硬件信息:
- NVIDIA GeForce MX230（显存2GB）
软件信息：
- win10
- 显卡驱动版本512.77
- CUDA驱动版本11.6.134（更换为旧版显卡驱动为441、CUDA10.2也存在同样的问题）
- python3.10.4
- jittor1.3.4.15
问题复现
-
创建空的conda环境，
pip install jittor安装最新版计图 -
第一次运行
python -m jittor.test.test_array，日志如下：[i 0624 21:46:28.068000 40 compiler.py:953] Jittor(1.3.4.15) src: c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor [i 0624 21:46:28.095000 40 compiler.py:954] cl at C:\Users\wawcac\.cache\jittor\msvc\VC\_\_\_\_\_\bin\cl.exe(19.29.30133) [i 0624 21:46:28.097000 40 compiler.py:955] cache_path: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default [i 0624 21:46:28.101000 40 install_cuda.py:53] cuda_driver_version: [11, 6, 0] [i 0624 21:46:28.126000 40 __init__.py:411] Found C:\Users\wawcac\.cache\jittor\jtcuda\cuda11.2_cudnn8_win\bin\nvcc.exe(11.2.67) at C:\Users\wawcac\.cache\jittor\jtcuda\cuda11.2_cudnn8_win\bin\nvcc.exe. [i 0624 21:46:28.282000 40 __init__.py:411] Found gdb(10.2) at C:\Program Files\TDM-GCC-64\bin\gdb.EXE. [i 0624 21:46:28.328000 40 __init__.py:411] Found addr2line(2.36.1) at C:\Program Files\TDM-GCC-64\bin\addr2line.EXE. [i 0624 21:46:28.367000 40 compiler.py:1008] cuda key:cu11.2.67 [i 0624 21:46:28.369000 40 __init__.py:227] Total mem: 7.82GB, using 2 procs for compiling. [i 0624 21:46:29.859000 40 jit_compiler.cc:28] Load cc_path: C:\Users\wawcac\.cache\jittor\msvc\VC\_\_\_\_\_\bin\cl.exe [i 0624 21:46:29.860000 40 init.cc:62] Found cuda archs: [61,] [i 0624 21:46:30.064000 40 compile_extern.py:516] mpicc not found, distribution disabled. ...[i 0624 21:46:31.270000 40 cuda_flags.cc:32] CUDA enabled. .[i 0624 21:46:31.278000 40 cuda_flags.cc:32] CUDA enabled. ..[i 0624 21:46:31.289000 40 cuda_flags.cc:32] CUDA enabled. [w 0624 21:46:33.938000 40 cuda_device_allocator.cc:29] Unable to alloc cuda device memory, use unify memory instead. This may cause low performance. [i 0624 21:46:33.939000 40 cuda_device_allocator.cc:31] === display_memory_info === total_cpu_ram: 15 B total_device_ram: 2GB hold_vars: 184 lived_vars: 733 lived_ops: 1318 name: sfrl is_device: 1 used: 159.7MB(18.1%) unused: 722.3MB(81.9%) total: 882MB name: sfrl is_device: 1 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 161KB(15.7%) unused: 863KB(84.3%) total: 1MB name: temp is_device: 0 used: 0 B(-nan(ind)%) unused: 0 B(-nan(ind)%) total: 0 B name: temp is_device: 1 used: 0 B(0%) unused: 442.8MB(100%) total: 442.8MB cpu&gpu: 1.41GB gpu: 1.351GB cpu: 60MB free: cpu( 0 B) gpu(68.87MB) =========================== [w 0624 21:46:34.296000 40 cuda_device_allocator.cc:29] Unable to alloc cuda device memory, use unify memory instead. This may cause low performance. [i 0624 21:46:34.296000 40 cuda_device_allocator.cc:31] === display_memory_info === total_cpu_ram: 15 B total_device_ram: 2GB hold_vars: 184 lived_vars: 665 lived_ops: 692 name: sfrl is_device: 1 used: 159.7MB(18.1%) unused: 722.3MB(81.9%) total: 882MB name: sfrl is_device: 1 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 98.5KB(9.62%) unused: 925.5KB(90.4%) total: 1MB name: temp is_device: 0 used: 0 B(-nan(ind)%) unused: 0 B(-nan(ind)%) total: 0 B name: temp is_device: 1 used: 0 B(0%) unused: 442.8MB(100%) total: 442.8MB cpu&gpu: 1.41GB gpu: 1.351GB cpu: 60MB free: cpu( 0 B) gpu(68.87MB) =========================== E.... Compiling Operators(2/2) used: 8.76s eta: 0s Compiling Operators(1/1) used: 4.37s eta: 0s [i 0624 21:46:47.603000 40 profiler.cc:468] Profile result, sorted by TotalTime ('it/s' represent number of iterations per sec) Name FileName Count TotalTime %,cum% AvgTime MinTime MaxTime Input Output InOut Compute Total time: 88.2us Total Memory Access: 80 B ?opkey0:array?T:int32?o:1?opkey1:broadcast_to?Tx:int32?DIM=1?BCAST=1?opkey2:binary?Tx:float32?Ty:int32?Tz:float32?OP:subtract?JIT:1?JIT_cuda:1?graph:000010,052020,010021,?var_info::11110121?index_t:int32 C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_array__T_int32__o_1__opkey1_broadcast_to__Tx_int32__DIM_1__BCAST_1__opkey2_binary___hash_42bbdae5e63a5b90_op.cc 1 88.2us(100%,100%) 88.2us 88.2us 88.2us 443KB/s 443KB/s 886KB/s 113Kit/s [i 0624 21:46:47.603000 40 profiler.cc:549] Memory profile result, sorted by CheckTimes Name FileName CheckTimes TLBMissRate .[i 0624 21:46:47.603000 40 cuda_flags.cc:32] CUDA enabled. [i 0624 21:46:47.604000 40 profiler.cc:468] Profile result, sorted by TotalTime ('it/s' represent number of iterations per sec) Name FileName Count TotalTime %,cum% AvgTime MinTime MaxTime Input Output InOut Compute Total time: 78.9us Total Memory Access: 80 B ?opkey0:array?T:int32?o:1?opkey1:broadcast_to?Tx:int32?DIM=1?BCAST=1?opkey2:binary?Tx:float32?Ty:int32?Tz:float32?OP:subtract?JIT:1?JIT_cuda:1?graph:000010,052020,010021,?var_info::11110121?index_t:int32 C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_array__T_int32__o_1__opkey1_broadcast_to__Tx_int32__DIM_1__BCAST_1__opkey2_binary___hash_42bbdae5e63a5b90_op.cc 1 78.9us(100%,100%) 78.9us 78.9us 78.9us 495KB/s 495KB/s 990KB/s 127Kit/s [i 0624 21:46:47.604000 40 profiler.cc:549] Memory profile result, sorted by CheckTimes Name FileName CheckTimes TLBMissRate [i 0624 21:46:47.604000 40 cuda_flags.cc:32] CUDA enabled. . Compiling Operators(4/4) used: 16.4s eta: 0s之后CPU占用会降到0，约一分钟后发生蓝屏，提示信息为driver power state failure。这一分钟里，即使按下
Ctrl-C，提示Caught SIGNAL 2, quick exit，时间到了依然会蓝屏。 -
重启后在同一环境中第二次运行
python -m jittor.test.test_array，不发生蓝屏，日志如下：[i 0624 22:00:03.576000 24 compiler.py:953] Jittor(1.3.4.15) src: c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor [i 0624 22:00:03.624000 24 compiler.py:954] cl at C:\Users\wawcac\.cache\jittor\msvc\VC\_\_\_\_\_\bin\cl.exe(19.29.30133) [i 0624 22:00:03.625000 24 compiler.py:955] cache_path: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default [i 0624 22:00:03.677000 24 install_cuda.py:53] cuda_driver_version: [11, 6, 0] [i 0624 22:00:03.716000 24 __init__.py:411] Found C:\Users\wawcac\.cache\jittor\jtcuda\cuda11.2_cudnn8_win\bin\nvcc.exe(11.2.67) at C:\Users\wawcac\.cache\jittor\jtcuda\cuda11.2_cudnn8_win\bin\nvcc.exe. [i 0624 22:00:03.946000 24 __init__.py:411] Found gdb(10.2) at C:\Program Files\TDM-GCC-64\bin\gdb.EXE. [i 0624 22:00:03.999000 24 __init__.py:411] Found addr2line(2.36.1) at C:\Program Files\TDM-GCC-64\bin\addr2line.EXE. [i 0624 22:00:04.040000 24 compiler.py:1008] cuda key:cu11.2.67 [i 0624 22:00:04.043000 24 __init__.py:227] Total mem: 7.82GB, using 2 procs for compiling. [i 0624 22:00:06.539000 24 jit_compiler.cc:28] Load cc_path: C:\Users\wawcac\.cache\jittor\msvc\VC\_\_\_\_\_\bin\cl.exe [i 0624 22:00:06.562000 24 init.cc:62] Found cuda archs: [61,] [i 0624 22:00:07.039000 24 compile_extern.py:516] mpicc not found, distribution disabled. ...[i 0624 22:00:10.926000 24 cuda_flags.cc:32] CUDA enabled. .[i 0624 22:00:10.978000 24 cuda_flags.cc:32] CUDA enabled. ..