Add check to verify if local shards in FSDP have tensors before accessing #7

syedshahbaaz · 2025-06-30T15:50:40Z

Fixes #PYTORCHDGQ-6374. The existing test (TestFSDPOptimState.test_interface_arguments) accesses local_shards[0] in FSDP for all ranks by default. Based on the number of ranks and data, not all ranks will have local shards with tensor every time. Accessing empty local_shards throws a list index out of range error. Adding a check before accessing elements resolves this for all cases.

daisyden · 2025-07-14T05:41:00Z

test/distributed/fsdp/test_fsdp_optim_state.py

                        continue
                    self.assertTrue(isinstance(s, ShardedTensor))
-                    if s._local_shards[0]:
+                    if s._local_shards:


You could change

if s._local_shards:

into

if len(s._local_shards) > 0:

@daisyden updated

For tensor with non-zero offset, it must be multiplied by element size Add regression test by creating Tensor in array of 6 elements with offset 3, which before the fix crashed with ``` C++ exception with description "setStorage: sizes [3, 3], strides [0, 1], storage offset 3, and itemsize 4 requiring a storage size of 24 are out of bounds for storage of size 15 Exception raised from checkInBoundsForStorage at /Users/nshulga/git/pytorch/pytorch/aten/src/ATen/native/Resize.h:123 (most recent call first): frame #0: c10::Error::Error(c10::SourceLocation, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>>) + 56 (0x104a9cd44 in libc10.dylib) frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&) + 120 (0x104a9a05c in libc10.dylib) frame #2: void at::native::checkInBoundsForStorage<long long>(c10::ArrayRef<long long>, c10::ArrayRef<long long>, long long, caffe2::TypeMeta const&, c10::Storage const&) + 656 (0x111dbd314 in libtorch_cpu.dylib) frame #3: void at::native::setStrided<long long>(at::Tensor const&, c10::ArrayRef<long long>, c10::ArrayRef<long long>, long long) + 152 (0x111dcd22c in libtorch_cpu.dylib) frame #4: at::native::as_strided_tensorimpl(at::Tensor const&, c10::ArrayRef<long long>, c10::ArrayRef<long long>, std::__1::optional<long long>) + 312 (0x111dccf98 in libtorch_cpu.dylib) frame #5: c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>), &at::(anonymous namespace)::(anonymous namespace)::wrapper_CPU__as_strided(at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>)>, at::Tensor, c10::guts::typelist::typelist<at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>>>, at::Tensor (at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>)>::call(c10::OperatorKernel*, c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>) + 104 (0x1129a1e94 in libtorch_cpu.dylib) frame #6: at::_ops::as_strided::call(at::Tensor const&, c10::ArrayRef<c10::SymInt>, c10::ArrayRef<c10::SymInt>, std::__1::optional<c10::SymInt>) + 476 (0x112200ad0 in libtorch_cpu.dylib) frame #7: at::Tensor::as_strided(c10::ArrayRef<long long>, c10::ArrayRef<long long>, std::__1::optional<long long>) const + 236 (0x1115db098 in libtorch_cpu.dylib) frame #8: at::native::expand(at::Tensor const&, c10::ArrayRef<long long>, bool) + 348 (0x111dcc0d4 in libtorch_cpu.dylib) frame #9: c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool), &torch::ADInplaceOrView::(anonymous namespace)::expand(c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool)>, at::Tensor, c10::guts::typelist::typelist<c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool>>, at::Tensor (c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool)>::call(c10::OperatorKernel*, c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool) + 116 (0x1157ac410 in libtorch_cpu.dylib) frame #10: c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool), &torch::autograd::VariableType::(anonymous namespace)::expand(c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool)>, at::Tensor, c10::guts::typelist::typelist<c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool>>, at::Tensor (c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool)>::call(c10::OperatorKernel*, c10::DispatchKeySet, at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool) + 992 (0x114e8b010 in libtorch_cpu.dylib) frame #11: at::_ops::expand::call(at::Tensor const&, c10::ArrayRef<c10::SymInt>, bool) + 316 (0x112743c90 in libtorch_cpu.dylib) frame #12: at::expand_size(at::Tensor const&, c10::ArrayRef<long long>) + 164 (0x1047d82b4 in basic) frame #13: BasicTest_TestForBlobResizeCPU_Test::TestBody() + 284 (0x1047d8048 in basic) ``` Pull Request resolved: pytorch#158690 Approved by: https://github.com/angelayi

syedshahbaaz · 2025-07-28T16:23:40Z

@daisyden Can you please re-review this PR?

