The big idea is to add `create_unbacked_symfloat` and `create_unbacked_symint` to ShapeEnv, allowing you to allocate symbolic floats/ints corresponding to data you don't know about at compile time. Then, instead of immediately erroring out when you try to call local_scalar_dense on a FakeTensor, we instead create a fresh symint/symfloat and return that.
There a bunch of odds and ends that need to be handled:
* A number of `numel` calls converted to `sym_numel`
* When we finally return from item(), we need to ensure we actually produce a SymInt/SymFloat when appropriate. The previous binding code assumed that you would have to get a normal Python item. I add a pybind11 binding for Scalar (to PyObject only) and refactor the code to use that. There is some trickiness where you are NOT allowed to go through c10::SymInt if there isn't actually any SymInt involved. See comment.
* One of our unit tests tripped an implicit data dependent access which occurs when you pass a Tensor as an argument to a sizes parameter. This is also converted to support symbolic shapes
* We now support tracking bare SymInt/SymFloat returns in proxy tensor mode (this was already in symbolic-shapes branch)
* Whenever we allocate an unbacked symint, we record the stack trace it was allocated at. These get printed when you attempt data dependent access on the symint (e.g., you try to guard on it)
* Subtlety: unbacked symints are not necessarily > 1. I added a test for this.
These unbacked symints are not very useful right now as you will almost always immediately raise an error later when you try to guard on them. The next logical step is adding an assertion refinement system that lets ShapeEnv learn facts about unbacked symints so it can do a better job eliding guards that are unnecessary.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90624
Approved by: https://github.com/Skylion007, https://github.com/voznesenskym
* Skip a unittest that needs FFT if not built with FFT
* Mark a test with "slow": `python test/test_ops.py -k TestCompositeComplianceCUDA.test_forward_ad_svd_lowrank_cuda_float32` took >5min on my machine.
* Skip a flaky test that's marked "expectedFailure", similar to #90233
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90609
Approved by: https://github.com/soumith
This lowers the `reduce_dtype` retrieval to the `handle` instead of the `state` in preparation for `fully_shard`, and this adds a guard to avoid a no-op `to()` call.
Note that this change pretty much gets overridden in following PRs.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90615
Approved by: https://github.com/rohan-varma
Use register_state_dict_pre_hook in FSDP to simplify state_dict implementations & remove hacks. This removes `def state_dict` entirely and paves the path for composable API as well.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90436
Approved by: https://github.com/fegin
This saves a data structure `_stream_to_name: Dict[torch.cuda.Stream, str]` that maps each FSDP stream to its name. This can help in debugging by checking `_stream_to_name[torch.cuda.current_stream()]` to see if it is `"default"` or `"unshard"` in the post-backward hook for example.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90611
Approved by: https://github.com/rohan-varma
Optimizes the nccl python bindings to reserve space when converting PythonObject* into Tensors. This should reduce the number of unnecessary allocations in the nccl bindings as the std::vector grows.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88203
Approved by: https://github.com/ezyang
Instead of inferring shape mappings from a bunch of data structures that were plumbed in InstructionTranslator, we instead work out mappings by just iterating over the GraphArgs and mapping symbols to arguments as they show up. If multiple argument sizes/strides/offset map to the same symbol, this means they are duck sized, so we also generate extra equality tests that they must be equal. Finally, we generate 0/1 specialization guards. The resulting code is much shorter, and I think also easier to understand.
TODO: Delete all the tensor ref tracking code, it's unnecessary
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90528
Approved by: https://github.com/voznesenskym
So, uh, I have a new strategy for generating dupe guards, one where I don't actually need to allocate symints for every tensor that is fakeified. So I'm reverting the changes I made from earlier PRs in this one.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90381
Approved by: https://github.com/voznesenskym
Wow, I had to sweat so much to get this PR out lol.
This PR enforces the invariant that whenever we allocate SymInts as part of fakeification, the SymInt is associated with a Source, and in fact we store the string source name on SymbolWithSourceName. We use 'sname' as the shorthand for source name, as 'name' is already used by sympy to name symbols.
In order to store source names, we have to plumb source names from Dynamo to PyTorch. This made doing this PR a bit bone crushing, because there are many points in the Dynamo codebase where we are improperly converting intermediate tensors into fake tensors, where there is no source (and there cannot be, because it's a frickin' intermediate tensor). I've fixed all of the really awful cases in earlier PRs in the stack. This PR is just plumbing in source names from places where we do have it.
Signed-off-by: Edward Z. Yang <ezyang@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90295
Approved by: https://github.com/voznesenskym
Summary:
This pull request makes some LazyGraphExecutor private data structures protected such that XLAGraphExecutor can reuse them.
Here is the list:
1. DeviceLocker.
2. DeviceLockerArena.
3. DataCacheArena. In addition, it also introduces LazyGraphExecutor::ResetTrimCounter() such that XLAGraphExecutor can reuse the trim counter.
