Question

我正在尝试使用Keras构建CNN模型，并且我下载了Tensorflow的gpu版本（因为我有GTX 1050Ti）。一切都正常运行（导入和设置模型），直到我想将数据拟合到模型，并发生如下错误。我还在只有cpu的环境中测试了代码，一切正常。我想知道我的gpu设置是否有问题？如果是，如何解决？提前谢谢！

An error ocurred while starting the kernel
2018 01:54:24.393675: I T:\src\github\tensorflow\tensorflow\core\platform\cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
2018 01:54:24.660254: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1356] Found device 0 with properties: 
name: GeForce GTX 1050 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.62
pciBusID: 0000:01:00.0
totalMemory: 4.00GiB freeMemory: 3.29GiB
2018 01:54:24.662362: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1435] Adding visible gpu devices: 0
2018 01:54:25.310183: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:923] Device interconnect StreamExecutor with strength 1 edge matrix:
2018 01:54:25.310808: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:929] 0 
2018 01:54:25.311107: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:942] 0: N 
2018 01:54:25.311512: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1053] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 3026 MB memory) ‑> physical GPU (device: 0, name: GeForce GTX 1050 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1)
2018 01:54:28.587566: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:28.588197: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:28.601085: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:28.601751: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:28.602208: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:28.602798: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_blas.cc:462] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED
2018 01:54:31.121650: E T:\src\github\tensorflow\tensorflow\stream_executor\cuda\cuda_dnn.cc:455] could not create cudnn handle: CUDNN_STATUS_ALLOC_FAILED
2018 01:54:31.122851: F T:\src\github\tensorflow\tensorflow\core\kernels\conv_ops.cc:713] Check failed: stream‑>parent()‑>GetConvolveAlgorithms( conv_parameters.ShouldIncludeWinogradNonfusedAlgo(), &algorithms)

Answer 1

我认为GPU可以从日志中看到内存不足。在CPU上运行时，数据将被提供给系统内存，我认为在您的系统中将大于4GB，因此没有错误。为了减少内存需求，您可以在将模型拟合到数据集时尝试较小的数据批量。

使用gpu运行Keras时发生错误

1 个答案: