This is a resubmission of my previous PR that I accidentally deleted, apology in advance if any inconvenience caused. Below are details of this PR.
Fix an issue when torch.addmv behaves inconsistent between torch.compile mode and eager mode. Here is the code to reproduce:
```
import torch
import numpy as np
@torch.compile
def test_optimized(input, mat, vec):
return torch.addmv(input, mat, vec)
def test(input, mat, vec):
return torch.addmv(input, mat, vec)
input = torch.tensor([2], dtype=torch.int32)
mat = torch.tensor(np.random.randn(0, 0), dtype=torch.int32)
vec = torch.tensor([])
origin_out = test(input, mat, vec)
optimized_out = test_optimized(input, mat, vec)
print(origin_out) # tensor([2.])
print(optimized_out) # tensor([])
```
According to the equation (https://pytorch.org/docs/stable/generated/torch.addmv.html), when matrix and vector is empty, returning `[2.]` seems more reasonable to me.
Following the cpu implementation of this API:e97b97af56/aten/src/ATen/native/Blas.cpp (L62)
I add an additional branch to handle empty matrix
Pull Request resolved: https://github.com/pytorch/pytorch/pull/143792
Approved by: https://github.com/jgong5, https://github.com/leslie-fang-intel, https://github.com/jansel
Fix: #141251
This PR adds a few static guard checks when decomposing and lowering the `slice`
operation, so that we avoid adding unnecessary guards. Specifically, when clamping the end
values.
In summary, the changes are:
- `slice` dynamo decomposition: checks `end >= sizes[dim]` statically. If we don't know
that, the following guard ensures that we (don't) need clamping.
- `evaluate_min` inductor `sizevar` function: checks whether we can solve it statically or
not, before actually creating a new guard.
The latter had to be changed because `evaluate_min` (called by `ir.SliceView` constructor)
would always try to create a guard based on the hints operation result. However, if both
`left` and `right` hints were true, it would default to `left <= right` guard. By checking
the guards statically before, we can avoid that.
```python
N = 16
@torch.compile(backend="inductor", dynamic=False, fullgraph=True)
def fn(x):
splits = torch.ops.aten.split.Tensor(x, N)
first = splits[0]
return torch.ops.aten.slice.Tensor(first, 0, 0, N)
x = torch.arange(N)
torch._dynamo.mark_dynamic(x, 0)
fn(x)
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/142372
Approved by: https://github.com/ezyang
Fixes https://github.com/pytorch/pytorch/issues/141332
`F.logsigmoid` will return two outputs: `output` and `buffer`.
For `F.logsigmoid` cpu path, it will use buffer to store some intermediate values and use them when computing gradients, so it returns a `buffer` tensor with nonzero size. For cuda and xpu paths, buffer is useless, so the `buffer ` tensor size of xpu `F.logsigmoid` will be zero, just like cuda. The root cause of the issue is that the codes in `decompositions.py` (ref:https://github.com/pytorch/pytorch/blob/main/torch/_decomp/decompositions.py#L2803) only handle the cuda cases, when the a fake tensor with device is xpu run to here, it will use the cpu path and return a `buffer` with nonzero size, which is conflict to the implementation of intel xpu concrete tensor. Therefore this pr add conditions to handle xpu cases. Make sure the two returned buffer sizes match each other.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141333
Approved by: https://github.com/guangyey, https://github.com/EikanWang, https://github.com/ezyang
For better tracking, we need to make maybe aliasing/mutating ops with proper tag. We need to special case native_batch_norm because it is not a CIA but has a wrong schema. I guess native_batch_norm will be removed at some point, so until then we just keep it around.
D60347117
Pull Request resolved: https://github.com/pytorch/pytorch/pull/131990
Approved by: https://github.com/bdhirsh
Previously the split decomp would return the input when there were no splits. this errors in torch.compile (or FakeTensorMode) with :
> RuntimeError: View operation returned a tensor that is the same as the input base tensor. This is no longer allowed; you must explicitly create a new tensor (e.g., using .detach()). As a user, you could have made a mistake implementing __torch_dispatch__ or a Python operator decomposition or meta registration; if that's not the case, please report a bug to PyTorch or the backend you are using.
Fix for https://github.com/pytorch/pytorch/issues/133394
Differential Revision: [D65635070](https://our.internmc.facebook.com/intern/diff/D65635070)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140065
Approved by: https://github.com/bdhirsh
Previously the split decomp would return the input when there were no splits. this errors in torch.compile (or FakeTensorMode) with :
> RuntimeError: View operation returned a tensor that is the same as the input base tensor. This is no longer allowed; you must explicitly create a new tensor (e.g., using .detach()). As a user, you could have made a mistake implementing __torch_dispatch__ or a Python operator decomposition or meta registration; if that's not the case, please report a bug to PyTorch or the backend you are using.
Fix for https://github.com/pytorch/pytorch/issues/133394
Differential Revision: [D65635070](https://our.internmc.facebook.com/intern/diff/D65635070)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140065
Approved by: https://github.com/bdhirsh
**Summary**
In the case of LLaMA2, for a linear operation with an activation size of `(4, 1, 4096)` and a stride of `(4096, 128, 1)` which has been decomposed into `matmul`. And the decomposition of `matmul` results in `bmm` due to a strict continuity check. We can align the continuity check with ATen by skip dim of size 1 to enable decomposition into `mm` instead.
