This PR fixes some issues with NJT backward / compile backward tests:
1. `requires_grad` was not being propagated appropriately during `SampleInput` generation, so a LOT of backward cases were untested before (sad times). This PR utilizes a helper function `_clone()` to clone() / detach() NJTs for SampleInputs while preserving `requires_grad` status. Note: the clone() / detach() stuff is for autograd; can't have two SampleInputs as part of the same autograd graph.
2. Per-sample skips weren't -fully- working; the op logic would still be invoked even with a skip. I found this out thanks to `split_with_sizes`, which segfaults during backwards because it tries to use an NST-specific formula. As annoying as it is, I tried a ton of things but ultimately had to split the `subtest_ctx` into that + a `skip_xfail_ctx` to run the subtests within.
* Updated all uses of per-sample skips / xfails: 4 in `test_nestedtensor.py` and 1 in `test_vmap.py`
3. Added the appropriate skips / xfails to get everything passing. There are a shitton of bugs to fix!
Pull Request resolved: https://github.com/pytorch/pytorch/pull/143072
Approved by: https://github.com/cpuhrsch, https://github.com/soulitzer
This is the initial foreach map HOP for pointwise ops which will be extended in the future to support grouped GEMMs and other ops.
This PR utilizes PrimHOPBase class to represent foreach_map as a HOP with a single subgraph. The way this is implemented is that the user API `foreach_map` provides a single pointwise torch op, and internally this function calls a polyfill which has the same semantics as a foreach op (ie iterates over lists of operands applying the op elementwise). The higher order op is passed through the stack down to inductor where a lowering in essence inlines the subgraph into the main graph. This is done by interpreting it with a pointwise subgraph lowering, grouping the outputs by device, and registering the output buffers as foreach groups as applicable. For testing I was able to reuse the existing foreach tests by creating a wrapper function which matches the foreach op interfaces for those tests and then run all of the existing foreach tests on foreach_map.
TODO before landing:
* Add tests for general functions
* Test warning if unsupported op will block fusion
Followups:
* I need to add tests for backwards (this will be a followup PR because backwards will require other work as well)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/142098
Approved by: https://github.com/eellison
Motivation: Generalize unit tests so that can be executed for cuda and non cuda devices.
Depedency : #133209 Merged now.
There was a #135242 for these changes and closed due to in correct commits. I have incoroprated the changes as suggested in comments.
@kwen2501 @zeshengzong Please review the changes.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139184
Approved by: https://github.com/kwen2501
Co-authored-by: Yu, Guangye <guangye.yu@intel.com>
**Overview**
This PR moves `torch/distributed/_composable/fsdp` to `torch/distributed/fsdp/_fully_shard` and makes public APIs available from `torch.distributed.fsdp`, e.g.:
```
from torch.distributed.fsdp import fully_shard
```
This is targeting 2.6 release. I rewrote some of the documentation with (hopefully) improved phrasing.
**Changes for Reland**
- Preserved the public objects from `torch/distributed/_composable/fsdp/fully_shard.py` so that the import path still works internally
- Added a unit test that we can do `from torch.distributed._composable.fsdp.fully_shard import FSDPModule`
Differential Revision: [D66890387](https://our.internmc.facebook.com/intern/diff/D66890387)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141868
Approved by: https://github.com/kwen2501, https://github.com/wconstab, https://github.com/weifengpy, https://github.com/fegin, https://github.com/XilunWu
Co-authored-by: Svetlana Karslioglu <svekars@meta.com>
Notable new features for SDPA operators on AMD systems from AOTriton 0.8b:
1. Nestedtensor support;
2. MQA/GQA support;
3. Restore Efficient attention support for causal=True and seqlen_q != seqlen_k cases;
+ The kernel should use top-left alignment, bottom right alignment will be added later
4. Move gfx1100 (RX7900/W7800/W7900) out of experimental support status.
However, users are strongly recommended to update to ROCM 6.2.4, notably for
its firmware updates.
Related unit tests are enabled as well.
Notable related changes from AOTriton 0.8b:
1. AOTriton 0.8b moves the GPU kernel out of libaotriton.so to a separate directory `aotriton.images`;
2. LZMA replaces ZSTD as GPU kernel compression algorithm for better compression ratio: aotriton0.8b (.so + aotriton.images take 350MB) compared to aotriton0.7b .so: 800MB
3. The compression cannot be disabled now, and `liblzma` is hard run-time dependency.
+ Should not be a problem, since `lzma` is part of Python Standard Library
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140172
Approved by: https://github.com/jithunnair-amd, https://github.com/jeffdaily
Co-authored-by: Jithun Nair <37884920+jithunnair-amd@users.noreply.github.com>
There are a number of cases where pattern matching differs based on the presence of ACL, causing the tests to fail. This adds `TEST_ACL` and `skipIfACL` so that these tests can still run with different values or be entirely skipped if necessary.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141921
Approved by: https://github.com/malfet
Co-authored-by: Nikita Shulga <2453524+malfet@users.noreply.github.com>
**Background:** conversion from outer dim -> inner dim makes the (previously valid) assumption that the ragged dim is immediately next to the batch dim. This is no longer the case after #137125.
This PR:
* Updates the outer dim -> inner dim conversion logic to match the actual ragged_idx. Since ragged_idx tells us where the packed ragged / batch dim is, both ragged and batch outer dims should map to this inner dim. The conversion logic must now take in `ragged_idx` to make this possible, so the PR updates all call-sites to pass this.
* Fixes outputs across keepdim settings when reducing over ragged / batch dims.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/142173
Approved by: https://github.com/drisspg
**Overview**
This PR moves `torch/distributed/_composable/fsdp` to `torch/distributed/fsdp/_fully_shard` and makes public APIs available from `torch.distributed.fsdp`, e.g.:
```
from torch.distributed.fsdp import fully_shard
```
This is targeting 2.6 release. I rewrote some of the documentation with (hopefully) improved phrasing.
**Follow-Ups**
- [x] Add some explanation in the docs about FSDP1 vs. FSDP2
- [ ] Move unit tests from `test/distributed/_composable/fsdp` to `test/distributed/fsdp/fully_shard/`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141868
Approved by: https://github.com/kwen2501, https://github.com/wconstab, https://github.com/weifengpy
Co-authored-by: Svetlana Karslioglu <svekars@meta.com>
Large n input caused a regression starting in ROCm 6.1. The for loop will run for an excessive number of iterations. The root cause seems to be how static_cast<int64_t> behaves for large float values such as 1e20 that certain unit tests will use. The workaround is to break out of the loop once the returned value reaches nan.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141150
Approved by: https://github.com/eqy, https://github.com/malfet
Using EC2 G6 instance, based on NVIDIA L4, added to scale config in https://github.com/pytorch/test-infra/pull/5376
To enable more balanced sharding, had to push 148ae19935
Added `@xfailIfSM89` to the following tests:
- test_fp8_pattern_2
- test_original_aten_preserved_split_addmm
- test_sparse_semi_structured_scaled_mm
- test_sparse_semi_structured_scaled_mm_fp8
- test_sparse_fp8fp8_mm
Increased tolerance to 2e-4 for `RNNTest.BidirectionalMultilayerGRU_CPU_vs_CUDA`
Skipped following inductor tests (that either flaky OOMs or timeouts):
- test_reduction_fn_std_float64
- test_reduction_fn_var_mean_float64
- test_multi_output_unbacked_custom_op
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140305
Approved by: https://github.com/wdvr, https://github.com/ZainRizvi
Certain `cpp_wrapper`-enabled tests were OOM-ing in the CI pipeline, with error messages suggesting that sufficient memory was accessible. This ultimately resulted from an internal memory limitation that was not queryable in the API. This PR adds querying for that limit.
Additionally, the failing tests had incorrect memory availability checks, and are updated with measured memory requirements.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140620
Approved by: https://github.com/malfet, https://github.com/eqy
ghstack dependencies: #141367
Previously, the reference inputs for div with rounding mode did not supply the rounding_mode keyword argument. This didn't match the sample inputs for this op.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136308
Approved by: https://github.com/albanD
Co-authored-by: Xia, Weiwen <weiwen.xia@intel.com>
Co-authored-by: Bob Ren <bobren@meta.com>
Co-authored-by: Xilun Wu <12968408+XilunWu@users.noreply.github.com>
Co-authored-by: siahuat0727 <tansiahuat@gmail.com>
Summary:
float8 dtypes were missing from this map, adding
Test Plan:
CI, and unbreaks debugging in torchao
If there is an existing test I can add this to - lmk
Reviewers:
Subscribers:
Tasks:
Tags:
Fixes #ISSUE_NUMBER
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141385
Approved by: https://github.com/soulitzer
This PR contains three `unsqueeze()`-related fixes for NJT:
1. Adjusts the output's `_ragged_idx` when `unsqueeze()` inserts a dim before the ragged dim
2. Corrects the unbind reference for `unsqueeze()` after the last input dim. For this case, the dim kwarg canonicalization logic needs to be applied wrt `inp.dim() + 1` to account for `dim=-1` properly
3. Adds ragged dim support to `unsqueeze()`, allowing for e.g. `(B, j1, D) -> (B, 1, j1, D)`. This is okay now after #137125
Note that `unsqueeze()` still doesn't support batch dim operation, and arguably should never support this.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141392
Approved by: https://github.com/cpuhrsch
ghstack dependencies: #141500, #140736, #140161
This PR introduces `ExtraOpData`, a structure that contains op metadata regarding whether the op is a view and the dim-related args it accepts. It also populates a huge database for dim-wise / view ops with this info.
Test logic (sample input generation, references) have been updated to utilize this data. It allows for a fairly generic set of sample inputs & a reference for the class of ops that accept a single NJT and operate dim-wise (AKA "unary dimwise ops").
Testing is added over the following ops:
* `chunk()`
* `narrow()`
* `select()`
* `split()`
* `split_with_sizes()`
* `squeeze()`
* `unflatten()`
* `unsqueeze()`
Most of the above do not operate on the ragged / batch dims or on non-contiguous NJTs, so the proper xfails are added as needed.
I also slipped in a couple minor fixes (sorry):
1. The `_wrap_jagged_dim()` helper now avoids assuming the `nt._ragged_idx == 1` and allows for a batch dim to be a valid input, disambiguating the converted inner dim as necessary through an additional `operating_on_batch` return value (i.e. both dim=0 and dim=1 map to dim=0 on the inner values tensor, since that dim represents a packed ragged dim for all batch items)
2. Padded dense -> NJT conversion requires shape gymnastics to operate with the restrictive FBGEMM kernel. The gymnastics were slightly wrong for the transposed NJT case, and this PR fixes that
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140161
Approved by: https://github.com/Skylion007, https://github.com/cpuhrsch
ghstack dependencies: #141500, #140736
**Background:** It's common to use `scalar_tensor()` in the input to `where()` to convert any scalars present to compatible tensors with matching options, *including layout*. This shows up in various places, notably including derivative formulas ([example](78491d6afc/tools/autograd/derivatives.yaml (L432-L434))). It causes problems for NJTs because they have `layout=torch.jagged` and it never makes sense to create a scalar tensor with this layout. Some of the breakage only seems to happen in CI for reasons I don't fully understand (see the revert of #140736 due to softshrink's derivative formula).
**This PR:**
* Allows non-contiguous NJT inputs to `where()` + adds tests for this
* Handles scalar tensor / dense tensor inputs for `condition` / `other` + adds tests for this
* Uses limited `broadcast_tensors()` / `broadcast_to()` support
* Improves `expand()` to work on non-contig NJTs
* Changes `scalar_tensor()` to use `torch.strided` instead of `torch.jagged` in both eager and torch.compile (i.e. meta registration)
* Changes backward formulas for `sinc`, `pow`, `special.i1`, and `special.i1e` to uses `scalar_tensor()` instead of e.g. `zeros({})`
**Alternative approach:** Update all problematic usages of `scalar_tensor()` to avoid ever passing `layout=torch.jagged`. This is an extensive change and includes `torch.where()` logic, a bunch of derivative formulas, and likely other places not yet discovered.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141500
Approved by: https://github.com/malfet, https://github.com/cpuhrsch, https://github.com/soulitzer
Summary: `TestFunc` is annotated as `Callable[[object], object]` which represents a callable that takes a single argument of any type (`object`) and returns a value of any type (`object`). However, in reality, `TestFunc` could be any number of arguments, as a result, the corret typing should be `Callable[[...], object]` instead which represents a callable that takes any number of arguments (including zero) and returns a value of any type (`object`).