Test Plan:
CI.
P.S. This is to re-land #90457.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90598
Approved by: https://github.com/JackCaoG
This adds a d3-based interactive visualization for exploring the memory
allocation traces that the caching allocator can capture. This visualization
code can also be attached to kineto trace information in the future to also
provide visualization for the memory events captured there, which come with
addition information about the graph.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90348
Approved by: https://github.com/robieta
Summary: Modified replace_pattern in the subgraph rewriter to return a list of pairs of matches along with their corresponding replacement nodes in the modified graph (`List[Tuple[Match, List[Node]]]`). This allows us to easily modify the replaced nodes, including setting the metadata.
Test Plan: CI
Differential Revision: D41737056
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90244
Approved by: https://github.com/SherlockNoMad
Summary:
Optimize the shape padding in the following perspectives:
- Add BFloat16 support for AMP training and Float16 support for inference
- Optimize microbenchmark to avoid peak memory issue, and include profiling memory ops to make more accurate decision
- Add a flag to turn off/on padding dims N and M in `torch.bmm` due to expensive memory copy of `.contiguous` to avoid peak memory issues in internal models
Test Plan: CI
Differential Revision: D41724868
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90425
Approved by: https://github.com/jianyuh
Continuation after https://github.com/pytorch/pytorch/pull/90163.
Here is a script I used to find all the non-existing arguments in the docstrings (the script can give false positives in presence of *args/**kwargs or decorators):
_Edit:_
I've realized that the indentation is wrong for the last `break` in the script, so the script only gives output for a function if the first docstring argument is wrong. I'll create a separate PR if I find more issues with corrected script.
``` python
import ast
import os
import docstring_parser
for root, dirs, files in os.walk('.'):
for name in files:
if root.startswith("./.git/") or root.startswith("./third_party/"):
continue
if name.endswith(".py"):
full_name = os.path.join(root, name)
with open(full_name, "r") as source:
tree = ast.parse(source.read())
for node in ast.walk(tree):
if isinstance(node, ast.FunctionDef):
all_node_args = node.args.args
if node.args.vararg is not None:
all_node_args.append(node.args.vararg)
if node.args.kwarg is not None:
all_node_args.append(node.args.kwarg)
if node.args.posonlyargs is not None:
all_node_args.extend(node.args.posonlyargs)
if node.args.kwonlyargs is not None:
all_node_args.extend(node.args.kwonlyargs)
args = [a.arg for a in all_node_args]
docstring = docstring_parser.parse(ast.get_docstring(node))
doc_args = [a.arg_name for a in docstring.params]
clean_doc_args = []
for a in doc_args:
clean_a = ""
for c in a.split()[0]:
if c.isalnum() or c == '_':
clean_a += c
if clean_a:
clean_doc_args.append(clean_a)
doc_args = clean_doc_args
for a in doc_args:
if a not in args:
print(full_name, node.lineno, args, doc_args)
break
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90505
Approved by: https://github.com/malfet, https://github.com/ZainRizvi
Fixes#88074
Several datapipes have their lengths cached on being executed for the first time. However, source datapipes might change in length (most prominently, whenever `apply_sharding` is called). The behaviour is counter-intuitive because we do not expect `__len__` to have side-effects.
This PR makes `__len__` dynamically computed.
Changes:
- Add note to the `datapipes` README that `__len__` should be dynamic and why.
- Remove caching of length computations in `ConcaterIterDataPipe`, `MultiplexerIterDataPipe`, `ZipperIterDataPipe`, `BatcherIterDataPipe`, `ConcaterMapDataPipe`, and `BatcherMapDataPipe`.
- This required removal of the `length` attribute in setstate/getstate of `MultiplexerIterDataPipe`. I am unsure whether to remove this completely and risk breaking saved checkpoints (as I did) or whether to just ignore the `length` of the loaded `state`.
- This also means the classes above no longer have a `length` attribute. I have found no uses of this, though.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88302
Approved by: https://github.com/NivekT
Summary:
This pull request makes some LazyGraphExecutor private data structures protected such that XLAGraphExecutor can reuse them.
Here is the list:
1. DeviceLocker.
2. DeviceLockerArena.
3. DataCacheArena.
In addition, it also introduces LazyGraphExecutor::ResetTrimCounter() such that XLAGraphExecutor can reuse the trim counter.
Test Plan:
CI.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/90457
Approved by: https://github.com/JackCaoG
- Adds `log_level` to aot's config
- Outputs log to `<graph_name>_<log_level>.log` in aot_torchinductor subfolder of the debug directory
- Modifies the Inductor debug context to use the graph name when naming the folder instead of the os pid
- Adds `TORCH_COMPILE_DEBUG` flag to enable it, (as well as separate dynamo and inductor logs)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88987
Approved by: https://github.com/Chillee