**Test Plan**
```
python -u -m pytest -s -v test/inductor/test_mkldnn_pattern_matcher.py -k test_linear_input_non_contiguous_3D_wo_bias
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139172
Approved by: https://github.com/jgong5, https://github.com/ezyang
In this PR, we implement lazy dictionary for export decomp behaviour for following reasons:
1. Custom op loading can happen after import time, as a result, the decomp table might not be able to pick up the decomp. Therefore we try to delay materialization as late as possible.
I intentionally seperated out the core_aten_decomp to not have any custom CIA ops in this PR to mitigate the risk of getting reverted but in the future, core_aten_decomp under torch/_decomp will exist as an alias to official export table (torch.export.default_decompositions)
Differential Revision: [D64140807](https://our.internmc.facebook.com/intern/diff/D64140807)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137650
Approved by: https://github.com/justinchuby, https://github.com/bdhirsh
Summary:
# context
* enable the `_get_user_embeddings` function
* run failed at P1610151892
```
torch._dynamo.exc.BackendCompilerFailed: backend='inductor' raised:
GuardOnDataDependentSymNode: Could not guard on data-dependent expression u22 <= 0 (unhinted: u22 <= 0). (Size-like symbols: u22)
ATTENTION: guard_size_oblivious would fix the error, evaluating expression to False.
Maybe you need to add guard_size_oblivious to framework code, see doc below for more guidance.
Potential framework code culprit (scroll up for full backtrace):
File "/data/users/hhy/fbsource/buck-out/v2/gen/fbcode/38472faba4e3e6c1/aps_models/ads/icvr/__icvr_launcher_live__/icvr_launcher_live#link-tree/torch/_decomp/decompositions.py", line 1692, in native_layer_norm_backward
if M <= 0 or N <= 0:
```
```
N = prod(inner_dims) # type: ignore[arg-type]
M = prod(outer_dims) # type: ignore[arg-type]
if M <= 0 or N <= 0:
return (
input.new_zeros(input_shape) if output_mask[0] else None,
input.new_zeros(input_shape[axis:]) if output_mask[1] else None,
input.new_zeros(input_shape[axis:]) if output_mask[2] else None,
)
```
# changes
* use guard_size_oblivious since the new_zeros return is kind of optimization, shouldn't impact the correctness of the follow up code logic.
* the size `ret[i][j]` could be zero, so the change in V1 isn't valid
* for more details: [post](https://fb.workplace.com/groups/6829516587176185/permalink/8003616173099548/)
```
from torch.fx.experimental.symbolic_shapes import guard_size_oblivious
if guard_size_oblivious(M <= 0) or guard_size_oblivious(N <= 0):
```
# past
* found `u22` was introduced at
```
def _wait_impl(self) -> List[List[int]]:
# Can not use is_torchdynamo_compiling(), as every such condition should be independent for compilation with graph breaks.
if isinstance(self._splits_awaitable, dist.Work):
self._splits_awaitable.wait()
ret = self._output_tensor.view(self.num_workers, -1).T.tolist() # <------ u22 introduced here
if not torch.jit.is_scripting() and is_torchdynamo_compiling():
for i in range(len(ret)):
for j in range(len(ret[i])):
torch._check_is_size(ret[i][j]) # <---------- my question: why the _check_is_size isn't enough??
torch._check(ret[i][j] > 0) # <------ added by diff V1
```
Test Plan:
# run command
```
TORCH_SHOW_CPP_STACKTRACES=1 TORCHDYNAMO_EXTENDED_DEBUG_CPP=1 TORCH_LOGS="+graph_code,output_code,dynamic,aot,guards,verbose_guards,recompiles,graph_breaks" TORCH_TRACE=/var/tmp/tt buck2 run fbcode//mode/opt fbcode//aps_models/ads/icvr:icvr_launcher_live -- mode=fmc/local_ig_fm_v4_mini training.pipeline_type=pt2 2>&1 | tee -a `tagT`.`tagH`.log
```
# results
* before
**without enabling `_get_user_embeddings`**
[14 Failures and Restarts](https://interncache-all.fbcdn.net/manifold/tlparse_reports/tree/logs/.tmp2eNI7p/failures_and_restarts.html)
log: P1610151892
{F1889387940}
* V1
enable `_get_user_embeddings`
with `torch._check(ret[i][j] > 0)`
[13 Failures and Restarts](https://interncache-all.fbcdn.net/manifold/tlparse_reports/tree/logs/.tmp6J1iY9/failures_and_restarts.html)
{F1889388378}
* V2
enable `_get_user_embeddings`
with `if guard_size_oblivious(M <= 0) or guard_size_oblivious(N <= 0):`
[tlparse](https://interncache-all.fbcdn.net/manifold/tlparse_reports/tree/logs/.tmpFhZZyC/index.html)
if guard_size_oblivious(M <= 0) or guard_size_oblivious(N <= 0):
Differential Revision: D63424929
Pull Request resolved: https://github.com/pytorch/pytorch/pull/136798
Approved by: https://github.com/ezyang
