Inconsistent behavior for _refs.fft.* operators when input is an empty tensor

[ekamiti]

🐛 Describe the bug

In eager mode when an empty tensor input is used for a torch.fft.* function it seems that the expectation is for it to fail as per this test:

pytorch/test/test_spectral_ops.py

Lines 236 to 244 in a361fce

	# Note: NumPy will throw a ValueError for an empty input
	@onlyNativeDeviceTypes
	@ops(spectral_funcs, allowed_dtypes=(torch.half, torch.float, torch.complex32, torch.cfloat))
	def test_empty_fft(self, device, dtype, op):
	t = torch.empty(1, 0, device=device, dtype=dtype)
	match = r"Invalid number of data points \([-\d]*\) specified"
	with self.assertRaisesRegex(RuntimeError, match):
	op(t)

In eager mode this seems to be validated here:

pytorch/aten/src/ATen/native/SpectralOps.cpp

Line 236 in a361fce

TORCH_CHECK(n >= 1, "Invalid number of data points (", n, ") specified");

When trying to enable this test for the version of the operators under torch/_refs the behavior is inconsistent

FAILED [0.0393s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_fft_cpu_complex64 - AssertionError: RuntimeError not raised
FAILED [0.0237s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_fft_cpu_float32 - AssertionError: RuntimeError not raised
FAILED [0.0109s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_hfft_cpu_complex64 - AssertionError: "Invalid number of data points \([-\...
FAILED [0.0404s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_hfft_cpu_float32 - AssertionError: "Invalid number of data points \([-\d]...
FAILED [0.0115s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_ifft_cpu_complex64 - ZeroDivisionError: division by zero
FAILED [0.0120s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_ifft_cpu_float32 - ZeroDivisionError: division by zero
FAILED [0.0114s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_ihfft_cpu_float32 - ZeroDivisionError: division by zero
FAILED [0.0109s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_irfft_cpu_complex64 - AssertionError: "Invalid number of data points \([-...
FAILED [0.0111s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_irfft_cpu_float32 - AssertionError: "Invalid number of data points \([-\d...
FAILED [0.0114s] test/test_spectral_ops.py::TestFFTCPU::test_empty_fft__refs_fft_rfft_cpu_float32 - AssertionError: RuntimeError not raised

Repro of the difference:

>>> t = torch.empty(1, 0, device='cpu', dtype=torch.float32)
>>> torch.fft.fft(t)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
RuntimeError: Invalid number of data points (0) specified
>>> torch._refs.fft.fft(t)
tensor([], size=(1, 0), dtype=torch.complex64)
>>> torch._refs.fft.hfft(t)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/mnt/d/Programming/pytorch/torch/_prims_common/wrappers.py", line 227, in _fn
    result = fn(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^
  File "/mnt/d/Programming/pytorch/torch/_refs/fft.py", line 238, in hfft
    return _fft_c2r("hfft", input, n, dim, norm, forward=True)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/d/Programming/pytorch/torch/_refs/fft.py", line 119, in _fft_c2r
    torch._check(
  File "/mnt/d/Programming/pytorch/torch/__init__.py", line 987, in _check
    _check_with(RuntimeError, cond, message)
  File "/mnt/d/Programming/pytorch/torch/__init__.py", line 970, in _check_with
    raise error_type(message_evaluated)
RuntimeError: Invalid number of data points (None) specified

Versions

Collecting environment information...
PyTorch version: 2.1.0a0+git45bcd74
Is debug build: True
CUDA used to build PyTorch: Could not collect
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.2 LTS (x86_64)
GCC version: (Ubuntu 12.1.0-2ubuntu1~22.04) 12.1.0
Clang version: 15.0.7
CMake version: version 3.26.4
Libc version: glibc-2.35

Python version: 3.11.3 (tags/v3.11.3:f3909b8bc8, May 16 2023, 16:08:28) [Clang 15.0.7 ] (64-bit runtime)
Python platform: Linux-5.10.102.1-microsoft-standard-WSL2-x86_64-with-glibc2.35
Is CUDA available: False
CUDA runtime version: 12.1.105
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: GPU 0: NVIDIA GeForce GTX 1060 6GB
Nvidia driver version: 531.61
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.1
/usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.1
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 48 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 12
On-line CPU(s) list: 0-11
Vendor ID: AuthenticAMD
Model name: AMD Ryzen 7 2700X Eight-Core Processor
CPU family: 23
Model: 8
Thread(s) per core: 2
Core(s) per socket: 6
Socket(s): 1
Stepping: 2
BogoMIPS: 7386.26
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl tsc_reliable nonstop_tsc cpuid extd_apicid pni pclmulqdq ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy svm cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext ssbd ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero xsaveerptr virt_ssbd arat npt nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold v_vmsave_vmload
Virtualization: AMD-V
Hypervisor vendor: Microsoft
Virtualization type: full
L1d cache: 192 KiB (6 instances)
L1i cache: 384 KiB (6 instances)
L2 cache: 3 MiB (6 instances)
L3 cache: 8 MiB (1 instance)
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Full AMD retpoline, IBPB conditional, STIBP disabled, RSB filling
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected

Versions of relevant libraries:
[pip3] flake8==6.0.0
[pip3] flake8-bugbear==23.3.23
[pip3] flake8-comprehensions==3.12.0
[pip3] flake8-executable==2.1.3
[pip3] flake8-logging-format==0.9.0
[pip3] flake8-pyi==23.3.1
[pip3] flake8-simplify==0.19.3
[pip3] mypy==0.960
[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.24.3
[pip3] pytorch-triton==2.1.0+7d1a95b046
[pip3] torch==2.1.0a0+git2563111
[conda] magma-cuda121 2.6.1 1 pytorch
[conda] mkl 2023.1.0 h6d00ec8_46342
[conda] mkl-include 2023.1.0 h06a4308_46342

cc @ezyang @mruberry @lezcano @peterbell10 @SherlockNoMad

[lezcano]

It looks like the reference is missing that check. We would accept a fix for it.