syedshahbaaz · 2025-08-12T13:18:33Z

PR #11

…rch#165479) These happen when building with CMAKE_BUILD_TYPE=RelWithAssert This should fix two types of failures that started with pytorch#163665 Disclaimer that I used a lot of AI since I don't how pybind works or what refcounts and pointers are, so idk if this is a good solution, or even a solution at all (fwiw the tests pass now) The first one type is Truncated: ``` default_pg, _ = _new_process_group_helper( File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2096, in _new_process_group_helper backend_class = creator_fn(dist_backend_opts, backend_options) File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/distributed/fake_pg.py", line 25, in _create_fake_pg return FakeProcessGroup._create_internal( RuntimeError: new_refcount != 1 INTERNAL ASSERT FAILED at "/var/lib/jenkins/workspace/c10/util/intrusive_ptr.h":319, please report a bug to PyTorch. intrusive_ptr: Cannot increase refcount after it reached zero. Exception raised from retain_ at /var/lib/jenkins/workspace/c10/util/intrusive_ptr.h:319 (most recent call first): C++ CapturedTraceback: #4 std::_Function_handler<std::shared_ptr<c10::LazyValue<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > const> (), c10::SetStackTraceFetcher(std::function<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > ()>)::{lambda()#1}>::_M_invoke(std::_Any_data const&) from Logging.cpp:0 #5 c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) from ??:0 #6 c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) from ??:0 #7 c10::detail::torchInternalAssertFail(char const*, char const*, unsigned int, char const*, char const*) from ??:0 #8 void pybind11::class_<c10d::FakeProcessGroup, (anonymous namespace)::IntrusivePtrNoGilDestructor<c10d::FakeProcessGroup> >::init_instance<(anonymous namespace)::IntrusivePtrNoGilDestructor<c10d::FakeProcessGroup>, 0>(pybind11::detail::instance*, void const*) from init.cpp:0 #9 pybind11::detail::type_caster_generic::cast(void const*, pybind11::return_value_policy, pybind11::handle, pybind11::detail::type_info const*, void* (*)(void const*), void* (*)(void const*), void const*) from :0 #10 pybind11::cpp_function::initialize<torch::distributed::c10d::(anonymous namespace)::c10d_init(_object*, _object*)::{lambda(int, int, c10::intrusive_ptr<c10d::FakeProcessGroup::Options, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup::Options> >)pytorch#127}, c10::intrusive_ptr<c10d::FakeProcessGroup, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup> >, int, int, c10::intrusive_ptr<c10d::FakeProcessGroup::Options, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup::Options> >, pybind11::name, pybind11::scope, pybind11::sibling, pybind11::arg, pybind11::arg, pybind11::arg_v>(torch::distributed::c10d::(anonymous namespace)::c10d_init(_object*, _object*)::{lambda(int, int, c10::intrusive_ptr<c10d::FakeProcessGroup::Options, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup::Options> >)pytorch#127}&&, c10::intrusive_ptr<c10d::FakeProcessGroup, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup> > (*)(int, int, c10::intrusive_ptr<c10d::FakeProcessGroup::Options, c10::detail::intrusive_target_default_null_type<c10d::FakeProcessGroup::Options> >), pybind11::name const&, pybind11::scope const&, pybind11::sibling const&, pybind11::arg const&, pybind11::arg const&, pybind11::arg_v const&)::{lambda(pybind11::detail::function_call&)#3}::_FUN(pybind11::detail::function_call&) from init.cpp:0 ``` and I fix it here by getting rid of `DontIncreaseRefcount` and using make_intrusive to do the ref count handling instead. However, I also had to move the constructor to be public, which I think is not good, based on the reasoning of the original PR The other one type is ``` Traceback (most recent call last): File "/var/lib/jenkins/workspace/test/test_testing.py", line 2415, in test_no_warning_on_import self.assertEqual(out, "") File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/common_utils.py", line 4233, in assertEqual raise error_metas.pop()[0].to_error( # type: ignore[index] AssertionError: String comparison failed: "/opt/conda/envs/py_3.10/lib/python3.10/s[352 chars]):\n" != '' - /opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/distributed/__init__.py:29: FutureWarning: pybind11-bound class 'torch._C._distributed_c10d.FakeProcessGroup' is using an old-style placement-new '__init__' which has been deprecated. See the upgrade guide in pybind11's docs. This message is only visible when compiled in debug mode. - if is_available() and not torch._C._c10d_init(): To execute this test, run the following from the base repo dir: python test/test_testing.py TestImports.test_no_warning_on_import ``` which I fix by getting rid of the `__init__` which I think is ok since it'll just error if you try to make one? Pull Request resolved: pytorch#165479 Approved by: https://github.com/ezyang

