Python下在多显卡机器上指定Tensorflow使用哪几块设备
以下代码指定使用0号和1号显卡,分配70%的显存。实际使用时请根据显卡的互相连接情况调整设备号
import os import tensorflow as tf from keras.backend.tensorflow_backend import set_session os.environ["CUDA_VISIBLE_DEVICES"] = '0, 1' config = tf.ConfigProto() config.gpu_options.allocator_type = 'BFC' #A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc. config.gpu_options.per_process_gpu_memory_fraction = 0.7 config.gpu_options.allow_growth = True set_session(tf.Session(config=config))
本文出自 Tech Trace,转载时请注明出处及相应链接。
本文永久链接: https://www.qiujiahui.com/2017/12/12/python%e4%b8%8b%e5%9c%a8%e5%a4%9a%e6%98%be%e5%8d%a1%e6%9c%ba%e5%99%a8%e4%b8%8a%e6%8c%87%e5%ae%9atensorflow%e4%bd%bf%e7%94%a8%e5%93%aa%e5%87%a0%e5%9d%97%e8%ae%be%e5%a4%87/