pytorch - 💡(How to fix) Fix XPU: Fatal Level Zero `RuntimeErrors` instead of recoverable `OutOfMemoryError` [9 comments, 4 participants]

🐛 Describe the bug

Unlike the CUDA backend, PyTorch on XPU fails to raise recoverable exceptions such as torch.OutOfMemoryError when physical memory is exceeded. Instead, it triggers fatal RuntimeErrors linked to Level Zero backend failures--including UR_RESULT_ERROR_DEVICE_LOST, UR_RESULT_ERROR_OUT_OF_DEVICE_MEMORY and UR_RESULT_ERROR_OUT_OF_RESOURCES--which cause the entire process to terminate.

Proposed Fix

While implementing pre-flight checks in oneAPI Level Zero may be resource-intensive, we can mitigate this within PyTorch. I’ve observed that calling torch.xpu.set_per_process_memory_fraction(1.0) at the start of a script prevents these fatal crashes. Should we consider integrating this call into the default XPU runtime initialization to ensure more stable memory management?

import torch
device = torch.device("xpu")

print(f"Attempting to force OOM...")
tensors = []
size = (16384, 16384)

# Keep allocating tensors until we run out of memory
for i in range(100):
    memory_allocated = torch.xpu.memory_allocated(device) / (1024**3)
    print(f"  Allocated tensor {i+1}: {memory_allocated:.2f} GB total")
    try:
        tensor = torch.randn(size, dtype=torch.float32, device=device)
    except torch.OutOfMemoryError as e:
        print(f"✓ OOM triggered at iteration {i+1}: {e}")
        break
    else:
        tensors.append(tensor)

Attempting to force OOM...
  Allocated tensor 1: 0.00 GB total
  Allocated tensor 2: 1.00 GB total
  Allocated tensor 3: 2.00 GB total
  Allocated tensor 4: 3.00 GB total
  Allocated tensor 5: 4.00 GB total
  Allocated tensor 6: 5.00 GB total
  Allocated tensor 7: 6.00 GB total
  Allocated tensor 8: 7.00 GB total
  Allocated tensor 9: 8.00 GB total
  Allocated tensor 10: 9.00 GB total
  Allocated tensor 11: 10.00 GB total
  Allocated tensor 12: 11.00 GB total
  Allocated tensor 13: 12.00 GB total
Traceback (most recent call last):
  File "/home/weilinxu/coder/xpu/xpu_reset/device_loss_demo.py", line 13, in <module>
    tensor = torch.randn(size, dtype=torch.float32, device=device)
RuntimeError: level_zero backend failed with error: 20 (UR_RESULT_ERROR_DEVICE_LOST)

Versions

Collecting environment information... PyTorch version: 2.10.0+xpu Is debug build: False CUDA used to build PyTorch: Could not collect ROCM used to build PyTorch: N/A

OS: Ubuntu 25.10 (x86_64) GCC version: (Ubuntu 15.2.0-4ubuntu4) 15.2.0 Clang version: Could not collect CMake version: version 4.3.0 Libc version: glibc-2.42

Python version: 3.13.8 (main, Oct 13 2025, 15:28:49) [GCC 11.4.0] (64-bit runtime) Python platform: Linux-6.17.0-14-generic-x86_64-with-glibc2.42 Is CUDA available: False CUDA runtime version: Could not collect CUDA_MODULE_LOADING set to: N/A GPU models and configuration: GPU 0: NVIDIA GeForce RTX 2080 Ti Nvidia driver version: 580.126.09 cuDNN version: Could not collect Is XPU available: True XPU used to build PyTorch: 20250301 Intel GPU driver version:

• intel-opencl-icd: 25.31.34666.3-1ubuntu1
• libze1: 1.24.1-2 Intel GPU models onboard: N/A Intel GPU models detected:
• [0] _XpuDeviceProperties(name='Intel(R) Arc(TM) B580 Graphics', platform_name='Intel(R) oneAPI Unified Runtime over Level-Zero V2', type='gpu', device_id=0xE20B, uuid=86800be2-0000-0000-6a00-000000000000, driver_version='1.6.34666', total_memory=11605MB, max_compute_units=160, gpu_eu_count=160, gpu_subslice_count=20, max_work_group_size=1024, max_num_sub_groups=64, sub_group_sizes=[16 32], has_fp16=1, has_fp64=1, has_atomic64=1) 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: 46 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 20 On-line CPU(s) list: 0-19 Vendor ID: GenuineIntel Model name: Intel(R) Core(TM) i9-9820X CPU @ 3.30GHz CPU family: 6 Model: 85 Thread(s) per core: 2 Core(s) per socket: 10 Socket(s): 1 Stepping: 4 CPU(s) scaling MHz: 30% CPU max MHz: 4200.0000 CPU min MHz: 1200.0000 BogoMIPS: 6599.98 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 pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 pti ssbd mba ibrs ibpb stibp tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi md_clear flush_l1d arch_capabilities Virtualization: VT-x L1d cache: 320 KiB (10 instances) L1i cache: 320 KiB (10 instances) L2 cache: 10 MiB (10 instances) L3 cache: 16.5 MiB (1 instance) NUMA node(s): 1 NUMA node0 CPU(s): 0-19 Vulnerability Gather data sampling: Vulnerable Vulnerability Ghostwrite: Not affected Vulnerability Indirect target selection: Not affected Vulnerability Itlb multihit: KVM: Mitigation: Split huge pages Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable Vulnerability Meltdown: Mitigation; PTI Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable Vulnerability Old microcode: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Mitigation; IBRS Vulnerability Spec rstack overflow: Not affected 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; IBRS; IBPB conditional; STIBP conditional; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsa: Not affected Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable Vulnerability Vmscape: Mitigation; IBPB before exit to userspace

Versions of relevant libraries: [pip3] Could not collect [conda] Could not collect

cc @gujinghui @EikanWang @fengyuan14 @guangyey