml_omniglot | 1.14 |
| torchbench | hf_Bert | 1.13 |
| torchbench | hf_DistilBert | 1.13 |
| torchbench | hf_T5_large | 1.12 |
| huggingface | MT5ForConditionalGeneration | 1.11 |

Dynamic shapes
Single-threaded
| Model Family | Model Name | Speedup |
|--------------|------------|-------|
| timm_models | mixer_b16_224 | 1.52 |
| timm_models | convit_base | 1.5 |
| huggingface | MobileBertForQuestionAnswering | 1.49 |
| torchbench | fastNLP_Bert | 1.42 |
| torchbench | timm_vision_transformer_large | 1.28 |
| timm_models | swin_base_patch4_window7_224 | 1.27 |
| torchbench | llama | 1.26 |
| huggingface | MobileBertForMaskedLM | 1.25 |
| timm_models | vit_base_patch16_224 | 1.25 |
| timm_models | beit_base_patch16_224 | 1.24 |
| timm_models | jx_nest_base | 1.2 |
| torchbench | dlrm | 1.19 |
| timm_models | pit_b_224 | 1.13 |
| timm_models | twins_pcpvt_base | 1.13 |
| torchbench | hf_Bert_large | 1.12 |
| torchbench | hf_BigBird | 1.11 |
| huggingface | Speech2Text2ForCausalLM | 1.11 |
| timm_models | eca_botnext26ts_256 | 1.11 |
| timm_models | botnet26t_256 | 1.1 |

Multi-threaded
| Model Family | Model Name | Speedup |
|--------------|------------|-------|
| torchbench | BERT_pytorch | 1.18 |
| torchbench | lennard_jones | 2.18 |
| torchbench | hf_Albert | 1.37 |
| torchbench | soft_actor_critic | 1.31 |
| huggingface | GPT2ForSequenceClassification | 1.29 |
| torchbench | hf_T5 | 1.28 |
| torchbench | fastNLP_Bert | 1.27 |
| torchbench | hf_Bart | 1.21 |
| torchbench | hf_Bert_large | 1.19 |
| torchbench | hf_T5_large | 1.19 |
| torchbench | hf_Bert | 1.16 |
| torchbench | hf_GPT2 | 1.16 |
| huggingface | CamemBert | 1.16 |
| torchbench | hf_GPT2_large | 1.13 |
| torchbench | functorch_maml_omniglot | 1.12 |
| huggingface | BertForMaskedLM | 1.12 |
| huggingface | MT5ForConditionalGeneration | 1.12 |
| torchbench | hf_DistilBert | 1.11 |
| timm_models | mixnet_l | 1.11 |
| timm_models | tf_mixnet_l | 1.11 |

No perf regressions.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127195
Approved by: https://github.com/jansel
 2024-06-21 07:21:47 +00:00
Animesh Jain
c5e0b84484 [dynamo][trace_rules] Remove incorrectly classified Ingraph functions (#128428) 
Co-authored-by: Laith Sakka <lsakka@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128428
Approved by: https://github.com/yanboliang, https://github.com/mlazos
 2024-06-19 00:06:46 +00:00
PyTorch MergeBot
5bc9835d64 Revert "[dynamo][trace_rules] Remove incorrectly classified Ingraph functions (#128428)" 
This reverts commit c52eda896e.

Reverted https://github.com/pytorch/pytorch/pull/128428 on behalf of https://github.com/anijain2305 due to luca saw bad compile time ([comment](https://github.com/pytorch/pytorch/pull/128453#issuecomment-2176877667))
 2024-06-18 20:09:00 +00:00
Animesh Jain
b0282071c4 [dynamo] override torch.nn.modules.activation._is_make_fx_tracing (#128748) 
Discovered while inlining `MultiHeadAttention` nn Module.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/128748
Approved by: https://github.com/jansel
ghstack dependencies: #128315
 2024-06-17 08:49:29 +00:00
Laith Sakka
d0c08926d1 allow inlining functions in _python_dispatch and _is_make_fx_tracing (#128485) 
This fix grab breaks in torch_multimodal_clip benchmark.

Co-authored-by: Animesh Jain <anijain@umich.edu>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128485
Approved by: https://github.com/anijain2305
ghstack dependencies: #128428
 2024-06-13 09:56:39 +00:00
Animesh Jain
c52eda896e [dynamo][trace_rules] Remove incorrectly classified Ingraph functions (#128428) 
Co-authored-by: Laith Sakka <lsakka@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128428
Approved by: https://github.com/yanboliang, https://github.com/mlazos
ghstack dependencies: #126578, #128440, #128470, #128453, #128484
 2024-06-13 06:08:56 +00:00
Xu Han
9e39c62908 correct avx512_vnni isa name. (#128318) 
`x86` has two vnni isa currently: `avx2_vnni` and `avx512_vnni`.
This PR correct the function name to `avx512_vnni`.

Co-authored-by: Jiong Gong <jiong.gong@intel.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128318
Approved by: https://github.com/leslie-fang-intel, https://github.com/jgong5, https://github.com/desertfire
 2024-06-12 16:12:49 +00:00
Kulin Seth
8df56afc20 Add support in Python API for the recommended max working set size. (#128289) 
Adds ways for users to request recommended max size for Metal on Mac. It plumbs through
https://developer.apple.com/documentation/metal/mtldevice/2369280-recommendedmaxworkingsetsize?language=objc

Can be used like
```
        max_memory = torch.mps.recommended_max_memory()
        print ("Recommended Max Memory : ", (max_memory/(1024*1024*1024)), "GB")
```

Co-authored-by: Nikita Shulga <2453524+malfet@users.noreply.github.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128289
Approved by: https://github.com/malfet
 2024-06-12 16:03:57 +00:00
Sam Larsen
fa8ec8e718 [dynamo] handle hashable exceptions in trace_rules lookup (#128078) 
Summary: Found during user empathy day when attempting to hash a fractions.Fraction object before it was fully constructed. See https://github.com/pytorch/pytorch/issues/128075

Pull Request resolved: https://github.com/pytorch/pytorch/pull/128078
Approved by: https://github.com/anijain2305
 2024-06-10 18:23:22 +00:00
Aaron Orenstein
dcfa7702c3 Flip default value for mypy disallow_untyped_defs [1/11] (#127838) 
See #127836 for details.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127838
Approved by: https://github.com/oulgen
 2024-06-08 18:16:33 +00:00
PyTorch MergeBot
94165dba7b Revert "[dynamo] Inline the getattr of fx graph and proxy graph (#128172)" 
This reverts commit 662a78f957.

Reverted https://github.com/pytorch/pytorch/pull/128172 on behalf of https://github.com/anijain2305 due to pippy tests fail ([comment](https://github.com/pytorch/pytorch/pull/128172#issuecomment-2155835201))
 2024-06-08 06:29:36 +00:00
Xu Han
ba81c3c290 [inductor] add cpp builder code. (take 2) (#125849) 
Fully manual rebase the code of PR: https://github.com/pytorch/pytorch/pull/124045
The old PR seems crashed due to too many commits, and too many times rebase. Please reference: https://github.com/pytorch/pytorch/pull/124045#issuecomment-2103744588

-------
It is the first step of RFC https://github.com/pytorch/pytorch/issues/124245.
Changes:
1. Add cpp builder code, the new cpp_builder support Windows OS.
2. Add CPU ISA checker which is cross OS and exported from backend cpuinfo.
3. Switch compiler ISA checker to new cpp builder.
4. CppCodeCache use the new ISA checker.
5. Add temprary `test_new_cpp_build_logical` UT to help on transfer to new code.
<img width="1853" alt="Image" src="https://github.com/pytorch/pytorch/assets/8433590/ce6519ab-ba92-4204-b1d6-7d15d2ba2cbe">

Pull Request resolved: https://github.com/pytorch/pytorch/pull/125849
Approved by: https://github.com/jgong5, https://github.com/desertfire
 2024-06-07 20:49:58 +00:00
Animesh Jain
662a78f957 [dynamo] Inline the getattr of fx graph and proxy graph (#128172) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128172
Approved by: https://github.com/yanboliang
ghstack dependencies: #128001, #126578, #128158
 2024-06-07 17:14:58 +00:00
Xuehai Pan
c97e3ebb96 Fix wrongly exposed variables in torch/__init__.py (#127795) 
<img width="609" alt="image" src="https://github.com/pytorch/pytorch/assets/16078332/964c6707-1856-4c2c-8cd8-ce1d96d38d36">

This PR removes temporary variables in `torch/__init__.py`.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127795
Approved by: https://github.com/albanD
 2024-06-06 08:31:41 +00:00
PyTorch MergeBot
d1fad416a8 Revert "Add aten._unsafe_masked_index (#116491)" 
This reverts commit f03f8bc901.

Reverted https://github.com/pytorch/pytorch/pull/116491 on behalf of https://github.com/PaliC due to breaking onnx tests ([comment](https://github.com/pytorch/pytorch/pull/116491#issuecomment-2145557724))
 2024-06-03 15:51:50 +00:00
Isuru Fernando
f03f8bc901 Add aten._unsafe_masked_index (#116491) 
To generate masked indexing operations that would generate
masked loads in triton code

Pull Request resolved: https://github.com/pytorch/pytorch/pull/116491
Approved by: https://github.com/lezcano, https://github.com/peterbell10
 2024-06-03 14:44:03 +00:00
lezcano
48538d3d14 Implement svd_lowrank and pca_lowrank for complex numbers (#125580) 
We fix a number of bugs previously present in the complex
implementation.

We also heavily simplify the implementation, using, among
other things, that we now have conjugate views.

I saw there is a comment regarding how slow some checks on this
function are. As such, I removed quite a few of the combinations of inputs
to make the OpInfo lighter. I still left a couple relevant examples to not regress
coverage though.

Fixes https://github.com/pytorch/pytorch/issues/122188

Pull Request resolved: https://github.com/pytorch/pytorch/pull/125580
Approved by: https://github.com/pearu, https://github.com/peterbell10
 2024-05-30 14:45:58 +00:00
Andrew M. James
80a8fc07b2 [dynamo] Handle np.iinfo/finfo/dtype as input (#124482) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/124482
Approved by: https://github.com/lezcano
ghstack dependencies: #124481
 2024-05-29 16:00:15 +00:00
Animesh Jain
1507d5205a [dynamo][fsdp] Skip Dynamo tracing of __getattr__ if its top-level frame (#127263) 
The generated bytecode for the first frame is below. Inlined comments about the LOAD_ATTR which causes Dynamo to trigger again on `__getattr__`.

~~~
[__bytecode] MODIFIED BYTECODE fn /data/users/anijain/pytorch2/test/dynamo/test_activation_checkpointing.py line 1129
[__bytecode] 1129           0 COPY_FREE_VARS           1
[__bytecode]                2 RESUME                   0
[__bytecode]                4 PUSH_NULL
[__bytecode]                6 LOAD_GLOBAL             10 (__compiled_fn_1)
[__bytecode]               18 LOAD_FAST                0 (x)
[__bytecode]               20 LOAD_DEREF               1 (mod)
[__bytecode]               22 LOAD_ATTR                6 (_checkpoint_wrapped_module)
[__bytecode]               32 LOAD_CONST               1 (0)
[__bytecode]               34 BINARY_SUBSCR
[__bytecode]               44 LOAD_ATTR                7 (weight)
[__bytecode]               54 LOAD_DEREF               1 (mod)
[__bytecode]               56 LOAD_ATTR                6 (_checkpoint_wrapped_module)
[__bytecode]               66 LOAD_CONST               1 (0)
[__bytecode]               68 BINARY_SUBSCR
[__bytecode]               78 LOAD_ATTR                8 (bias)

# When this optimized bytecode is executed, these two lines call the __getattr__ of ActivationWrapper module.
# Dynamo gets invoked on __getattr__.

# If we had inlined __getattr__ during the tracing, we would have seen the LOAD_ATTR
# on more low level data structures like _modules, obviating the need for CPython
# to call python overriden __getattr__. But today, UnspecializedNNModuleVariable
# calls python getattr at tracing time (instead of inlining it), resulting in LOAD_ATTR
# on the module itself.

# To prevent Dynamo to skip tracing of __Getattr__ on the optimized bytecode,
# we can check if its top level frame and just skip it.

[__bytecode]               88 LOAD_DEREF               1 (mod)
[__bytecode]               90 LOAD_ATTR                0 (a)

[__bytecode]              100 PRECALL                  4
[__bytecode]              104 CALL                     4
[__bytecode]              114 UNPACK_SEQUENCE          1
[__bytecode]              118 RETURN_VALUE
~~~~

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127263
Approved by: https://github.com/yf225
 2024-05-28 08:16:53 +00:00
Yu, Guangye
e7a42702f9 generalize custom_fwd&custom_bwd to be device-agnostic (#126531) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126531
Approved by: https://github.com/jgong5, https://github.com/gujinghui, https://github.com/albanD, https://github.com/EikanWang
ghstack dependencies: #126527
 2024-05-25 06:48:16 +00:00
Tugsbayasgalan (Tugsuu) Manlaibaatar
72f0bdcc22 Remove torch._constrain_as_value (#127103) 
Summary: This API doesn't do anything useful and should be subsumed by torch._check.

Test Plan: CI

Differential Revision: D57786740

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127103
Approved by: https://github.com/angelayi
 2024-05-24 22:49:46 +00:00
Oguz Ulgen
a6155d23d1 [easy] Delete dead code global (#126903) 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126903
Approved by: https://github.com/aorenste
ghstack dependencies: #126083
 2024-05-23 08:29:29 +00:00
Oguz Ulgen
cc61d03ac9 Do not trace into triton/backends (#126083) 
Fixes #125807

Pull Request resolved: https://github.com/pytorch/pytorch/pull/126083
Approved by: https://github.com/yanboliang, https://github.com/jansel
 2024-05-23 08:29:29 +00:00
Jack Taylor
d30cdc4321 [ROCm] amdsmi library integration (#119182) 
Adds monitoring support for ROCm using amdsmi in place of pynvml.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/119182
Approved by: https://github.com/jeffdaily, https://github.com/malfet, https://github.com/xw285cornell
 2024-05-21 01:59:26 +00:00
Jason Ansel
f9de510121 [dynamo] Graph break on set_num_threads (#126623) 
Fixes #125364

Pull Request resolved: https://github.com/pytorch/pytorch/pull/126623
Approved by: https://github.com/yanboliang
 2024-05-20 17:44:32 +00:00
PyTorch MergeBot
315389bfed Revert "Remove deprecated _aminmax operator (#125995)" 
This reverts commit 0116ffae7f.

Reverted https://github.com/pytorch/pytorch/pull/125995 on behalf of https://github.com/huydhn due to Sorry for reverting your change but we need to reland this after I get rid of all usage of _aminmax internally in Meta ([comment](https://github.com/pytorch/pytorch/pull/125995#issuecomment-2113769497))
 2024-05-16 01:45:37 +00:00
cyy
0116ffae7f Remove deprecated _aminmax operator (#125995) 
It has been deprecated for a long time.

Co-authored-by: Edward Z. Yang <ezyang@meta.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125995
Approved by: https://github.com/ezyang
 2024-05-12 17:50:17 +00:00
PyTorch MergeBot
0d4fdb0bb7 Revert "[ROCm] amdsmi library integration (#119182)" 
This reverts commit 85447c41e3.

Reverted https://github.com/pytorch/pytorch/pull/119182 on behalf of https://github.com/huydhn due to Sorry for reverting your change, but the ROCm failed test is legit 85447c41e3 ([comment](https://github.com/pytorch/pytorch/pull/119182#issuecomment-2103433197))
 2024-05-09 21:18:21 +00:00
Jack Taylor
85447c41e3 [ROCm] amdsmi library integration (#119182) 
Adds monitoring support for ROCm using amdsmi in place of pynvml.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/119182
Approved by: https://github.com/jeffdaily, https://github.com/malfet, https://github.com/xw285cornell
 2024-05-09 18:21:38 +00:00
Michael Lazos
1b1b18a7a4 Add LRScheduler Composability E2E Tests (#125653) 
adds tests to verify the LRSchedulers correctly update the compiled optimizers without recompiles.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/125653
Approved by: https://github.com/yanboliang
ghstack dependencies: #123751, #123752, #123753, #125383
 2024-05-09 00:52:43 +00:00
PyTorch MergeBot
2e237fcd70 Revert "[inductor] add cpp builder code. (#124045)" 
This reverts commit 469383755f.

Reverted https://github.com/pytorch/pytorch/pull/124045 on behalf of https://github.com/clee2000 due to broke inductor/test_codecache and inductor/test_max_autotune 469383755f https://github.com/pytorch/pytorch/actions/runs/8996772350/job/24724775182 ([comment](https://github.com/pytorch/pytorch/pull/124045#issuecomment-2100851419))
 2024-05-08 15:33:20 +00:00
Xu Han
469383755f [inductor] add cpp builder code. (#124045) 
Previous full PR https://github.com/pytorch/pytorch/pull/115248 is failed to merge due to fb_code is hard to debug.
I also tried to submit them as two pieces, https://github.com/pytorch/pytorch/pull/118514 https://github.com/pytorch/pytorch/pull/118515. And they have passed PreCI at that time.

Now I tried to split https://github.com/pytorch/pytorch/pull/115248 into smaller piece, and it is the first step of RFC https://github.com/pytorch/pytorch/issues/124245.
Changes:
1. Add cpp builder code, the new cpp_builder support Windows OS.
2. Add CPU ISA checker which is cross OS and exported from backend cpuinfo.
3. Switch compiler ISA checker to new cpp builder.
4. CppCodeCache use the new ISA checker.
5. Add temprary `test_new_cpp_build_logical` UT to help on transfer to new code.
<img width="1853" alt="Image" src="https://github.com/pytorch/pytorch/assets/8433590/ce6519ab-ba92-4204-b1d6-7d15d2ba2cbe">

Pull Request resolved: https://github.com/pytorch/pytorch/pull/124045
Approved by: https://github.com/jgong5, https://github.com/jansel
 2024-05-08 05:27:15 +00:00