Tensorflow 是否支持 Tesla K80
Does Tensorflow support Tesla K80
我是 运行 Tesla K80 上的 Tensorflow 0.8,带有 CUDA 7.5 和 CUDNN v5。一切正常,但两台设备无法相互访问。
下面列出了警告日志。谢谢。
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 0 to device ordinal 2
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 0 to device ordinal 3
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 1 to device ordinal 2
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 1 to device ordinal 3
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 2 to device ordinal 0
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 2 to device ordinal 1
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 3 to device ordinal 0
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 3 to device ordinal 1
是的,TensorFlow 应该可以在 Tesla K80 GPU 上运行(尽管它不是 officially tested devices). As Robert Crovella points out in 之一,这些警告消息仅说明您的系统未配置为在各种 GPU 设备,但您应该仍然可以在它们上 运行 TensorFlow。
我敢打赌,您一定有像这样的多路配置:
每个 K80 都没有共享同一个 PCIe 根联合体。
然后,允许从 GPU0 到 GPU1 的点对点访问,但不允许从 GPU0 到 GPU2/GPU3。
Tensorflow 应该能够检测到这种系统并在 GPU 之间执行手动复制。
我是 运行 Tesla K80 上的 Tensorflow 0.8,带有 CUDA 7.5 和 CUDNN v5。一切正常,但两台设备无法相互访问。
下面列出了警告日志。谢谢。
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 0 to device ordinal 2
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 0 to device ordinal 3
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 1 to device ordinal 2
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 1 to device ordinal 3
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 2 to device ordinal 0
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 2 to device ordinal 1
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 3 to device ordinal 0
I tensorflow/core/common_runtime/gpu/gpu_init.cc:59] cannot enable peer access from device ordinal 3 to device ordinal 1
是的,TensorFlow 应该可以在 Tesla K80 GPU 上运行(尽管它不是 officially tested devices). As Robert Crovella points out in
我敢打赌,您一定有像这样的多路配置:
每个 K80 都没有共享同一个 PCIe 根联合体。 然后,允许从 GPU0 到 GPU1 的点对点访问,但不允许从 GPU0 到 GPU2/GPU3。
Tensorflow 应该能够检测到这种系统并在 GPU 之间执行手动复制。