[i 0624 22:00:10.993000 24 cuda_flags.cc:32] CUDA enabled. [w 0624 22:00:14.356000 24 cuda_device_allocator.cc:29] Unable to alloc cuda device memory, use unify memory instead. This may cause low performance. [i 0624 22:00:14.357000 24 cuda_device_allocator.cc:31] === display_memory_info === total_cpu_ram: 15 B total_device_ram: 2GB hold_vars: 184 lived_vars: 733 lived_ops: 1318 name: sfrl is_device: 1 used: 159.7MB(18.1%) unused: 722.3MB(81.9%) total: 882MB name: sfrl is_device: 1 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 161KB(15.7%) unused: 863KB(84.3%) total: 1MB name: temp is_device: 0 used: 0 B(-nan(ind)%) unused: 0 B(-nan(ind)%) total: 0 B name: temp is_device: 1 used: 0 B(0%) unused: 442.8MB(100%) total: 442.8MB cpu&gpu: 1.41GB gpu: 1.351GB cpu: 60MB free: cpu( 0 B) gpu(68.87MB) =========================== [w 0624 22:00:14.910000 24 cuda_device_allocator.cc:29] Unable to alloc cuda device memory, use unify memory instead. This may cause low performance. [i 0624 22:00:14.911000 24 cuda_device_allocator.cc:31] === display_memory_info === total_cpu_ram: 15 B total_device_ram: 2GB hold_vars: 184 lived_vars: 665 lived_ops: 692 name: sfrl is_device: 1 used: 159.7MB(18.1%) unused: 722.3MB(81.9%) total: 882MB name: sfrl is_device: 1 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 58MB(98.3%) unused: 1MB(1.69%) total: 59MB name: sfrl is_device: 0 used: 98.5KB(9.62%) unused: 925.5KB(90.4%) total: 1MB name: temp is_device: 0 used: 0 B(-nan(ind)%) unused: 0 B(-nan(ind)%) total: 0 B name: temp is_device: 1 used: 0 B(0%) unused: 442.8MB(100%) total: 442.8MB cpu&gpu: 1.41GB gpu: 1.351GB cpu: 60MB free: cpu( 0 B) gpu(68.87MB) =========================== E....[i 0624 22:00:15.152000 24 profiler.cc:468] Profile result, sorted by TotalTime ('it/s' represent number of iterations per sec) Name FileName Count TotalTime %,cum% AvgTime MinTime MaxTime Input Output InOut Compute Total time: 146us Total Memory Access: 80 B ?opkey0:array?T:int32?o:1?opkey1:broadcast_to?Tx:int32?DIM=1?BCAST=1?opkey2:binary?Tx:float32?Ty:int32?Tz:float32?OP:subtract?JIT:1?JIT_cuda:1?graph:000010,052020,010021,?var_info::11110121?index_t:int32 C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_array__T_int32__o_1__opkey1_broadcast_to__Tx_int32__DIM_1__BCAST_1__opkey2_binary___hash_42bbdae5e63a5b90_op.cc 1 146us(100%,100%) 146us 146us 146us 267KB/s 267KB/s 534KB/s 68.3Kit/s [i 0624 22:00:15.153000 24 profiler.cc:549] Memory profile result, sorted by CheckTimes Name FileName CheckTimes TLBMissRate .[i 0624 22:00:15.153000 24 cuda_flags.cc:32] CUDA enabled. [i 0624 22:00:15.154000 24 profiler.cc:468] Profile result, sorted by TotalTime ('it/s' represent number of iterations per sec) Name FileName Count TotalTime %,cum% AvgTime MinTime MaxTime Input Output InOut Compute Total time: 36.3us Total Memory Access: 80 B ?opkey0:array?T:int32?o:1?opkey1:broadcast_to?Tx:int32?DIM=1?BCAST=1?opkey2:binary?Tx:float32?Ty:int32?Tz:float32?OP:subtract?JIT:1?JIT_cuda:1?graph:000010,052020,010021,?var_info::11110121?index_t:int32 C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_array__T_int32__o_1__opkey1_broadcast_to__Tx_int32__DIM_1__BCAST_1__opkey2_binary___hash_42bbdae5e63a5b90_op.cc 1 36.3us(100%,100%) 36.3us 36.3us 36.3us 1.05MB/s 1.05MB/s 2.1MB/s 276Kit/s [i 0624 22:00:15.154000 24 profiler.cc:549] Memory profile result, sorted by CheckTimes Name FileName CheckTimes TLBMissRate [i 0624 22:00:15.154000 24 cuda_flags.cc:32] CUDA enabled. ..[i 0624 22:00:15.231000 24 cuda_flags.cc:32] CUDA enabled. [i 0624 22:00:15.231000 24 cuda_flags.cc:32] CUDA enabled. .... ====================================================================== ERROR: test_memcopy_overlap (__main__.TestArray) ---------------------------------------------------------------------- Traceback (most recent call last): File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 125, in inner ret = func(*args, **kw) File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py", line 49, in test_memcopy_overlap a.sync() RuntimeError: [f 0624 22:00:14.826000 24 executor.cc:665] Execute fused operator(128/186) failed. [JIT Source]: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_broadcast_to__Tx_float32__DIM_7__BCAST_19__opkey1_reindex__Tx_float32__XDIM_4__YD___hash_b0347dac92a33895_op.cc [OP TYPE]: fused_op:( broadcast_to, reindex, binary.multiply, reduce.add,) [Input]: float32[256,256,3,3,]layer3.0.conv2.weight, float32[100,256,14,14,], [Output]: float32[100,256,14,14,], [Async Backtrace]: not found, please set env JT_SYNC=1, trace_py_var=3 [Reason]: [f 0624 22:00:14.825000 24 helper_cuda.h:128] CUDA error at c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor\src\mem\allocator\cuda_device_allocator.cc:32 code=2( cudaErrorMemoryAllocation ) cudaMallocManaged(&ptr, size) ********** Async error was detected. To locate the async backtrace and get better error report, please rerun your code with two enviroment variables set: cmd: >>> set JT_SYNC=1 >>> set trace_py_var=3 powershell: >>> $env:JT_SYNC=1 >>> $env:trace_py_var=3 During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 124, in inner with self: File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 145, in __exit__ setattr(flags, k, v) RuntimeError: [f 0624 22:00:15.051000 24 executor.cc:665] Execute fused operator(68/126) failed. [JIT Source]: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_broadcast_to__Tx_float32__DIM_7__BCAST_19__opkey1_reindex__Tx_float32__XDIM_4__YD___hash_b0347dac92a33895_op.cc [OP TYPE]: fused_op:( broadcast_to, reindex, binary.multiply, reduce.add,) [Input]: float32[256,256,3,3,]layer3.0.conv2.weight, float32[100,256,14,14,], [Output]: float32[100,256,14,14,], [Async Backtrace]: not found, please set env JT_SYNC=1, trace_py_var=3 [Reason]: [f 0624 22:00:15.051000 24 helper_cuda.h:128] CUDA error at c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor\src\mem\allocator\cuda_device_allocator.cc:32 code=2( cudaErrorMemoryAllocation ) cudaMallocManaged(&ptr, size) ********** Async error was detected. To locate the async backtrace and get better error report, please rerun your code with two enviroment variables set: cmd: >>> set JT_SYNC=1 >>> set trace_py_var=3 powershell: >>> $env:JT_SYNC=1 >>> $env:trace_py_var=3 ---------------------------------------------------------------------- Ran 18 tests in 4.337s FAILED (errors=1) -
按照提示设置
JT_SYNC和trace_py_var后的报错信息如下：====================================================================== ERROR: test_memcopy_overlap (__main__.TestArray) ---------------------------------------------------------------------- Traceback (most recent call last): File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 125, in inner ret = func(*args, **kw) File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py", line 49, in test_memcopy_overlap a.sync() RuntimeError: [f 0624 22:06:50.515000 40 executor.cc:665] Execute fused operator(128/186) failed. [JIT Source]: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_broadcast_to__Tx_float32__DIM_7__BCAST_19__opkey1_reindex__Tx_float32__XDIM_4__YD___hash_b0347dac92a33895_op.cc [OP TYPE]: fused_op:( broadcast_to, reindex, binary.multiply, reduce.add,) [Input]: float32[256,256,3,3,]layer3.0.conv2.weight, float32[100,256,14,14,], [Output]: float32[100,256,14,14,], [Async Backtrace]: --- C:\ProgramData\Miniconda3\envs\jittor\lib\runpy.py:196 <_run_module_as_main> C:\ProgramData\Miniconda3\envs\jittor\lib\runpy.py:86 <_run_code> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py:212 <<module>> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\main.py:101 <__init__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\main.py:271 <runTests> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\runner.py:184 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:84 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:122 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:84 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:122 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:650 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:591 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:549 <_callTestMethod> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:125 <inner> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py:48 <test_memcopy_overlap> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:152 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:144 <_forward_impl> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\nn.py:2054 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:52 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\nn.py:847 <execute> [Reason]: [f 0624 22:06:50.514000 40 helper_cuda.h:128] CUDA error at c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor\src\mem\allocator\cuda_device_allocator.cc:32 code=2( cudaErrorMemoryAllocation ) cudaMallocManaged(&ptr, size) During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 124, in inner with self: File "C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py", line 145, in __exit__ setattr(flags, k, v) RuntimeError: [f 0624 22:06:50.709000 40 executor.cc:665] Execute fused operator(68/126) failed. [JIT Source]: C:\Users\wawcac\.cache\jittor\jt1.3.4\cl\py3.10.4\Windows-10-10.xd5\IntelRCoreTMi7x8f\default\cu11.2.67\jit\__opkey0_broadcast_to__Tx_float32__DIM_7__BCAST_19__opkey1_reindex__Tx_float32__XDIM_4__YD___hash_b0347dac92a33895_op.cc [OP TYPE]: fused_op:( broadcast_to, reindex, binary.multiply, reduce.add,) [Input]: float32[256,256,3,3,]layer3.0.conv2.weight, float32[100,256,14,14,], [Output]: float32[100,256,14,14,], [Async Backtrace]: --- C:\ProgramData\Miniconda3\envs\jittor\lib\runpy.py:196 <_run_module_as_main> C:\ProgramData\Miniconda3\envs\jittor\lib\runpy.py:86 <_run_code> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py:212 <<module>> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\main.py:101 <__init__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\main.py:271 <runTests> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\runner.py:184 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:84 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:122 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:84 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\suite.py:122 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:650 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:591 <run> C:\ProgramData\Miniconda3\envs\jittor\lib\unittest\case.py:549 <_callTestMethod> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:125 <inner> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\test\test_array.py:48 <test_memcopy_overlap> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:152 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:144 <_forward_impl> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\nn.py:2054 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\models\resnet.py:52 <execute> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\__init__.py:954 <__call__> C:\ProgramData\Miniconda3\envs\jittor\lib\site-packages\jittor\nn.py:847 <execute> [Reason]: [f 0624 22:06:50.709000 40 helper_cuda.h:128] CUDA error at c:\programdata\miniconda3\envs\jittor\lib\site-packages\jittor\src\mem\allocator\cuda_device_allocator.cc:32 code=2( cudaErrorMemoryAllocation ) cudaMallocManaged(&ptr, size) ---------------------------------------------------------------------- Ran 18 tests in 3.431s FAILED (errors=1)
蓝屏后强制重启，尚未保存文件会丢失。