torch.floor_divide not triggering ZeroDivisionError on GPU

[jiren-the-gray]

🐛 Describe the bug

Running torch.divide with 0 as the denominator does not throw ZeroDivisionError on GPU, neither does it result in inf. Executing on CPU throws ZeroDivisionError as expected.

Code snippet: Colab

Minimal repro:

import torch

output = torch.floor_divide(torch.tensor(1).cuda(), torch.tensor(0).cuda())
print(output)
# tensor(4294967295, device='cuda:0')

Versions

Collecting environment information... PyTorch version: 2.4.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A

OS: Ubuntu 20.04.6 LTS (x86_64) GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0 Clang version: 15.0.0 (git@github.com:llvm/llvm-project.git 4ba6a9c9f65bbc8bd06e3652cb20fd4dfc846137) CMake version: version 3.22.1 Libc version: glibc-2.31

Python version: 3.10.0 (default, Mar 3 2022, 09:58:08) [GCC 7.5.0] (64-bit runtime) Python platform: Linux-5.15.0-117-generic-x86_64-with-glibc2.31 Is CUDA available: True CUDA runtime version: Could not collect CUDA_MODULE_LOADING set to: LAZY GPU models and configuration: GPU 0: NVIDIA GeForce RTX 4090 Nvidia driver version: 555.42.02 cuDNN version: /usr/lib/x86_64-linux-gnu/libcudnn.so.7.6.5 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 Byte Order: Little Endian Address sizes: 39 bits physical, 48 bits virtual CPU(s): 24 On-line CPU(s) list: 0-23 Thread(s) per core: 1 Core(s) per socket: 16 Socket(s): 1 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 183 Model name: 13th Gen Intel(R) Core(TM) i7-13700KF Stepping: 1 CPU MHz: 5188.958 CPU max MHz: 5800.0000 CPU min MHz: 800.0000 BogoMIPS: 6835.20 Virtualization: VT-x L1d cache: 384 KiB L1i cache: 256 KiB L2 cache: 16 MiB NUMA node0 CPU(s): 0-23 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Mitigation; Clear Register File Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: 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; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req umip pku ospke waitpkg gfni vaes vpclmulqdq rdpid movdiri movdir64b fsrm md_clear serialize arch_lbr flush_l1d arch_capabilities

Versions of relevant libraries: [pip3] numpy==1.26.4 [pip3] optree==0.11.0 [pip3] torch==2.4.0 [pip3] triton==3.0.0 [conda] numpy 1.26.4 py310heeff2f4_0
[conda] numpy-base 1.26.4 py310h8a23956_0
[conda] optree 0.11.0 pypi_0 pypi [conda] torch 2.4.0 pypi_0 pypi [conda] triton 3.0.0 pypi_0 pypi

cc @malfet

[malfet]

This almost looks like an expected behavior to me, as execution on GPU is asynchronous and we don't want to check the inputs. And as there are no inf for integer types, INTMAX is the next best thing

[jiren-the-gray]

@malfet That makes sense, but the value returned here is the maximum 32-bit integer. For 64 bit integers, INTMAX would be 9223372036854775807. Wouldn't it make more sense to return that?

I also tried explicitly mentioning the dtype with torch.int64, it still gave me the max value for 32 bit:

import torch
import sys

output = torch.floor_divide(torch.tensor(1,dtype=torch.int64).cuda(), torch.tensor(0,dtype=torch.int64).cuda())
print(output)
# tensor(4294967295, device='cuda:0')
print(output.dtype)
# torch.int64
print(sys.maxsize)
# 9223372036854775807
print(torch.iinfo(torch.int64).max)
# 9223372036854775807

Also here: colab