Test Plan: Contbuild & OSS CI
Differential Revision: D66463705
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141513
Approved by: https://github.com/wz337, https://github.com/Skylion007
# Motivation
This PR add `XPUInductorQuantizer`, which would defined the recipe of int8 quantization at XPU backend.
# Detailed
The `XPUInductorQuantizer` is class derived from `X86InductorQuantizer` as both quantizer would take the advantage of highly optimized operators in oneDNN library(qconv, qlinear, qconv/qlinear fusion).
We share the same recipe as `X86InductorQuantizer`, so we would have same `annotate_xxxx` methods. So, in ideal situation, the `XPUInductorQuantizer` would have no class body as all implementation can inherit from base class.
In this PR, we override the `annotate_xxx` method for operators that has NOT be implemented. All operators XPU backend does not implement would be fallbacked to fp32 implementation as the node in graph is a `dq-op-q` pairs. This would help provide good OOB usability for XPU backend. On the other hand, the implemented operators would uses `annotate_op` implemented in base class and could be lowered successfully.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139578
Approved by: https://github.com/EikanWang, https://github.com/leslie-fang-intel, https://github.com/CuiYifeng, https://github.com/jerryzh168
ghstack dependencies: #133080
# Motivation
This PR enables the XPU quantized convolution. The operators it registers are `onednn::qconv_prepack`, `onednn::qconv1d_pointwise`, `onednn::qconv2d_pointwise`, `onednn::qconv3d_pointwise`. We share same operator schemas as Intel CPU backend as both would call kernels implemented in oneDNN library.
# Details
The implemented operators would be further integrated into pt2e quant flow. In this PR, we validated the kernel functionality via the UT in `test/inductor/test_mkldnn_pattern_matcher.py` where CPU backend defines a series of UT for quantized convolution. Also, we extend the device support for inductor lowering pass and inductor IR defined in `torch/_inductor/fx_passes/quantization.py` and `torch/_inductor/mkldnn_ir.py`. The overall picture would be that, CPU and GPU backend could share the general optimization pass(op fusion) and quantization inductor IR. After lowering, the final kernel would be dispatched to different implementation in oneDNN library.
In this PR, we share the same int8 quantizer in CPU, namely, `X68InductorQuantizer`. In next PR #139578, we will append a `XPUIndcutorQuantizer` which will customized the pt2e behaviors at XPU backend. The capability of `XPUInductorQuantizer` would gradually grow along with the development of quantized operators in XPU.
# Validation
* UT testing
```bash
python test/inductor/test_mkldnn_pattern_matcher.py -v \
-k test_qconv2d_xpu \
-k test_qconv2d_silu_xpu \
-k test_qconv2d_relu6_xpu \
-k test_qconv2d_hardtanh_xpu \
-k test_qconv2d_hardswish_xpu
```
* Runtime exemplification
```bash
#qconv2d
onednn_verbose,primitive,exec,gpu:0,convolution,jit:ir,forward_training,src_u8::blocked:acdb::f0 wei_s8::blocked:acdb::f0 bia_undef::undef::: dst_f32::blocked:acdb::f0,attr-scratchpad:user attr-scales:src0:0:f32+wei:1:f32 attr-zero-points:src0:0:s32 attr-post-ops:binary_add:f32:2+eltwise_linear:1,alg:convolution_direct,mb1_ic128oc128_ih6oh4kh3sh1dh0ph0_iw6ow4kw3sw1dw0pw0,0.0668945
#qconv2d_silu
onednn_verbose,primitive,exec,gpu:0,convolution,jit:ir,forward_training,src_u8::blocked:acdb::f0 wei_s8::blocked:acdb::f0 bia_undef::undef::: dst_u8::blocked:acdb::f0,attr-scratchpad:user attr-scales:src0:0:f32+wei:1:f32 attr-zero-points:src0:0:s32 attr-post-ops:eltwise_swish:1+binary_add:f32:2+eltwise_linear:0.0124779:22,alg:convolution_direct,mb1_ic3oc128_ih8oh6kh3sh1dh0ph0_iw8ow6kw3sw1dw0pw0,0.0881348
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133080
Approved by: https://github.com/guangyey, https://github.com/EikanWang, https://github.com/atalman
This PR contains three `unsqueeze()`-related fixes for NJT:
1. Adjusts the output's `_ragged_idx` when `unsqueeze()` inserts a dim before the ragged dim
2. Corrects the unbind reference for `unsqueeze()` after the last input dim. For this case, the dim kwarg canonicalization logic needs to be applied wrt `inp.dim() + 1` to account for `dim=-1` properly
3. Adds ragged dim support to `unsqueeze()`, allowing for e.g. `(B, j1, D) -> (B, 1, j1, D)`. This is okay now after #137125
Note that `unsqueeze()` still doesn't support batch dim operation, and arguably should never support this.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141392
Approved by: https://github.com/cpuhrsch
ghstack dependencies: #140736, #140161
This PR introduces `ExtraOpData`, a structure that contains op metadata regarding whether the op is a view and the dim-related args it accepts. It also populates a huge database for dim-wise / view ops with this info.
Test logic (sample input generation, references) have been updated to utilize this data. It allows for a fairly generic set of sample inputs & a reference for the class of ops that accept a single NJT and operate dim-wise (AKA "unary dimwise ops").
Testing is added over the following ops:
* `chunk()`
* `narrow()`
* `select()`
* `split()`
* `split_with_sizes()`
* `squeeze()`
* `unflatten()`
* `unsqueeze()`
Most of the above do not operate on the ragged / batch dims or on non-contiguous NJTs, so the proper xfails are added as needed.
I also slipped in a couple minor fixes (sorry):
1. The `_wrap_jagged_dim()` helper now avoids assuming the `nt._ragged_idx == 1` and allows for a batch dim to be a valid input, disambiguating the converted inner dim as necessary through an additional `operating_on_batch` return value (i.e. both dim=0 and dim=1 map to dim=0 on the inner values tensor, since that dim represents a packed ragged dim for all batch items)
2. Padded dense -> NJT conversion requires shape gymnastics to operate with the restrictive FBGEMM kernel. The gymnastics were slightly wrong for the transposed NJT case, and this PR fixes that
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140161
Approved by: https://github.com/Skylion007, https://github.com/cpuhrsch
ghstack dependencies: #140736
Adding `destroy_pg_upon_exit` property to allow derived Test classes to control whether auto destroy is desired.
(Otherwise, derived test classes will need to rewrite the `_run()` method, leading to duplicated code of `_run()` and if one needs to add things to `_run` in the future, more code change is needed.)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141192
Approved by: https://github.com/wconstab
* Automatically applies ruff rule 401. Turns loops into equivalent list comprehensions which are faster and do not leak the scope of the loop variables.
* list comprehensions not only often have better typing, but are 50+% faster than for loops on overhead. They also preserve length information etc and are better for the interpreter to optimize.
* Manually went back and made mypy happy after the change.
* Also fixed style lints in files covered by flake8 but not by pyfmt
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140980
Approved by: https://github.com/justinchuby, https://github.com/malfet
### Background
This PR adds the functionality to xfail / skip on a per-`SampleInput` basis for `OpInfo` tests. See #89354 and #82669 for some requests asking for this type of functionality.
This was originally landed for NJT in #138370 and is generalized and slightly tweaked here.
### Design
#### Principles
* Clean separation among `SampleInput` generation logic, test logic that uses the `SampleInput`s, and xfail / skip logic (which will change as bugs are addressed).
* Flexibility in xfail / skip predicate specification - ideally each bug can be handled by a single skip / xfail, even if it surfaces across a specific class of ops.
* This is important in practice for NJT, where it's common to have a bug that affects all binary ops, for example.
* Opt-in with minimal test logic changes + no substantial impact on other tests.
#### Details
The core new concept is a `SampleRule`, which can be either an `XFailRule` or `SkipRule`.
```python
@dataclass
class SampleRule(ABC):
# function to indicate whether the rule applies to this op; return True if so
# NB: str arg of callable is device_type
op_match_fn: Callable[[str, OpInfo], bool] = None
# function to indicate whether the rule applies to this sample; return True if so
sample_match_fn: Callable[[torch.device, SampleInput], bool] = None
# optional name for identifying the rule
name: str = ""
@dataclass
class XFailRule(SampleRule):
# expected error type
error_type: TypeVar = Exception
# expected error message
error_msg: str = ".*"
@dataclass
class SkipRule(SampleRule):
...
```
* See below for example usage details, but at a high level: each test should have a corresponding list of `sample_skips_and_xfails`.
* The list of `sample_skips_and_xfails` is traversed in order, and the first rule that matches (if any) is applied, so order can matter.
* The PR includes a logging mechanism for matched rules accessible by setting the loglevel to `DEBUG`.
* The split between `op_match_fn` and `sample_match_fn` is made to allow pre-filtering of the list of rules to get only those that apply to the op under test.
* Each `SampleInput` is run within a subtest context so they can be individually skipped / xfailed as needed. This also means that a test will no longer stop after the first erroring `SampleInput`; all samples will be run through test logic.
### Example Usage
Consider the following OpInfo test:
```python
class MyTestCase(TestCase):
@ops(op_db)
def test_foo(self, device, dtype, op):
for sample in op.sample_inputs(device, dtype, requires_grad=False):
# do some SampleInput-based test logic
output = op.op(sample.input, *sample.args, **sample.kwargs)
...
```
This is a common pattern for such tests; simply generate a list of `SampleInputs` and run them through the op. Now say you want to xfail one of these `SampleInput`s for a given op. Today, you have to xfail the entire test or hack around this in the test logic.
This PR lets you do this to get very flexible xfail / skips based on op / sample input properties:
```python
# NB: Define rules for per-SampleInput xfails / skips. These can also be defined in-line in the @ops decorator, but
# it can be more readable to maintain these somewhere else. These are attempted to be matched in order and
# the first one that matches applies, so order can matter.
FOO_SKIPS_AND_XFAILS = [
XFailRule(
error_type=ValueError,
error_mg="2D inputs not supported",
op_match_fn=lambda device, op: (
# NB: logic for which ops this rule applies to goes here
op.full_name == "add"
),
sample_match_fn=lambda device, sample: (
# NB: logic which samples this rule applies to goes here
sample.input.dim() == 2
),
# NB: optional rule identifier can help with debugging matched rules
name="add_with_2D_inputs_not_supported",
),
# NB: This follows a similar structure as XFailRule but without error_type / error_msg. Obviously
# this skips a particular SampleInput instead of xfailing :)
SkipRule(...),
...
]
class MyTestCase(TestCase):
@ops(op_db)
@sample_skips_and_xfails(FOO_SKIPS_AND_XFAILS)
# NB: the @ops decorator automatically filters out any rules that don't apply to this op
def test_foo(self, device, dtype, op):
for sample, subtest_ctx in op.sample_inputs(
# NB: use_subtests=True is required for skips / xfails to work. If skips / xfails are defined and use_subtests != True,
# an informative error will be thrown.
device, dtype, requires_grad=False, use_subtests=True
):
# NB: this subtest context manager runs each sample input as a "subtest" and handles skips / xfails appropriately
with subtest_ctx(self):
# do some SampleInput-based test logic
output = op.op(sample.input, *sample.args, **sample.kwargs)
...
```
More examples can be seen in `test/test_nestedtensor.py`, where this system is used in practice.
I also demonstrate usage of syntactic sugar over this system in `test/functorch/test_vmap.py`. Here, a skip for the `to()` operator is replaced with a granular xfail for `test_vmap_exhaustive()`:
```python
...
# pre-existing xfail
xfail("item"),
# new granular xfail using syntactic sugar over the general system
xfailIf(
"to",
lambda sample: (
sample.kwargs["memory_format"] == torch.channels_last
),
),
...
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140443
Approved by: https://github.com/janeyx99, https://github.com/zou3519
ghstack dependencies: #140160, #138370
Per discussion with @malfet, only allow weights_only unpickler to load NJT if `torch.nested` and `torch._dynamo` are imported
(this is slightly weird as technically `torch.nested` is actually imported by default and `torch._dynamo.decorators._DimRange` is actually what needs to be imported)
we can't import this from `torch.nested` as this would
- undo dynamo lazy import
- cause circular import
===========================
Redo of https://github.com/pytorch/pytorch/pull/140304 caused issues as `torch.nested._internal.foo` needs to be imported, which causes issues like
```python
torch/_weights_only_unpickler.py", line 339, in load
if full_path in _get_allowed_globals():
torch/_weights_only_unpickler.py", line 188, in _get_allowed_globals
torch.nested._internal.nested_tensor.NestedTensor
AttributeError: module 'torch.nested' has no attribute '_internal'
```
**This likely wasn't caught in our CI because imports are global during unit tests(?), so we use subprocess to properly test this time**
Differential Revision: [D65961691](https://our.internmc.facebook.com/intern/diff/D65961691)
@jbschlosser
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140739
Approved by: https://github.com/malfet
# Motivation
This pr is an extension of #131758. As described in #131758, these changes are looking to make distributed UTs more accessible to users of all device types.
It is a demonstration of a few changes discussed by @kwen2501 and @jgong5 in the discussion for #131758(https://github.com/pytorch/pytorch/pull/131758#discussion_r1762422784)
This PR contains two types of changes, the first is to the common distributed folder where we have added a new class derived from MultiProcessTestCase which helps abstracts out the process group creation /deletion and other functionality for a given device.
The new generalized content can be added by deriving from this base class.
Also includes other misc changes for gaudi support
The second changed file is test_functional_api. a test file in common distributed. This file is a POC for how we can use this new class to write more device agnostic distributed test cases.
The following changes have been made to test_functional_api.py:
-Functionality has been added to test for non cuda devices using intel HPU as an example
-Multiple set up steps previously required by MultiProcessTestCase have been abstracted out
-Misc adaptations to allow for general call to accelerators while adding test skips instead explicitly skipping for multiple GPUs
-Skipifhpu flags have been added to enable skipping a few Multithreaded test cases which are as yet not supported on HPUs
NOTE: Within test functional api, there are tests which require the use of some multithreading functions which are as yet not supported on HPUs. These have been skipped for hpu using skipHPU decorator.
I will be raising a separate PR to improve usability pf said decorators in a device agnostic setting in the manner suggested by @kwen2501 in a comment on this PR.
This pr is a cleaned up version of a previous PR(#136988) which I closed due to human error. I have addressed some of the comments made by @kwen2501 in this as well
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138216
Approved by: https://github.com/kwen2501, https://github.com/guangyey
Faced with an annoying string of warnings like this when running tests,
<img width="1644" alt="Screenshot 2024-11-15 at 11 23 21 AM" src="https://github.com/user-attachments/assets/91ff4e1d-3c29-4510-9a61-46e7df68a212">
My choices seem to be (1) call destroy_process_group() at the end of
each test fn, (2) do this in some wrapper, (3) do it in the base test
class.
Since tests in MultiProcessTestCase are responsible for calling
init_process_group themselves, they should also be responsible for
calling destroy (or at least method (3) would be asymmetric and may
result in double-destroy).
But it doesn't feel worth it to go add a destroy call manually to each
test, and try/except for a possible second destroy call seems like a
happy middle ground.
Note: tests that want to ensure that destroy runs cleanly can and should
still call destroy _inside_ the test, and this change does not affect
that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140820
Approved by: https://github.com/fegin
Faced with an annoying string of warnings like this when running tests,
<img width="1644" alt="Screenshot 2024-11-15 at 11 23 21 AM" src="https://github.com/user-attachments/assets/91ff4e1d-3c29-4510-9a61-46e7df68a212">
My choices seem to be (1) call destroy_process_group() at the end of
each test fn, (2) do this in some wrapper, (3) do it in the base test
class.
Since tests in MultiProcessTestCase are responsible for calling
init_process_group themselves, they should also be responsible for
calling destroy (or at least method (3) would be asymmetric and may
result in double-destroy).
But it doesn't feel worth it to go add a destroy call manually to each
test, and try/except for a possible second destroy call seems like a
happy middle ground.
Note: tests that want to ensure that destroy runs cleanly can and should
still call destroy _inside_ the test, and this change does not affect
that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140820
Approved by: https://github.com/fegin
ghstack dependencies: #140460, #140815
This PR adds caching for user defined triton kernels by putting the transitive closure of source code in node.meta along with constant arguments.
One HUGE hack we do here is a node looks like
```
triton_kernel_wrapper_functional_proxy = torch.ops.higher_order.triton_kernel_wrapper_functional(kernel_idx = 0, constant_args_idx = 1, grid = [(1, 1, 1)], tma_descriptor_
metadata = {}, kwargs = {'in_ptr0': arg0_1, 'in_ptr1': arg1_1, 'out_ptr': arg0_1}, tensors_to_clone = ['out_ptr']);
```
so we use regex to remove `kernel_idx = 0, constant_args_idx = 1` parts as they are not relevant to cache hash. This is horrible and I'd like to eventually not use pickle as a hashing alternative but this is a longer project.
Differential Revision: [D65895744](https://our.internmc.facebook.com/intern/diff/D65895744)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140326
Approved by: https://github.com/zou3519
Implementation of the `softmax_backward_data` operator for the CPU backend produces incorrect results when the `output` argument is non-contiguous.
Here is a test case that demonstrates this issue:
```python
torch.manual_seed(0)
op = torch.ops.aten._softmax_backward_data
grad_output = torch.ones(3, 3, 3)
temp = torch.randn(3, 10, 3)
out = temp[:, :3, :]
out = out.contiguous()
print(out.is_contiguous())
grad_input = op(grad_output, out, 1, torch.float32)
print(grad_input)
```
In this test case, the variable `grad_input` yields incorrect results if the line `out = out.contiguous()` is commented out. With this fix, `grad_input` consistently produces the same results whenever `output` is contiguous.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139740
Approved by: https://github.com/zou3519
Functionally two decorators are very similar, but one should rely on expectedFailure as much as possible to get signal when something is fixed.
- Move `product_version` variable from `test_mps` to common_utils, but call it `MACOS_VERSION`
- Introduce `skipIfMPSOnMacOS13` to decorate the hard crashes that happens only on MacOS13 (which at this point will not get any fixes and will be deprecated soon)
- Add `device_type='mps'` to all `skipIfMPS` per https://github.com/pytorch/pytorch/issues/140560
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139940
Approved by: https://github.com/janeyx99, https://github.com/huydhn
@ezyang noticed this exercises a multithreading bug that is causing tests to become disabled:
```
2024-11-13T21:05:55.8363582Z inductor/test_torchinductor_opinfo.py::TestInductorOpInfoCPU::test_comprehensive_fft_ihfftn_cpu_int32 /opt/conda/envs/py_3.9/lib/python3.9/site-packages/_pytest/threadexception.py:73: PytestUnhandledThreadExceptionWarning: Exception in thread Thread-3
2024-11-13T21:05:55.8364857Z
2024-11-13T21:05:55.8364974Z Traceback (most recent call last):
2024-11-13T21:05:55.8365491Z File "/opt/conda/envs/py_3.9/lib/python3.9/threading.py", line 980, in _bootstrap_inner
2024-11-13T21:05:55.8366003Z self.run()
2024-11-13T21:05:55.8366371Z File "/opt/conda/envs/py_3.9/lib/python3.9/threading.py", line 917, in run
2024-11-13T21:05:55.8366858Z self._target(*self._args, **self._kwargs)
2024-11-13T21:05:55.8367518Z File "/opt/conda/envs/py_3.9/lib/python3.9/site-packages/fbscribelogger/__init__.py", line 176, in _run_event_loop
2024-11-13T21:05:55.8368189Z self.loop.run_until_complete(self.task)
2024-11-13T21:05:55.8368774Z File "/opt/conda/envs/py_3.9/lib/python3.9/asyncio/base_events.py", line 647, in run_until_complete
2024-11-13T21:05:55.8369348Z return future.result()
2024-11-13T21:05:55.8369980Z File "/opt/conda/envs/py_3.9/lib/python3.9/site-packages/fbscribelogger/__init__.py", line 214, in _worker
2024-11-13T21:05:55.8370603Z message = await asyncio.wait_for(
2024-11-13T21:05:55.8371090Z File "/opt/conda/envs/py_3.9/lib/python3.9/asyncio/tasks.py", line 442, in wait_for
2024-11-13T21:05:55.8371573Z return await fut
2024-11-13T21:05:55.8372156Z File "/opt/conda/envs/py_3.9/lib/python3.9/asyncio/queues.py", line 166, in get
2024-11-13T21:05:55.8372613Z await getter
2024-11-13T21:05:55.8374010Z RuntimeError: Task <Task pending name='Task-1' coro=<FbScribeLogger._worker() running at /opt/conda/envs/py_3.9/lib/python3.9/site-packages/fbscribelogger/__init__.py:214> cb=[_run_until_complete_cb() at /opt/conda/envs/py_3.9/lib/python3.9/asyncio/base_events.py:184]> got Future <Future pending> attached to a different loop
2024-11-13T21:05:55.8375366Z
2024-11-13T21:05:55.8375603Z warnings.warn(pytest.PytestUnhandledThreadExceptionWarning(msg))
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140717
Approved by: https://github.com/ezyang, https://github.com/zxiiro
Summary:
`with_comms()` is mostly used as a decorator with an optional input argument `eager_init`. The problem of a decorator with input argument is that it has to be used with invocation always, i.e., you have to use as `with_comms()` rather than `with_comms` which majority of the existing usages.
This diff tries to provide a solution such that we could use `with_comms`, `with_comms()`, `with_comms(eager_init=False)`, and `with_comms(eager_init=True)`.
Test Plan: Contbuild & OSS CI
Differential Revision: D65385700
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139637
Approved by: https://github.com/wz337
This PR updates OpInfo-based tests for NJTs:
* Adds extensive coverage across non-contiguous NJTs (both non-contiguous transposed and non-contiguous with holes)
* The `_sample_njts()` helper that `sample_input_func`s utilize now produces non-contig NJTs as well
* Utilizes a `SampleInput`-based xfail system for granular classification of bugs. For example, it's possible to indicate that a class of ops is expected to fail only on non-contig with holes NJT inputs.
* I decided on adding `SampleInput`s and utilizing this system over using test parametrization for two reasons:
* Test perf - adding `SampleInput`s is faster than generating entire new tests
* Avoiding the possibility of `sample_input_func`s not respecting the non-contig test parameter - this would result in silently incorrect passing of these tests. Keeping the responsibility for `SampleInput` generation firmly within each `OpInfo`'s `sample_input_func` means weirdness like this isn't possible
* Improves `SampleInput` naming for a bunch of `sample_input_func`s. This makes it easier to xfail them as needed. For example, binary / unary / other ops now use the new `_describe_njt()` helper to get a string repr that uniquely defines the type of NJT being passed to the op
* Adds appropriate `XFailRule`s to get tests passing for forward / backward / forward compile / backward compile. In general, each xfail corresponds to some bug that needs to be fixed
```python
# Represents a rule indicating how to xfail a particular test. It allows granularity
# at the device, dtype, op, and individual sample levels. This flexibility allows entire
# bugs to be represented by a single rule, even if this corresponds with multiple conceptual
# test cases across multiple ops.
@dataclass
class XFailRule:
# expected error type
error_type: TypeVar = Exception
# expected error message
error_msg: str = ".*"
# function to indicate whether the rule applies; return True if so
match_fn: Callable[[torch.device, torch.dtype, OpInfo, SampleInput], bool] = None
# optional name for identifying the rule
name: str = ""
def match(self, device, dtype, op, sample) -> bool:
return self.match_fn(device, dtype, op, sample)
```
Example:
```python
# Bug when broadcasting a binary op with non-contiguous with holes NJT + dense
# tensor with 1 in ragged dim.
XFailRule(
error_type=RuntimeError,
error_msg="cannot call binary pointwise function .* with inputs of shapes",
match_fn=lambda device, dtype, op, sample: (
isinstance(op, BinaryUfuncInfo)
and "noncontig_holes" in sample.name
and "broadcasting 1 over ragged" in sample.name
),
name="binary_noncontig_holes_broadcasting_1_over_ragged",
),
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138370
Approved by: https://github.com/cpuhrsch, https://github.com/soulitzer
ghstack dependencies: #140160
Summary:
I was looking into why the non-standard bool value will fail for msort - it makes sense for argsort and sort to fail, because we're randomly generating uint8 so the order will be different (and thus the indices will be different). But msort should work.
After some digging, it's interesting that even though scalar_t is bool, when the actual value is a uint8_t, the comparison will treat them as signed. I tried lhs=255 and rhs=0: lhs < rhs is equivalent to -1 < 0 which is true (but it's supposed to be False)
Therefore we add an explicit type cast.
Test Plan: Remove the test skip
Differential Revision: D65472170
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139870
Approved by: https://github.com/Skylion007, https://github.com/davidberard98
Based on discussion here: https://github.com/pytorch/pytorch/pull/138731
Introducing ability for subclass implement type convertion to expected_type.
```
def __coerce_same_metadata_as_tangent__(
self, expected_metadata: Any, expected_type: Optional[Type] = None
):
```
Here if `expected_type=None` means `SubclassClass` is expected.
E.g. for `DTensor` we may find tangent `AsyncCollectiveTensor` where we expected `Tensor` - in this case
`expected_type=Tensor` will be called during runtime
Adding implementation to AsyncCollectiveTensor, that just triggers `wait()`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139095
Approved by: https://github.com/bdhirsh
This allows Configs to handle setting their defaults (or overriding
themselves) via environment variables.
The environment variables are resolved at install time (which is usually
import time). This is done 1) to avoid any race conditions between
threads etc..., but 2) to help encourage people to just go modify the
configs directly, vs overriding environment variables to change
pytorch behaviour.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138956
Approved by: https://github.com/ezyang
ghstack dependencies: #138766
As MacOS-15 or newer supports those out of the box. This significantly reduces memory requirements and improves performance for some stable diffision networks.
Test plan: Run
```python
from diffusers import StableDiffusionXLPipeline, AutoencoderKL, EulerAncestralDiscreteScheduler
import torch
import time
vae = AutoencoderKL.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0",
subfolder='vae',
torch_dtype=torch.bfloat16,
force_upcast=False).to('mps')
pipe = StableDiffusionXLPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", vae=vae,
torch_dtype=torch.bfloat16, variant="fp16").to('mps')
pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
start_time = time.time()
start_mps_mem = torch.mps.driver_allocated_memory()
image = pipe(prompt="Spherical cow in vacuum",
num_inference_steps=10,
guidance_scale=8,
generator=torch.Generator("mps").manual_seed(42),
).images[0]
end_mps_mem = torch.mps.driver_allocated_memory()
run_time = time.time() - start_time
print(f"run time in {run_time:.2f} sec, end_mps_mem {end_mps_mem/1024.0**2:.2f} Mb mem increase {(end_mps_mem-start_time)/1024.0**2:.2f} Mb")
image.save(f'bfloat16.png')
```
Before the change total memory use were 16Gb and needed 65 sec to complete, after it drops down to 14Gb and takes 50 sec to finish on M2Pro, though generated image remains the same:
Fixes https://github.com/pytorch/pytorch/issues/139389
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139791
Approved by: https://github.com/drisspg, https://github.com/Skylion007
ghstack dependencies: #139788, #139784, #139763
This teaches install_config_module (and the underlying code) to
understands Config objects. Additionally we've added a JK option to this
which resolves the JK.
This config gets stored within the _ConfigEntry class and is evaluated
when __getattr__ is called. If justknobs is set, it'll call
justknobs_check to see the result.
Due to preceeding work, basically everything works correctly here and we
had to update a couple of tests, and modify the getattr behaviour.
Note that we are updating the justknob_check function to support a
default option, to make default work.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138766
Approved by: https://github.com/ezyang
I'm sick of reductions not working properly - spotty dim coverage, missing backwards, etc. This PR fixes quite a bit.
It applies to the following ops:
* `sum` / `mean` / `prod`
* `all` / `any`
* `amin` / `amax`
* `min` / `max`
* `argmin` / `argmax`
The general reduction logic has been factored out into a helper `_apply_reduction(func, func_name, identity_element, *args, **kwargs)`. The idea is that by providing a valid identity element, we can utilize conversions to padded dense when needed for reducing over the ragged dim.
Extensive test coverage includes:
* reductions across ragged dim
* reductions across non-batch, non-ragged dims
* reductions across both batch and ragged dims
* multiple dim reductions (for ops that support this)
* full reduction -> scalar
Bonus: the PR includes backwards fixes for `sum` and `mean`, which have never worked.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139317
Approved by: https://github.com/cpuhrsch
**Background:** The `@parametrize` decorator enjoys widespread usage as a convenient tool for ensuring extensive test coverage. One particular feature that makes this easy is the ability to stack such decorators, testing over the cross-product of inputs. Example:
```python
class MyTestClass(TestCase):
@parametrize("x", range(3))
@parametrize("y", [False, True])
def test_foo(self, x, y):
# Invoked with:
# x=0, y=False
# x=1, y=False
# x=2, y=False
# x=0, y=True
# x=1, y=True
# x=2, y=True
...
```
Note that the `@ops` and `@modules` decorators employ the same underlying machinery for parametrizing over `OpInfo` / `ModuleInfo` entries. These decorators also parametrize over op-specific `device` / `dtype` info *according to what is supported for each op*.
```python
class MyTestClass(TestCase):
@ops(op_db)
def test_foo(self, op, device, dtype):
# Invoked each OpInfo in the db along with each device / dtype that corresponds
# with this op according to the OpInfo entry.
...
```
Note that this in contrast to the naive cross product between ops and devices / dtypes, which would generate too many tests. Certain use cases benefit from a similar type of flexible parametrization that is more intelligent than simple cross-product composition. It is expensive to generate / run too many tests, even if the unneeded ones are skipped appropriately.
This PR attempts to generalize such flexible parametrization and satisfy these use cases through the introduction of a `@reparametrize` decorator, which operates on an existing parametrizer and allows for customized on-the-fly parametrization through the use of an `adapter_fn`. Examples:
```python
# adapter_fn that adds a new arg
def include_is_even_arg(test_name, param_kwargs):
x = param_kwargs["x"]
is_even = x % 2 == 0
new_param_kwargs = dict(param_kwargs)
new_param_kwargs["is_even"] = is_even
is_even_suffix = "_even" if is_even else "_odd"
new_test_name = f"{test_name}{is_even_suffix}"
yield (new_test_name, new_param_kwargs)
# adapter_fn that excludes certain values
def exclude_odds(test_name, param_kwargs):
x = param_kwargs["x"]
is_even = x % 2 == 0
yield None if not is_even else (test_name, param_kwargs)
class MyTestClass(TestCase):
@reparametrize(parametrize("x", range(5)), include_is_even_arg)
def test_foo(self, x, is_even):
# Invoked with both the x value and the new is_even arg
...
@reparametrize(parametrize("x", range(5)), exclude_odds)
def test_bar(self, x):
# Only invoked with even x values
...
```
For a more real-world use case, imagine you want to write a set of OpInfo tests that parametrize over additional op-specific things beyond `device` / `dtype` (in NJT's case, this includes contiguity type, whether to operate over the batch / ragged / other dims, etc.). The `@reparametrize` decorator allows you to customize the `@ops` parametrization to add in these additional args as they make sense on a per-op basis.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138369
Approved by: https://github.com/janeyx99
We should only pass the `device_id` when the backend is `nccl`. Otherwise, we would run into the following error:
```
RuntimeError: No backend for the parent process group or its backend does not support splitting
```
This also fixes test failure is not asserted when using `with_comms()` or `with_comms(eager_init=False)`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139097
Approved by: https://github.com/XilunWu
This PR adds FlexAttention + NJT support. In particular:
* To handle raggedness, treats the packed sequence dim of input NJTs as a giant "stacked sequence". To ensure user `score_mod` / `mask_mod` functions can still be written in the original NJT sequence space, this PR handles conversions for indices within the giant "stacked sequence" -> sequence relative indices automatically.
* Provides `py_impls` for `NestedTensor` to the HOPs for flex attention forward / backward that simply wrap / unwrap NJTs appropriately
* Adds barebones `new_empty()` support to NJT since FlexAttention utilizes this repeatedly; right now, only `new_empty()` with a shape of `()` is supported
* Tests that FlexAttention with a causal mask matches causal SDPA
* Adds a new public API for FlexAttention usage:
* `create_nested_block_mask(mask_mod, B, H, njt, BLOCK_SIZE, _compile)` - NJT analogue for `create_block_mask()` that utilizes the `njt`'s ragged structure to create an appropriately-sized block mask (e.g. `(1, 1, total_seqlen, total_seqlen)`). This function handles the index conversion from "stacked sequence" space -> relative sequence space.
* Minor note: as this is a public API, this function is purposefully named with "nested" instead of "njt" to keep the latter as an informal, mostly internal-only term.
Example usage:
```python
def causal_mask(b, h, q_idx, kv_idx):
return q_idx >= kv_idx
query = ... # NJT of shape (B, H, S*, D)
key = ... # NJT of shape (B, H, S*, D)
value = ... # NJT of shape (B, H, S*, D)
# create_nested_block_mask() automatically converts indices from "stacked sequence" space -> relative sequence space
block_mask = create_nested_block_mask(causal_mask, 1, 1, query) # block mask conceptual shape is (B, H, sum(S*), sum(S*))
output = flex_attention(query, key, value, block_mask=block_mask)
def causal_score_mod(score, b, h, q_idx, kv_idx):
return torch.where(q_idx >= kv_idx, score, float("-inf"))
# flex_attention() automatically converts indices from "stacked sequence" space -> relative sequence space for NJT inputs
output2 = flex_attention(query, key, value, score_mod=causal_score_mod)
```
TODO:
* ~~Determine the right level of abstraction for public API helpers + move them alongside other helpers~~ Verify this with others though
* ~~Some cleanup~~
* ~~`njt_score_mod_adapter`~~
* ~~Q: should `create_njt_block_mask()` call `njt_mask_mod_adapter()` so we don't need two calls?~~
* Can we avoid materializing the `sum(s)` length `seq_idx` used for conversion between stacked sequence -> sequence relative indices?
* Not for now, although future work may deepen the integration between Flex + NJT (possibly requiring custom templates). We should try to cache this though.
* ~~Demonstrate non-causal mask~~
* Support non-contiguous NJTs with holes (**booted to future PR**)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136792
Approved by: https://github.com/drisspg
ghstack dependencies: #138841
dot reference implementation should be consistent with the cpu / cuda implementations since it may be used for meta dispatch
i.e.
```python
import torch
x = torch.tensor([1,2,3], dtype=torch.float32)
y = torch.tensor([4,5,6], dtype=torch.float16)
x.dot(y)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
RuntimeError: dot : expected both vectors to have same dtype, but found Float and Half
```
However the below does not raise an exception
```python
x.to("meta").dot(y.to("meta"))
```
Fixes #ISSUE_NUMBER
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138596
Approved by: https://github.com/bdhirsh
This PR combines a number of cleanups in one PR. If any of the specific cleanups don't seem to make sense, let me know and I can remove them.
Cleanups
- This PR adds a set of test suites for the config module code, which handles basically all the APIs and ways it is used. Please let me know if you see anything critical that is not tested that I missed. This test suite is primarily used as the regression test suite for later changes in this diff. Note that there is some dynamo specific testing of the config module, but it isn't as verbose.
- I removed all internal usage of shallow_copy_dict. Those usages could all use the deep copy, and did not depend on the reference behavior of certain config values that shallow_copy_dict allows.
- I removed shallow copy semantics for configuration with a deprecation warning. I think this requires a release note, so hopefully I did that correctly. Let me know if we want to continue to expose shallow copy value semantics, but I just can't find a case where I expect anyone would want it. It also complicated later internal changes to the API (i.e. breaking apart various layers of the config changes).