Previously g3 = NVIDIA Tesla M60 Now g6 = NVIDIA L4 Also change cuda arch list accordingly Pros: More memory, newer GPU Cons: That was one of the few remaining tests on g3 runners, so we probably lost coverage? We can probably run more tests in parallel now but I'm not going to do that here Disabled a bunch of sparse tests and nestedtensor tests that were previously skipped due to not having sufficient hardware? They are now failing with ``` Traceback (most recent call last): File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/common_utils.py", line 3293, in wrapper method(*args, **kwargs) File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/common_utils.py", line 3292, in wrapper with policy(): File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/testing/_internal/common_utils.py", line 2532, in __enter__ self.beforeStreams[-1].synchronize() File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/cuda/streams.py", line 105, in synchronize super().synchronize() torch.AcceleratorError: CUDA error: device-side assert triggered Search for `cudaErrorAssert' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information. CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. Exception raised from stream_synchronize at /var/lib/jenkins/workspace/c10/cuda/CUDAFunctions.h:120 (most recent call first): C++ CapturedTraceback: #4 std::_Function_handler<std::shared_ptr<c10::LazyValue<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > const> (), c10::SetStackTraceFetcher(std::function<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > ()>)::{lambda()#1}>::_M_invoke(std::_Any_data const&) from Logging.cpp:0 #5 c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) from ??:0 #6 c10::cuda::c10_cuda_check_implementation(int, char const*, char const*, unsigned int, bool) [clone .cold] from CUDAException.cpp:0 #7 THCPStream_synchronize(_object*, _object*) from Stream.cpp:0 #8 cfunction_vectorcall_NOARGS from /usr/local/src/conda/python-3.10.14/Objects/methodobject.c:489 #9 _PyObject_VectorcallTstate from /usr/local/src/conda/python-3.10.14/Include/cpython/abstract.h:114 #10 _PyEval_EvalFrame from /usr/local/src/conda/python-3.10.14/Include/internal/pycore_ceval.h:46 #11 _PyObject_VectorcallTstate from /usr/local/src/conda/python-3.10.14/Include/cpython/abstract.h:114 #12 _PyEval_EvalFrame from /usr/local/src/conda/python-3.10.14/Include/internal/pycore_ceval.h:46 ``` when run with cuda launch blocking I got a ton of stuff like ``` /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [5,3,0], thread: [2,7,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [5,3,0], thread: [3,7,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [0,0,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [1,0,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [2,0,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [3,0,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [0,1,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [1,1,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [3,1,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [0,2,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [2,2,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [3,2,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [0,3,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [1,3,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [1,4,0] Assertion `value < upper_bound` failed. /var/lib/jenkins/workspace/third_party/cutlass/include/cutlass/integer_subbyte.h:124: cutlass::integer_subbyte<Bits, Signed>::integer_subbyte(unsigned int) [with int Bits = 2; __nv_bool Signed = false]: block: [3,8,0], thread: [3,4,0] Assertion `value < upper_bound` failed. ``` Pull Request resolved: pytorch#165158 Approved by: https://github.com/seemethere

Update UT to check local shards only if it exists

c3f9fc0

syedshahbaaz changed the title ~~[Distributed] Add check to verify if local shards in FSDP have tensors before accessing~~ Add check to verify if local shards in FSDP have tensors before accessing Jun 30, 2025

syedshahbaaz requested a review from daisyden June 30, 2025 19:55

daisyden reviewed Jul 14, 2025

View reviewed changes

Modify empty local_shards check condition

873c082

syedshahbaaz requested a review from daisyden July 14, 2025 18:38

syedshahbaaz closed this Aug 12, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add check to verify if local shards in FSDP have tensors before accessing #7

Add check to verify if local shards in FSDP have tensors before accessing #7

Uh oh!

syedshahbaaz commented Jun 30, 2025

Uh oh!

daisyden Jul 14, 2025

Uh oh!

syedshahbaaz Jul 14, 2025

Uh oh!

syedshahbaaz commented Jul 28, 2025

Uh oh!

syedshahbaaz commented Aug 12, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Add check to verify if local shards in FSDP have tensors before accessing #7

Add check to verify if local shards in FSDP have tensors before accessing #7

Uh oh!

Conversation

syedshahbaaz commented Jun 30, 2025

Uh oh!

daisyden Jul 14, 2025

Choose a reason for hiding this comment

Uh oh!

syedshahbaaz Jul 14, 2025

Choose a reason for hiding this comment

Uh oh!

syedshahbaaz commented Jul 28, 2025

Uh oh!

syedshahbaaz commented Aug 12, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants