🐛 Describe the bug
Description: In the function compiled by torch.compile, torch.compile still generates uninitialized data when torch.use_deterministic_algorithms(True) is run before. This is inconsistent with the eager mode behavior
A minimal reproduction:
import torch
torch._inductor.config.fallback_random=True
torch.use_deterministic_algorithms(True)
def fn():
return torch.empty_like(torch.randn(4, 4))
cfunc = torch.compile(fn)
out1 = fn()
out2 = cfunc()
torch.testing.assert_close(out1, out2, equal_nan=True)This will generate the following inconsistency:
"""
out1: tensor([[nan, nan, nan, nan],
[nan, nan, nan, nan],
[nan, nan, nan, nan],
[nan, nan, nan, nan]])
out2: tensor([[1.4462e-31, 4.5877e-41, 3.7315e-31, 0.0000e+00],
[0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[1.0322e-36, 0.0000e+00, 1.0315e-36, 0.0000e+00],
[1.0322e-36, 0.0000e+00, 1.0315e-36, 0.0000e+00]])
Traceback (most recent call last):
File "$FILE", line 14, in <module>
torch.testing.assert_close(out1, out2, equal_nan=True)
File "$ENVIRONMENT/lib/python3.11/site-packages/torch/testing/_comparison.py", line 1600, in assert_close
raise error_metas[0].to_error(msg)
AssertionError: Tensor-likes are not close!
Mismatched elements: 16 / 16 (100.0%)
Greatest absolute difference: nan at index (0, 0) (up to 1e-05 allowed)
Greatest relative difference: nan at index (0, 0) (up to 1.3e-06 allowed)
"""
According to the documentation of empty_like here, floating point and complex tensors should be filled with NaN.
There is a similar issue #113707, but this has been already fixed in torch 2.10.0, which will generate the error message RuntimeError: This compiled backward function is being run with torch.use_deterministic_algorithms(True), but it was previously generated during the forward function while torch.use_deterministic_algorithms(False) was set. now.
Error logs
No response
Versions
PyTorch version: 2.10.0+cu126
Is debug build: False
CUDA used to build PyTorch: 12.6
ROCM used to build PyTorch: N/A
OS: AlmaLinux 10.1 (Heliotrope Lion) (x86_64)
GCC version: (GCC) 14.3.1 20250617 (Red Hat 14.3.1-2)
Clang version: 19.1.7 (/home/runner/work/llvm-project/llvm-project/clang cd708029e0b2869e80abe31ddb175f7c35361f90)
CMake version: version 3.30.5
Libc version: glibc-2.39
Python version: 3.11.11 (main, Dec 11 2024, 16:28:39) [GCC 11.2.0] (64-bit runtime)
Python platform: Linux-6.12.0-124.20.1.el10_1.x86_64-x86_64-with-glibc2.39
Is CUDA available: True
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to:
GPU models and configuration:
GPU 0: NVIDIA RTX 6000 Ada Generation
GPU 1: NVIDIA RTX 6000 Ada Generation
Nvidia driver version: 590.44.01
cuDNN version: Could not collect
Is XPU available: False
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
Caching allocator config: N/A
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): 256
On-line CPU(s) list: 0-255
Vendor ID: AuthenticAMD
Model name: AMD EPYC 7773X 64-Core Processor
CPU family: 25
Model: 1
Thread(s) per core: 2
Core(s) per socket: 64
Socket(s): 2
Stepping: 2
Frequency boost: enabled
CPU(s) scaling MHz: 62%
CPU max MHz: 3529.3369
CPU min MHz: 2200.0000
BogoMIPS: 4400.24
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 xtopology nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local user_shstk clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin brs arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold v_vmsave_vmload vgif v_spec_ctrl umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca fsrm debug_swap
Virtualization: AMD-V
L1d cache: 4 MiB (128 instances)
L1i cache: 4 MiB (128 instances)
L2 cache: 64 MiB (128 instances)
L3 cache: 1.5 GiB (16 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-63,128-191
NUMA node1 CPU(s): 64-127,192-255
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: 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: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Mitigation; Safe RET
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsa: Vulnerable: No microcode
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Mitigation; IBPB before exit to userspace
Versions of relevant libraries:
[pip3] jaxlib==0.4.14+cuda11.cudnn86
[pip3] numpy==1.26.0
[pip3] nvidia-cublas-cu11==11.11.3.6
[pip3] nvidia-cublas-cu12==12.6.4.1
[pip3] nvidia-cuda-cupti-cu11==11.8.87
[pip3] nvidia-cuda-cupti-cu12==12.6.80
[pip3] nvidia-cuda-nvrtc-cu12==12.6.77
[pip3] nvidia-cuda-runtime-cu11==11.8.89
[pip3] nvidia-cuda-runtime-cu12==12.6.77
[pip3] nvidia-cudnn-cu11==9.10.1.4
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cufft-cu11==10.9.0.58
[pip3] nvidia-cufft-cu12==11.3.0.4
[pip3] nvidia-curand-cu12==10.3.7.77
[pip3] nvidia-cusolver-cu11==11.4.1.48
[pip3] nvidia-cusolver-cu12==11.7.1.2
[pip3] nvidia-cusparse-cu11==11.7.5.86
[pip3] nvidia-cusparse-cu12==12.5.4.2
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.6.85
[pip3] nvidia-nvtx-cu12==12.6.77
[pip3] torch==2.10.0+cu126
[pip3] torchvision==0.25.0+cu126
[pip3] triton==3.6.0
[conda] jaxlib 0.4.14+cuda11.cudnn86 pypi_0 pypi
[conda] numpy 1.26.0 pypi_0 pypi
[conda] nvidia-cublas-cu11 11.11.3.6 pypi_0 pypi
[conda] nvidia-cublas-cu12 12.6.4.1 pypi_0 pypi
[conda] nvidia-cuda-cupti-cu11 11.8.87 pypi_0 pypi
[conda] nvidia-cuda-cupti-cu12 12.6.80 pypi_0 pypi
[conda] nvidia-cuda-nvrtc-cu12 12.6.77 pypi_0 pypi
[conda] nvidia-cuda-runtime-cu11 11.8.89 pypi_0 pypi
[conda] nvidia-cuda-runtime-cu12 12.6.77 pypi_0 pypi
[conda] nvidia-cudnn-cu11 9.10.1.4 pypi_0 pypi
[conda] nvidia-cudnn-cu12 9.10.2.21 pypi_0 pypi
[conda] nvidia-cufft-cu11 10.9.0.58 pypi_0 pypi
[conda] nvidia-cufft-cu12 11.3.0.4 pypi_0 pypi
[conda] nvidia-curand-cu12 10.3.7.77 pypi_0 pypi
[conda] nvidia-cusolver-cu11 11.4.1.48 pypi_0 pypi
[conda] nvidia-cusolver-cu12 11.7.1.2 pypi_0 pypi
[conda] nvidia-cusparse-cu11 11.7.5.86 pypi_0 pypi
[conda] nvidia-cusparse-cu12 12.5.4.2 pypi_0 pypi
[conda] nvidia-cusparselt-cu12 0.7.1 pypi_0 pypi
[conda] nvidia-nccl-cu12 2.27.5 pypi_0 pypi
[conda] nvidia-nvjitlink-cu12 12.6.85 pypi_0 pypi
[conda] nvidia-nvtx-cu12 12.6.77 pypi_0 pypi
[conda] torch 2.10.0+cu126 pypi_0 pypi
[conda] torchvision 0.25.0+cu126 pypi_0 pypi
[conda] triton 3.6.0 pypi_0 pypi
cc @ezyang @gchanan @kadeng @msaroufim @chauhang @penguinwu @voznesenskym @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @muchulee8 @amjames @aakhundov @coconutruben @jataylo
🐛 Describe the bug
Description: In the function compiled by
torch.compile,torch.compilestill generates uninitialized data whentorch.use_deterministic_algorithms(True)is run before. This is inconsistent with the eager mode behaviorA minimal reproduction:
This will generate the following inconsistency:
According to the documentation of
empty_likehere, floating point and complex tensors should be filled with NaN.There is a similar issue #113707, but this has been already fixed in
torch 2.10.0, which will generate the error messageRuntimeError: This compiled backward function is being run with torch.use_deterministic_algorithms(True), but it was previously generated during the forward function while torch.use_deterministic_algorithms(False) was set.now.Error logs
No response
Versions
cc @ezyang @gchanan @kadeng @msaroufim @chauhang @penguinwu @voznesenskym @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @muchulee8 @amjames @aakhundov @coconutruben @jataylo