- I fixed what I believe is a bug in how hashes are calculated on configs. In particular, if you got the hash, then made a config change, and then got the hash again, it would not update the hash. @oulgen, please let me know if I'm misunderstanding this behavior and it is desired.
- I switched our multiple implementations of iterating through the dictionary to a single one. This is primarily to make later changes easier, but it also makes it clear how inconsistent our various config ignoring options are. Let me know if people would be interested in me unifying the various options for ignoring config values.
- I updated the test patcher (not the performance critical one, just the normal one), to use __setattr__ and __getattr__ to remove direct API access to the underlying config fetcher.
For release notes, Not sure exactly how to communicate this, but something like
"ConfigModule.to_dict, and ConfigModule.shallow_copy_dict no longer retain their shallow copy semantics, which allowed reference values objects to be modified. If you wish to modify the config object, call load_config explicitly".
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138377
Approved by: https://github.com/ezyang, https://github.com/jansel, https://github.com/jovianjaison
This has the benefit that
1) It's much easier to aggregate test failure repros into say a CSV or shell script from scuba
2) We can do analysis (eg. set different two sets of tests across two PRs)
3) We can get results faster at the test-level granularity instead of job-level granularity we see in the HUD/GH.
I tested this by introducing a breaking change, adding ci-scribe label and then verifying that the failed tests were logged to scuba: https://fburl.com/scuba/torch_open_source_signpost/w6qt7qr9
I then reverted the breaking change and published this PR.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138394
Approved by: https://github.com/ezyang
Add an optional `eager_init` flag to `with_comms`.
When `eager_init` is True and backend is `nccl`, we pass the `device_id` to `init_process_group()` for eager initialization.
Otherwise, `device_id` is still `None` and this goes through the normal lazy call.
Default for `eager_init` is False.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138108
Approved by: https://github.com/kwen2501
The logsumexp tensor was considered for internal use only but apparently exposed to unit tests and inductors.
The stream should be selected after picking the current device. Otherwise the code is checking the default device's architecture.
Fixes#131316#137414
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137717
Approved by: https://github.com/drisspg
Co-authored-by: Jack Taylor <108682042+jataylo@users.noreply.github.com>
Our matmul support is abysmal - let's at least get this working and do it performantly later.
Bonus: implements `bmm` as well.
jagged <-> padded dense conversions are utilized when possible, and an unbind-based fallback otherwise (the former works with torch.compile and the latter doesn't). Some testing is missing because we don't have factory function support yet :(
Pull Request resolved: https://github.com/pytorch/pytorch/pull/138121
Approved by: https://github.com/cpuhrsch
This adds Dynamo tracing support for the host-side Triton TMA API (see `create_2d_tma_descriptor` calls on the host in the [Triton tutorial](https://triton-lang.org/main/getting-started/tutorials/09-persistent-matmul.html#sphx-glr-getting-started-tutorials-09-persistent-matmul-py)). A few notes:
- Here we assume the availability of the host-side TMA API added to upstream Triton in https://github.com/triton-lang/triton/pull/4498. As of time of writing, this is not a part of the PT2 OSS Triton pin (although back-ported internally). OSS Triton pin update should be done in December 2024.
- To capture the chain of calls `t.data_ptr() --> create_{1d,2d}_tma_descriptor(ptr, ...) --> kernel[grid](tma_desc, ...)`, we add three new variable trackers: `DataPtrVariable`, `CreateTMADescriptorVariable` (for the function), `TMADescriptorVariable` (for TMA descriptor object). This is to maintain the path back from the Triton kernel to the Tensor from which the TMA descriptor has been created.
- The newly introduced variables have `reconstruct` methods used in case of graph breaks.
- The `tma_descriptor_metadata` extracted from the captured `create_{1d,2d}_tma_descriptor` calls is propagated through the HOPs in Dynamo and AOTAutograd to be used by the downstream compiler (e.g., Inductor). See the unit tests for how the captured HOP arguments look like.
- In the Dynamo-captured fx graph, we replace the TMA descriptor arguments of the Triton kernel by the underlying Tensors, to be able to track the input/output relationships in terms of Tensors.
- In the Triton kernel mutation analysis pass (in AOTAutograd), we use the `tt.experimental_descriptor_store` TTIR op to detect mutations of the underlying tensors via TMA descriptors. So that downstream AOTAutograd can perform functionalizations as required.
- JIT Inductor and AOT Inductor support will be implemented in follow-up PRs.
Differential Revision: [D64404928](https://our.internmc.facebook.com/intern/diff/D64404928)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137677
Approved by: https://github.com/zou3519
### Fix 1: Throw async error during init wait
Previously we just busy wait for `ncclSuccess`, if the nonblocking init encountered error, we never report that. Added detection of async error via `ncclGetAsyncError`.
### Fix 2: Add wait after comm split
```
// After calling ncclCommSplit in non-blocking mode, we should wait for the
// source communicator to be out of ncclInProgress state.
// Reason 1:
// it's unsafe to call new operations on the parent comm while it's in
// ncclInProgress state.
// Reason 2:
// as of NCCL 2.23, the ptr value of child comm will not be filled until the
// state of parent comm is ncclSuccess. This may change in the future. See:
// https://github.com/NVIDIA/nccl/issues/1472
```
This wait does not mean the child comm is ready for use, neither does it block till that point.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137741
Approved by: https://github.com/shuqiangzhang
A proposal addressing Issue #1489: **Optimizer should track parameter names and not id.**
(also mentioned in here: [[RFC] Introducing FQNs/clarity eyeglasses to optim state_dict](https://dev-discuss.pytorch.org/t/rfc-introducing-fqns-clarity-to-optim-state-dict/1552)
## Summary
This PR introduces a backward-compatible enhancement where optimizers track parameter names instead of just their id.
Optimizers can be initialized with `named_parameters()` as:
```python
optimizer = optim.SGD(model.named_parameters(), lr=0.01, momentum=0.9)
```
This allows for greater clarity and ease when handling optimizers, as the parameters' names are preserved within the optimizer’s `state_dict` as:
```
state_dict =
{
'state': {
0: {'momentum_buffer': tensor(...), ...},
1: {'momentum_buffer': tensor(...), ...},
},
'param_groups': [
{
'lr': 0.01,
'weight_decay': 0,
...
'params': [0,1]
'param_names' ['layer.weight', 'layer.bias'] (optional)
}
]
}
```
Loading `state_dict` is not changed (backward-compatible) and the `param_names` key will be ignored.
## Key Features
#### Named Parameters in Optimizer Initialization:
Optimizers can accept the output of `model.named_parameters()` during initialization, allowing them to store parameter names directly.
#### Parameter Names in `state_dict`:
The parameter names are saved as a list in the optimizer’s `state_dict` with key `param_names`, alongside the `params` indices, ensuring seamless tracking of both names and parameters.
## Backward Compatibility
#### No Breaking Changes:
This change is fully backward-compatible. The added `param_names` key in the optimizer's `state_dict` is ignored when loading a state to the optimizer.
#### Customization with Hooks:
For more control, the loaded state_dict can be modified using a custom `register_load_state_dict_pre_hook`, providing flexibility for different design needs.
## Documentation Updates
Please refer to the documentation changes for more details on how this feature is implemented and how it can be used effectively.
## Solution Example:
A suggested solution to the problem mentioned in #1489, for the same parameters but in a different order.
The following `register_load_state_dict_pre_hook` should be added to the optimizer before loading to enable loading the state dict :
```python
def adapt_state_dict_ids(optimizer, state_dict):
# assuming a single param group.
current_state_group = optimizer.state_dict()['param_groups'][0]
loaded_state_group = state_dict['param_groups'][0]
# same number of params, same names, only different ordering
current_state_name_to_id_mapping = {} # mapping -- param_name: id
for i, name in enumerate(current_state_group['param_names']):
current_state_name_to_id_mapping[name] = current_state_group['params'][i]
# changing the ids of the loaded state dict to match the order of the given state dict.
for i, name in enumerate(current_state_group['param_names']):
loaded_state_group['params'][i] = current_state_name_to_id_mapping[name]
return state_dict
```
In this code, the loaded `state_dict` ids are adapted to match the order of the current optimizer `state_dict`.
Both the previous and the current optimizers are required to be initiated with `named_parameters()` to have the 'param_names' key in the dict.
### Note
This is my first contribution to PyTorch, and I wish to receive feedback or suggestions for improvement.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134107
Approved by: https://github.com/janeyx99
Co-authored-by: Jane (Yuan) Xu <31798555+janeyx99@users.noreply.github.com>
This PR contains multiple fixes for issue https://github.com/pytorch/pytorch/issues/135279:
## First part:
Moves the GPU guard (`cudaSetDevice`) before the `currentStreamCaptureStatusMayInitCtx` call.
As its name suggests, it May Init Ctx.
## Second part:
Even with the above fix, additional contexts are still observed during Work object destruction, e.g.
```
work = dist.all_reduce(tensor, async_op=True)
time.sleep(5) <-- no additional context yet
del work <-- additional context shows up
```
### Debug process
Chasing it down to destruction of a `Future` object -- a member variable of `Work`.
Then further down to the following member of `Future`:
```
std::vector<c10::Event> events_;
```
When the `events_` are destroyed, we hit the road down to:
1f3a793790/c10/cuda/impl/CUDAGuardImpl.h (L106-L121)
When there is no "preset" CUDA context (**which is the case for python garbage collector**), line 112: `c10::cuda::GetDevice(&orig_device)` will set `orig_device` to 0. Then, at line 120, `c10::cuda::SetDevice(orig_device)` will "officially" set the context to device 0 --
**that's where rank 1, 2, ... can create extra context on device 0!**
### Solution
This PR adds an explicit destructor to `Future`. In this destructor, destroy each event with a device guard.
## Test
Added test_extra_cuda_context, implemented via
- `pynvml` (if available), or
- memory consumption check.
`python test/distributed/test_c10d_nccl.py -k test_extra_cuda_context`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/135273
Approved by: https://github.com/fduwjj, https://github.com/wconstab, https://github.com/eqy
ghstack dependencies: #137161
This PR contains multiple fixes for issue https://github.com/pytorch/pytorch/issues/135279:
## First part:
Moves the GPU guard (`cudaSetDevice`) before the `currentStreamCaptureStatusMayInitCtx` call.
As its name suggests, it May Init Ctx.
## Second part:
Even with the above fix, additional contexts are still observed during Work object destruction, e.g.
```
work = dist.all_reduce(tensor, async_op=True)
time.sleep(5) <-- no additional context yet
del work <-- additional context shows up
```
### Debug process
Chasing it down to destruction of a `Future` object -- a member variable of `Work`.
Then further down to the following member of `Future`:
```
std::vector<c10::Event> events_;
```
When the `events_` are destroyed, we hit the road down to:
1f3a793790/c10/cuda/impl/CUDAGuardImpl.h (L106-L121)
When there is no "preset" CUDA context (**which is the case for python garbage collector**), line 112: `c10::cuda::GetDevice(&orig_device)` will set `orig_device` to 0. Then, at line 120, `c10::cuda::SetDevice(orig_device)` will "officially" set the context to device 0 --
**that's where rank 1, 2, ... can create extra context on device 0!**
### Solution
This PR adds an explicit destructor to `Future`. In this destructor, destroy each event with a device guard.
## Test
Added test_extra_cuda_context, implemented via
- `pynvml` (if available), or
- memory consumption check.
`python test/distributed/test_c10d_nccl.py -k test_extra_cuda_context`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/135273
Approved by: https://github.com/fduwjj, https://github.com/wconstab, https://github.com/eqy
Function `_get_pg_default_device` is being used outside of `distributed_c10d.py`.
A concern is that people may not be aware of what it actually does, due to bad naming of this function:
`Return the device to use with ``group`` for control flow usage (object collectives, barrier).`
The remediation is as follows:
- Added a deprecation warning to `_get_pg_default_device`;
- Added a private function `_get_object_coll_device` to undertake what it does;
- Added a `_device_capability` function for users who want to query the device support of a PG -- it returns a plain list, no more "default" choice.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136790
Approved by: https://github.com/H-Huang
Fixes#130154
This PR takes the strategy outlined in the above issue and clears out any cached sizes / strides PyCapsules before serialization. This affects the default subclass serialization logic.
The PyCapsule issue also affects `deepcopy`, so that's fixed here as well.
Note: I originally tried utilizing a context manager to remove / restore cached PyCapsules after serialization, but in practice the state returned from `_reduce_ex_internal()` references the actual `tensor.__dict__()`, so the problem persists once the cached values are restored. Instead, we have to be careful to remove the cached values in the right place so they're not re-cached when pulling out size / stride information for serialization.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137030
Approved by: https://github.com/albanD
## Motivation
The FSDP common code for FSDP UT execution is mostly written with cuda device in mind. However other devices such the intel Gaudi supports most of the functionality. We are generalizing the base content so that the UT content can be used for non-cuda device execution.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133209
Approved by: https://github.com/kwen2501
Fix two more leaks of the same variety as #136507 (see that PR desc and attached gdoc for debug details).
This time, also add a test-time check that helped to discover new leaks and ensure we won't accidently regress.
Adds `check_tensor_leak` util which internally asserts no tensors are being kept alive by other objects involved in py ref cycles.
Uses objgraph for a nice debug utility when a leak is found.
Credit to @H-Huang for pointing out objdump and helping debug the 'param_group["intermediates"]` leak.
I manually confirmed that all 3 of the leaks identified/fixed so far are caught by the unit test and checker.
Sample output, if I re-introduce a leak by commenting out `del param_group["intermediates"]` in _backward.py,
and run `python test/distributed/pipelining/test_schedule_multiproc.py -k test_schedule_with_native_zero_bubble`:
```
warnings.warn(
/data/users/whc/pytorch/torch/testing/_internal/common_utils.py:5341: UserWarning: 34 tensors were found in the garbage. Did you introduce a reference cycle?
warnings.warn(
/data/users/whc/pytorch/torch/testing/_internal/common_utils.py:5347: UserWarning: Dumping first 1 objgraphs of leaked tensors rendered to png
Graph written to /tmp/objgraph-ztz642h3.dot (19 nodes)
Graph viewer (xdot) not found, generating a png instead
Image generated as /tmp/objgraph-ztz642h3.png
```
rendering of ` /tmp/objgraph-ztz642h3.png`:
<img width="1671" alt="image" src="https://github.com/user-attachments/assets/9098ff29-224c-4533-935b-83c210ac2e22">
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136584
Approved by: https://github.com/kwen2501, https://github.com/H-Huang
ghstack dependencies: #136507
Co-authored-by: Howard Huang <howardhuang@fb.com>
Related: #132695
This PR uses padded dense <-> jagged conversions to handle binary pointwise broadcasting of (NT, T) and (T, NT). This includes:
* `(B, j0, D) + (1, 1, 1)`
* `(B, j0, D) + (B, 1, 1)`
* `(B, j0, D) + (B, 1, D)`
* etc.
This PR also adds (hacky) support for bool inputs to the jagged <-> padded dense conversions. The underlying CUDA kernels do not support integer / bool inputs; so the following workaround is employed: `convert input -> half, run conversion kernel, convert output -> bool`. Note that this bool support is needed specifically for the backward formula of `fmax`, and likely others.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133021
Approved by: https://github.com/cpuhrsch
Fixes#93843
`EmbeddingBag()` / `embedding_bag()` support 1D inputs with offsets to handle raggedness. NJT is a natural fit here as it already maintains offsets of the same form. This PR updates the python-side to support NJT and adds corresponding OpInfo-based NJT tests.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/135888
Approved by: https://github.com/cpuhrsch
Summary:
To facilitate PSS-2 upgrade, this uses `ndt.NDArray` instead of `nd.ndarray` in type annotations. In Numpy-1.19 (PSS-1) it's an alias to `nd.ndarray` -- a noop.
In Numpy-1.24, `ndt.NDArray` a proper generic type, and without this change uses of `nd.ndarray` generate this Pyre type error:
```counterexample
Invalid type parameters [24]: Generic type `np.ndarray` expects 2 type parameters.
```
Test Plan: Sandcastle plus visual inspection
Differential Revision: D62977370
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136288
Approved by: https://github.com/kit1980
skip_if_rocm is used only in multiprocess case (when UT test class is a child of MultiProcessTestCase). Each individual process can exit with a skip code. If used for single process UT, it will cause the UT to fail as the process returns a non-zero exit code. Use skipIfRocm in single process UTs.
To avoid the above confusion, this PR renamed skip_if_rocm to skip_if_rocm_multiprocess.
Fixes #ISSUE_NUMBER
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136161
Approved by: https://github.com/jithunnair-amd, https://github.com/kwen2501, https://github.com/fegin
`rms_norm()` is a nice-to-have for ViT :)
This PR:
* SymInt-ifies `rms_norm()`, allowing NJT to use the same decomp.
* Adds torch_function-based input validation logic for nested-specific stuff (no normalization supported over the ragged dim for now) on the python NJT side.
* Adds multi-dim support (on non-ragged, non-batch dims) to `mean()` for NJT.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/135872
Approved by: https://github.com/mikaylagawarecki
ghstack dependencies: #125947
Fixes#136090
* Add support for isin to tensor half dtypes for CPU (just add a few extra dispatches).
* Seems like the CUDA implementation for bfloat16 was mostly compiled and available all along (it just calls sort internally AND unique). To enable it, we just need to remove an assert to access it (since sort's functionality was updated since the assert was added) and add missing dtype support to unique.
* This unlocks more GPU functionality with minimal code bloat. I also added CPU kernels for the dtypes for parity.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136114
Approved by: https://github.com/malfet
This PR is a replacement for https://github.com/pytorch/pytorch/pull/133085 for pushing a quick fix for RMSNorm.
The original author is @kkontny
Previous PR summary:
Since FP16 has quite small dynamic range it is very easy to overflow while computing `at::pow(input, 2)` , and it happens in real world computation.
I've tried to use `nn.RMSNorm` fused implementation instead of `LlamaRMSNorm` inside `transformers` implementation of Llama (`src/transformers/models/llama/modeling_llama.py`). It started to give wrong answers in Fp16 while still giving good in FP32. I figured out happens due to overflow while computing square of the input tensor.
Original `LLamaRMSNorm` implementation upcasts input to fp32 to prevent this and give better numerical stability.
```
class LlamaRMSNorm(nn.Module):
def __init__(self, hidden_size, eps=1e-6):
"""
LlamaRMSNorm is equivalent to T5LayerNorm
"""
super().__init__()
self.weight = nn.Parameter(torch.ones(hidden_size))
self.variance_epsilon = eps
def forward(self, hidden_states):
input_dtype = hidden_states.dtype
hidden_states = hidden_states.to(torch.float32)
variance = hidden_states.pow(2).mean(-1, keepdim=True)
hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
return self.weight * hidden_states.to(input_dtype)
```
Proposed commit fixed the issue. FP16 in RMSNorm has to be treated in special way, to be usable in real world implementations.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134106
Approved by: https://github.com/mikaylagawarecki, https://github.com/eqy
Notable changes:
1. Enable CudaGraph related tests
2. Fix UT problems
3. EXPERIMENTAL Navi31 support. User should enable Navi31 support with Env Var `TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1`
Know Problem:
1. `test/test_transformers.py` will massive failures and/or NaN outputs with `--use-pytest`
+ Update: Confirmed skip `class TestSDPAPrivateUse1Only` can fix the problem with `--use-pytest`
Note:
AOTriton 0.7b adds support to nestedtenosrs+SDPA but need more work (and consequently a separate PR) to enable it.
Fixes#133540
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134498
Approved by: https://github.com/pruthvistony, https://github.com/jeffdaily, https://github.com/malfet
reland of https://github.com/pytorch/pytorch/pull/133113
I have to create a new PR because the previous reverted PR could not either be rebased, or imported successfully :(
----
Moving DTensor to be in the public namespace, to formally add the documentation page that includes all the public APIs. This includes:
* many path renames and path import fixes
* a dedicated doc page without too much content yet (adding in the next PRs)
* To preserve the BC for users still using the torch.distributed._tensor, I added a shim script to redirect old path calls to the new module
The BC preserving is evidented by the fact that all DTensor tests are still working without changing the public imports. So it's safe to land the changes
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134203
Approved by: https://github.com/tianyu-l
Hi,
I noticed the `unfold` operator was missing on MaskedTensor.
I tested that my change works when calling unfold and backward on a `MaskedTensor` but I didn't find the tests for the dispatch of such operation. Where is it?
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125262
Approved by: https://github.com/cpuhrsch
This PR is slightly a revival / update to the discussion from https://github.com/pytorch/pytorch/pull/98960:
Part of FSDP2's tracing strategy right now is that:
(1) it is painful/difficult to handle the case where we have multiple graph input tensors that are aliased to each other and at least one of them is duplicated
(2) we already have longstanding in logic to remove duplicate input tensors from the graph in dynamo. Morally, FSDP2 gives us duplicate input tensors in the backward graph for every `unsharded_param`, because we have (a) the `unsharded_param` being closed over by the backward hook to resize/allgather, and (b) the same `unsharded_param` being saved for backward by autograd (we now guarantee in the partitioner that we will always save the base tensor for backward and recompute views)
(3) However, we were still seeing cases where the `unsharded_param` showed up twice in the backward graph inputs, as distinct tensor objects (with different python ids) instead of being true duplicates that dynamo can de-dup.
It turns on that this was because we were `.detach()`ing the `unsharded_param` in AOTDispatcher before plumbing it through the compiled forward (and so autograd would save a detach'd version of the `unsharded_param`). This is precisely because of the logic from https://github.com/pytorch/pytorch/pull/98960.
However, re-reading the detailed comments, it seems unnecessary to do a detach() on a graph input that is a (leaf) `nn.Parameter`, even if it happens to get no gradients in the backward. Since it is a leaf, we don't have to worry about the autograd engine "continuing to backprop through the graph beyond the current tensor" (the leaf has no other grad_fn for autograd to backprop through).
So this PR makes us a bit less aggressive about calling detach() on inputs: we only do it when:
(1) our graph input statically will get a `None` gradient (and also has no metadata mutations, the existing state)
(2) **and** our graph input is a non-leaf tensor (so detach()ing is actually required to prevent autograd from incorrectly backpropping past the non-leaf.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134193
Approved by: https://github.com/yf225
Co-authored-by: Will Feng <yf225@cornell.edu>
Summary:
In graph of TestXNNPACKQuantizer.test_dynamic_linear_with_con test, some quantized_decomposed.quantize_per_tensor.default ops are becoming quantized_decomposed.dequantize_per_tensor.tensor ops when using the new training ir.
This is because we lift params/buffers before calling make_fx. So previously, for the graph that’s passed to make_fx,`graph.L__self___linear1.weight` is a tensor
now in training ir, graph.L__self___linear1.weight is a FakeTensor. This caused the node overload to be different.
Test Plan:
```
buck2 run 'fbcode//mode/dev-nosan' fbcode//caffe2/test/quantization:test_quantization -- -r test_dynamic_linear_with_conv
```
Differential Revision: D61364547
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134525
Approved by: https://github.com/tugsbayasgalan, https://github.com/jerryzh168
Summary:
When exporting for training with `tolist`, we do not hit `FunctionalTensor.tolist` since we do not functionalize. Unfortunately, this means we hit `FakeTensor.tolist`, which creates unbacked symints that are not backed by proxies.
Rather than trying to patch up this low-level implementation, we replace it with essentially what `FunctionalTensor.tolist` does, which is higher-level: we essentially desugar to `item()` calls and let it take care of unbacked symints.
Test Plan:
Some expected failures are gone now.
Also found a test for `tolist` that was written when `FunctionalTensor.tolist` was implemented but not really doing much; repurposed it now to exercise more modes.
Differential Revision: D62197742
Pull Request resolved: https://github.com/pytorch/pytorch/pull/135131
Approved by: https://github.com/ezyang
Differential Revision: D61506212
Use `skipCUDAIf` from `torch.testing._internal.common_device_type` if we create the test class with `instantiate_device_type_tests`.
`instantiate_device_type_tests` would make sure the class has attr device_type, which works with`skipCUDAIf` from `torch.testing._internal.common_device_type`.
