coolbei
coolbei
I have a full convolutional network. when I input variable data into the net, memory keeps growing. when I modify the souce code, to use origin caffe convolutional layer instead...
hi, Im using a really simple network, and this error happen: i modify the question, so that a small net can reproduce the error stably. ========================================================================== ========================================================================== update: this small...
for now, streamHolder is a global singleton. And all the primitives share one mkldnn stream. It has 2 problems: 1. not thread-safe. Even when a net is created and used...