Also skipping test_vertical_pointwise_reduction_fusion for cpu test class, since the test expects cuda.
FAILED [0.0026s] test/inductor/test_unbacked_symints.py::TestUnbackedSymintsCPU::test_vertical_pointwise_reduction_fusion_cpu - AttributeError: 'TestUnbackedSymintsCPU' object has no attribute 'device'
repro:
```
CUDA_VISIBLE_DEVICES="" pytest test/inductor/test_unbacked_symints.py -k cpu -v
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133936
Approved by: https://github.com/ColinPeppler, https://github.com/desertfire
The issue:
Const propagation checks only if arguments do not have FakeTensor. If argument is Subclass, it will pass this condition.
As a result Const Propogation execution happens without FakeTensorMode and having tensor factories inside Subclass.__torch_dispatch__ results that this Tensor is not Fakified.
Solution:
If we have subclasses arguments, do not count that const propagation is doable
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134855
Approved by: https://github.com/zou3519
This essentially undoes large skips on everything but MacOS Sequoia to nn.modules made by https://github.com/pytorch/pytorch/pull/128393
Instead it uses existing `xfail`, but guards it on `_macos15_or_newer` boolean
Before the change if run on MacOS 14:
```
% python3 ../test/test_modules.py -v -k Hardswish 2>&1|tail -n3
Ran 57 tests in 0.053s
OK (skipped=32)
```
After
```
% python3 ../test/test_modules.py -v -k Hardswish 2>&1|tail -n3
Ran 57 tests in 0.229s
OK (skipped=10, expected failures=2)
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134858
Approved by: https://github.com/janeyx99
Hi,
I noticed the `unfold` operator was missing on MaskedTensor.
I tested that my change works when calling unfold and backward on a `MaskedTensor` but I didn't find the tests for the dispatch of such operation. Where is it?
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125262
Approved by: https://github.com/cpuhrsch
See #121528 for additional context.
In #120682, we moved the attention kernels from meta_registrations to fake_impls with the intent of fixing the device handling for seed/offset: these are typically on CPU. We needed to put the registrations in fake_impls to do this because meta_registrations doesn't have a way to specify device, whereas fake_impls does. But when we tried to actually fix the device types (#120839), we had to revert the PR because it broke cudagraph handling (during which seed/offset _are_ on CUDA).
Now, we want to put the registrations back in meta_registrations so that we can call these kernels with meta tensors. The use case is later in this stack - we want to be able to use the flop counter with these kernels.
Also - I specifically skip the `compare_tensor_meta()` check in test_fake / test_fake_autocast tests for the `_efficient_attention_forward` and `_flash_attention_forward` kernels, which fails because of the device mismatch from the seed/offset tensors. Then we can un-skip these opinfos. I verified that the efficient_attention_forward bug (#120842) is now caught by these opinfos if I revert the fix from this PR.
Differential Revision: [D61687369](https://our.internmc.facebook.com/intern/diff/D61687369)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134288
Approved by: https://github.com/drisspg
Which fixes BatchNorm behavior for if called with empty tensors on MPS backed. Removed `expectedFailureMPS` in test_nn.py, deleted expected failure in `test_mps.py` and adjusted `skipIfMPS` to `expectedFailureMPS` in BatchNorm2d OpInfo decorator, but restrict it only to the memory format tests
Test Plan: CI + `python3 -c "import torch; print(torch.nn.BatchNorm2d(3, device='mps')(torch.rand(0, 3, 2, 2, device='mps')))"`
Fixes https://github.com/pytorch/pytorch/issues/134423
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134540
Approved by: https://github.com/Skylion007, https://github.com/albanD
## Context
In some user Triton kernels, we have this set-up for whatever reason.
```
@triton.jit
def mykernel(
param0,
param1,
param2,
param3: tl.constexpr, # autotuned
param4, # non-constexpr
):
...
```
This is an edge case because it's a general practice to declare all constexprs params at the end.
And this will be an issue for AOTI because it fails to codegen all 4 params. That will surface as a device-side error: CUDA IMA, invalid argument...
```
> void* kernel_args_var_0[] = {&var_0, &var_1, &var_2};
---
< CUdeviceptr var_3;
< AOTI_TORCH_ERROR_CODE_CHECK(aoti_torch_get_data_ptr(buf0, reinterpret_cast<void**>(&var_3)));
< void* kernel_args_var_0[] = {&var_0, &var_1, &var_2, &var_3};
```
## Root-cause
* `kernel.constexpr` from the Kernel side-table contains the indices for all `constexpr` params that includes autotuned params.
* `raw_args`, that gets passed to wrapper codegen, excludes autotuned args.
* In the wrapper codegen, we try to find non-constexpr args using `kernel.constexpr` & `raw_args`. This is okay unless there's a `raw_arg` after an autotuned param in the function signature.
79b7fff188/torch/_inductor/codegen/cpp_wrapper_cuda.py (L118-L126)
## Fix
We try to fix this, by calculating the right constexprs wrt `raw_args`.
An illustration
```
raw_args: [arg0, arg1, arg2, arg4]
kernel.arg_names: [param0, param1, param2, param3, param4]
kernel.constexprs: [3] # param3 is autotuned; this is correct wrt kernel.arg_names
constexpr_indices: [] # this is correct wrt raw_args
```
Differential Revision: [D61831625](https://our.internmc.facebook.com/intern/diff/D61831625)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134520
Approved by: https://github.com/oulgen
This PR increases test coverage by including the tests in `test/test_nn.py` in the test suite of MPS.
Some of the tests are decorated with `@expectedFailureMPS` for various reasons. Either that the op is not implemented, or that the outputs do not align. Those tests that contain differing results should be investigated further to rule out any live bugs.
```bash
$ python test/run_test.py --mps --verbose -k TestNN
Running test batch 'tests to run' cost 84.76 seconds
```
Ref #133520
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134184
Approved by: https://github.com/albanD, https://github.com/malfet
I had a night mare rewriting tests in test_misc.py specifically :
1. graphs can have comments that refers to my files "/lsakka/.." we really dont care about comments add option to ignore comments.
2. empty lines added when EXPECTTEST_ACCEPT=1 are changed with linter causing tests to fail or linter fail!
add flag to ignore empty lines.
3. EXPECTTEST_ACCEPT fails when the text have some not readable characters. those should not effect comparing strings, also those causes weird diffs comments when tests fails. I removed ansi_escape chars https://github.com/pytorch/pytorch/pull/133045
this is used in
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134248
Approved by: https://github.com/aorenste
ghstack dependencies: #133639, #134364
Windows file path use `\` as delimiter, it is also a escape character. We need translate all path `\` to `/`. which like Linux.
Reproduce UT:
```cmd
pytest test\dynamo\test_higher_order_ops.py -v -k test_vmap_grad_vmap_guard_fail
```
Error msg:
```cmd
________________________________________________________________________________________________________ HigherOrderOpVmapGuardTests.test_vmap_grad_vmap_guard_fail _________________________________________________________________________________________________________
Traceback (most recent call last):
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\site-packages\torch\testing\_internal\logging_utils.py", line 89, in test_fn
fn(self, records)
File "D:\xu_git\dnnl_cb\pytorch\test\dynamo\test_higher_order_ops.py", line 2714, in test_vmap_grad_vmap_guard_fail
munge_exc(record.getMessage()),
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\site-packages\torch\testing\_internal\common_utils.py", line 5252, in munge_exc
s = re.sub(file, os.path.basename(file), s)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\re.py", line 209, in sub
return _compile(pattern, flags).sub(repl, string, count)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\re.py", line 303, in _compile
p = sre_compile.compile(pattern, flags)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\sre_compile.py", line 788, in compile
p = sre_parse.parse(p, flags)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\sre_parse.py", line 955, in parse
p = _parse_sub(source, state, flags & SRE_FLAG_VERBOSE, 0)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\sre_parse.py", line 444, in _parse_sub
itemsappend(_parse(source, state, verbose, nested + 1,
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\sre_parse.py", line 526, in _parse
code = _escape(source, this, state)
File "C:\Users\Xuhan\.conda\envs\win_mkl_static\lib\sre_parse.py", line 370, in _escape
raise source.error("incomplete escape %s" % escape, len(escape))
re.error: incomplete escape \x at position 2
To execute this test, run the following from the base repo dir:
python test\dynamo\test_higher_order_ops.py HigherOrderOpVmapGuardTests.test_vmap_grad_vmap_guard_fail
This message can be suppressed by setting PYTORCH_PRINT_REPRO_ON_FAILURE=0
--------------------------------------------------------------------------------------------------------------------------- Captured stdout call ----------------------------------------------------------------------------------------------------------------------------
frames [('total', 2), ('ok', 2)]
inductor []
inline_call []
stats [('calls_captured', 38), ('unique_graphs', 2)]
--------------------------------------------------------------------------------------------------------------------------- Captured stderr call ----------------------------------------------------------------------------------------------------------------------------
V0824 01:29:00.148000 27840 torch\_dynamo\guards.py:2787] [0/1] [__recompiles] Recompiling function fn in D:\xu_git\dnnl_cb\pytorch\test\dynamo\test_higher_order_ops.py:2699
V0824 01:29:00.148000 27840 torch\_dynamo\guards.py:2787] [0/1] [__recompiles] triggered by the following guard failure(s):
V0824 01:29:00.148000 27840 torch\_dynamo\guards.py:2787] [0/1] [__recompiles] - 0/0: torch._functorch.pyfunctorch.compare_functorch_state([('Vmap', 1, 'error')]) # _dynamo\output_graph.py:479 in init_ambient_guards
========================================================================================================================== short test summary info ==========================================================================================================================
FAILED [0.7452s] test/dynamo/test_higher_order_ops.py::HigherOrderOpVmapGuardTests::test_vmap_grad_vmap_guard_fail - re.error: incomplete escape \x at position 2
```
Local test passed:
<img width="860" alt="image" src="https://github.com/user-attachments/assets/90f0d780-0639-4c03-8d7c-6f227c93a3fc">
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134348
Approved by: https://github.com/jansel
This PR switches to cuDSS library and has the same purpose of #127692, which is to add Sparse CSR tensor support to linalg.solve.
Fixes#69538
Minimum example of usage:
```
import torch
if __name__ == '__main__':
spd = torch.rand(4, 3)
A = spd.T @ spd
b = torch.rand(3).to(torch.float64).cuda()
A = A.to_sparse_csr().to(torch.float64).cuda()
x = torch.linalg.solve(A, b)
print((A @ x - b).norm())
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/129856
Approved by: https://github.com/amjames, https://github.com/lezcano, https://github.com/huydhn
Co-authored-by: Zihang Fang <zhfang1108@gmail.com>
Co-authored-by: Huy Do <huydhn@gmail.com>
As you can see, 'privateuse1' appears many times in out-of-tree extension codebase. I think that everything about the device type should be as same as other in-tree backends after registering the privateuse1 backend.
For example, after registering a privateuse1 backend named "foo", you should allow "foo" to be passed in as a valid device type.
```diff
- instantiate_device_type_tests(TestIndexing, globals(), only_for='privateuse1')
- instantiate_device_type_tests(NumpyTests, globals(), only_for='privateuse1')
+ instantiate_device_type_tests(TestIndexing, globals(), only_for='foo')
+ instantiate_device_type_tests(NumpyTests, globals(), only_for='foo')
```
> https://github.com/Ascend/pytorch/blob/master/test/test_indexing.py#L1654-L1655
The change is to map privateuse1 backend name to 'privateuse1' when calling `filter_desired_device_types()`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133082
Approved by: https://github.com/albanD
Summary: Defaulting TORCH_NCCL_DUMP_ON_TIMEOUT to "true" and adding a kilswitch in case we need to kill this feature in production.
Test Plan: Tests pass manually but need futher testing before this is rolled out fully everywhere.
Differential Revision: D61136320
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133237
Approved by: https://github.com/c00w
# UPDATE:
This is take 3 of https://github.com/pytorch/pytorch/pull/131863 which was landed via co dev but not applying correclty
# Summary
Changes the stance of SDPA on what to do for fully masked out rows
## Current Behavior
Several PyTorch users have expressed frustration over this issue:
- https://github.com/pytorch/pytorch/issues/41508
- https://github.com/pytorch/pytorch/issues/103749
- https://github.com/pytorch/pytorch/issues/103963
These are significant issues with extensive discussion but no satisfactory resolution. The PyTorch team's consensus, as stated here:
https://github.com/pytorch/pytorch/issues/24816#issuecomment-524415617
Can be paraphrased as follows:
When passing in fully masked out rows, attention becomes ambiguous. We have two main options:
1. Uniformly attend to all values:
```python
scores[masked_out_rows] = 1 / len(row)
out[masked_out_rows] = 1 / len(row) * value
```
2. Decide that attention between no queries (masked) and no keys (masked) is meaningless:
```python
output[fully_masked_rows] = NaN
```
We went with option 2. Partially because it was easier to implement, but also people argued that users can slice the output to remove the NaNs:
``` Python
>fill_value = -float("inf")
>row0 = torch.randn(4)
>row1 = torch.tensor([(fill_value for _ in range(4)])
>matrix = torch.stack([row0, row1]).requires_grad_(True)
>out = torch.softmax(matrix, 1)
>out = out[0]
>print(out)
tensor([0.5377, 0.2729, 0.0692, 0.1201])
```
Cool, problem solved. But what happends when you call backwards..
```Python
>out.backward(torch.ones_like(out))
>print(matrix.grad)
tensor([[3.0957e-08, 1.4157e-08, 7.7802e-10, 1.3713e-08],
[ nan, nan, nan, nan]])
```
Those pesky NaNs are back!
## Why do we see NaNs today?
The core of the problem revolves around using softmax function in sdpa:
```python
> row = torch.tensor([(-float("inf")) for _ in range(4)])
> torch.softmax(row, 0)
tensor([nan, nan, nan, nan])
```
## Quick Aside: Masking in Attention
Attention itself doesn't have a concept of masking. The `sdpa` function has an argument called `attn_mask`, which would be more accurately named `attn_bias`. This is because we don't actually "mask" entries when computing attention. Instead, due to implementation details([performance](https://github.com/pytorch/pytorch/issues/25110#issuecomment-524519087)), we add a value to the masked-out query/key pairs.
We use a large negative number (typically -inf) to decrease the attention weight, as softmax assigns more weight to larger values.
## Alternative Approaches
If we use a very large negative number instead of -inf:
```python
> row = torch.tensor([(-1e6) for _ in range(4)])
> torch.softmax(row, 0)
tensor([0.2500, 0.2500, 0.2500, 0.2500])
```
However if users always remembered to "slice" out their outputs i.e.:
```Python
>fill_value = -1e6
>...
>out.backward(torch.ones_like(out))
>print(matrix.grad)
tensor([[-0.0563, -0.0564, 0.1613, -0.0486],
[ 0.0000, 0.0000, 0.0000, 0.0000]])
```
This would bring us back into a better state.
## A Third Option
We don't necessarily need to alter the behavior of softmax for -inf or very large negative numbers. The fundamental goal is to exclude certain query/key pairs from attention, regardless of the underlying implementation.
This PR implements the new semantic for masking w/ attention in fully masked-out rows:
```python
out[masked_out_rows] = 0
```
**Important Note**: This idea isn't entirely new. The [MaskedTensor](https://pytorch.org/tutorials/prototype/maskedtensor_overview#safe-softmax) prototype, a tensor subclass, was designed to handle such cases. However, it remains a prototype feature and hasn't gained widespread adoption.
## Details
This PR stack does 3 things:
1. Adds a PRIVATE _safe_softmax op
2. Updates semantic for flash_cpu fused kernel
3. Updates semantic for efficient_cuda fused kernel
_safe_softmax is not supposed to be used generically and is only meant to be used within the context of SDPA. Due to this fact instead of decomposing softmax and checking for -inf rows we instead "cheat" and use nan_to_num.
Why I think this is okay? (please find a counter point if avail)
There are multiple ways NaNs can emerge. For the fully masked out rows case nan_to_num works. But what if there were other NaNs, wouldn't this silently remove them?
The only case that this can happen is if the input itself had a NaN or an Inf
For example:
```Python
a = torch.ones([4], requires_grad=False, dtype=torch.float16)
a[1] = torch.finfo(torch.float16).max
print(a.softmax(-1))
```
Will return
`tensor([0., 1., 0., 0.], dtype=torch.float16)`
Where
```Python
a = torch.ones([4], requires_grad=False, dtype=torch.float16)
a[1] = float("inf")
a.softmax(-1)
```
returns:
`tensor([nan, nan, nan, nan], dtype=torch.float16)`
If we dont want to even allow for the possibility of "inf" or "NaN" attention scores to be converted to 0 then we can implemented it something like this
```Python
max = torch.max(a, dim=-1, keepdim=True)
exp = torch.exp(a - max.values)
denom = torch.sum(exp, dim=-1, keepdim=True)
softmax = exp / denom
softmax = torch.where(max.values == float('-inf'), 0.0, softmax)
```
however we would be paying for this in math performance.
## Why Now
I think one point that has substantially changed where PyTorch should lie on this argument is the fact that we have fused implementations for SDPA now. And these fused implementations allow us to easily and performantly support this new semantic.
Differential Revision: [D61418679](https://our.internmc.facebook.com/intern/diff/D61418679)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133882
Approved by: https://github.com/soulitzer
Moving DTensor to be in the public namespace, to formally add the
documentation page that includes all the public APIs. This includes:
* many path renames and path import fixes
* a dedicated doc page without too much content yet (adding in the next
PRs)
* To preserve the BC for users still using the `torch.distributed._tensor`,
I added a shim script to redirect old path calls to the new module
The BC preserving is evidented by the fact that all DTensor tests are still
working without changing the public imports. So it's safe to land the
changes
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133113
Approved by: https://github.com/XilunWu
ghstack dependencies: #133305, #133306
This PR adds the foreach impl for Adafactor knowing that there are many ways to improve its runtime perf today (by adding more foreach support). After this PR:
- we have a foreach flag for Adafactor
- It is NOT the default. Why not? It is only slightly faster + uses O(n) more memory where n is the number of params in your max param group. People tend to use Adafactor for memory efficiency.
Next steps:
- make torch.compile possible on it
- make it faster (by adding more foreach apis)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/132336
Approved by: https://github.com/albanD
ghstack dependencies: #133360
# Summary
Changes the stance of SDPA on what to do for fully masked out rows
## Current Behavior
Several PyTorch users have expressed frustration over this issue:
- https://github.com/pytorch/pytorch/issues/41508
- https://github.com/pytorch/pytorch/issues/103749
- https://github.com/pytorch/pytorch/issues/103963
These are significant issues with extensive discussion but no satisfactory resolution. The PyTorch team's consensus, as stated here:
https://github.com/pytorch/pytorch/issues/24816#issuecomment-524415617
Can be paraphrased as follows:
When passing in fully masked out rows, attention becomes ambiguous. We have two main options:
1. Uniformly attend to all values:
```python
scores[masked_out_rows] = 1 / len(row)
out[masked_out_rows] = 1 / len(row) * value
```
2. Decide that attention between no queries (masked) and no keys (masked) is meaningless:
```python
output[fully_masked_rows] = NaN
```
We went with option 2. Partially because it was easier to implement, but also people argued that users can slice the output to remove the NaNs:
``` Python
>fill_value = -float("inf")
>row0 = torch.randn(4)
>row1 = torch.tensor([(fill_value for _ in range(4)])
>matrix = torch.stack([row0, row1]).requires_grad_(True)
>out = torch.softmax(matrix, 1)
>out = out[0]
>print(out)
tensor([0.5377, 0.2729, 0.0692, 0.1201])
```
Cool, problem solved. But what happends when you call backwards..
```Python
>out.backward(torch.ones_like(out))
>print(matrix.grad)
tensor([[3.0957e-08, 1.4157e-08, 7.7802e-10, 1.3713e-08],
[ nan, nan, nan, nan]])
```
Those pesky NaNs are back!
## Why do we see NaNs today?
The core of the problem revolves around using softmax function in sdpa:
```python
> row = torch.tensor([(-float("inf")) for _ in range(4)])
> torch.softmax(row, 0)
tensor([nan, nan, nan, nan])
```
## Quick Aside: Masking in Attention
Attention itself doesn't have a concept of masking. The `sdpa` function has an argument called `attn_mask`, which would be more accurately named `attn_bias`. This is because we don't actually "mask" entries when computing attention. Instead, due to implementation details([performance](https://github.com/pytorch/pytorch/issues/25110#issuecomment-524519087)), we add a value to the masked-out query/key pairs.
We use a large negative number (typically -inf) to decrease the attention weight, as softmax assigns more weight to larger values.
## Alternative Approaches
If we use a very large negative number instead of -inf:
```python
> row = torch.tensor([(-1e6) for _ in range(4)])
> torch.softmax(row, 0)
tensor([0.2500, 0.2500, 0.2500, 0.2500])
```
However if users always remembered to "slice" out their outputs i.e.:
```Python
>fill_value = -1e6
>...
>out.backward(torch.ones_like(out))
>print(matrix.grad)
tensor([[-0.0563, -0.0564, 0.1613, -0.0486],
[ 0.0000, 0.0000, 0.0000, 0.0000]])
```
This would bring us back into a better state.
## A Third Option
We don't necessarily need to alter the behavior of softmax for -inf or very large negative numbers. The fundamental goal is to exclude certain query/key pairs from attention, regardless of the underlying implementation.
This PR implements the new semantic for masking w/ attention in fully masked-out rows:
```python
out[masked_out_rows] = 0
```
**Important Note**: This idea isn't entirely new. The [MaskedTensor](https://pytorch.org/tutorials/prototype/maskedtensor_overview#safe-softmax) prototype, a tensor subclass, was designed to handle such cases. However, it remains a prototype feature and hasn't gained widespread adoption.
## Details
This PR stack does 3 things:
1. Adds a PRIVATE _safe_softmax op
2. Updates semantic for flash_cpu fused kernel
3. Updates semantic for efficient_cuda fused kernel
_safe_softmax is not supposed to be used generically and is only meant to be used within the context of SDPA. Due to this fact instead of decomposing softmax and checking for -inf rows we instead "cheat" and use nan_to_num.
Why I think this is okay? (please find a counter point if avail)
There are multiple ways NaNs can emerge. For the fully masked out rows case nan_to_num works. But what if there were other NaNs, wouldn't this silently remove them?
The only case that this can happen is if the input itself had a NaN or an Inf
For example:
```Python
a = torch.ones([4], requires_grad=False, dtype=torch.float16)
a[1] = torch.finfo(torch.float16).max
print(a.softmax(-1))
```
Will return
`tensor([0., 1., 0., 0.], dtype=torch.float16)`
Where
```Python
a = torch.ones([4], requires_grad=False, dtype=torch.float16)
a[1] = float("inf")
a.softmax(-1)
```
returns:
`tensor([nan, nan, nan, nan], dtype=torch.float16)`
If we dont want to even allow for the possibility of "inf" or "NaN" attention scores to be converted to 0 then we can implemented it something like this
```Python
max = torch.max(a, dim=-1, keepdim=True)
exp = torch.exp(a - max.values)
denom = torch.sum(exp, dim=-1, keepdim=True)
softmax = exp / denom
softmax = torch.where(max.values == float('-inf'), 0.0, softmax)
```
however we would be paying for this in math performance.
## Why Now
I think one point that has substantially changed where PyTorch should lie on this argument is the fact that we have fused implementations for SDPA now. And these fused implementations allow us to easily and performantly support this new semantic.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/131060
Approved by: https://github.com/jbschlosser
This PR fixes flaky internal tests:
- The AutoHeuristic test was sometimes failing because it required autotuning to happen for mixed_mm which didn't end up happening when there was a fx graph cache hit.
- The tests inside pattern_matcher failed because in some cases pad_mm decided to pad which made the mixed_mm pattern not match anymore (instead of cast -> mm, it was cast -> pad -> mm), and the tests also fail when is_big_gpu is false (which I haven't found an explanation for).
Differential Revision: [D60972176](https://our.internmc.facebook.com/intern/diff/D60972176)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133015
Approved by: https://github.com/Chillee, https://github.com/eellison
