Switch between CPU and GPU devices for cnn example

Open lzjpaul opened this issue 3 years ago • 3 comments

For cifar_distributed_cnn and cnn examples, the model is by default run on the GPU. Need to enable it to run on CPU for machines with only CPUs

Dec 05 '22 10:12 lzjpaul

Add a conditional judgement to decide whether use CPU or GPU and then apply the devices to datasets and models

Feb 21 '23 03:02 lzjpaul

https://github.com/apache/singa/commit/0a2d9b20c743b7b1501f1b50db32d36b00de85a0

Feb 21 '23 03:02 lzjpaul

modify line 114 for judging CPU and GPU in https://github.com/apache/singa/blob/dev/examples/cifar_distributed_cnn/train_cnn.py

Feb 24 '23 08:02 lzjpaul

This has been resolved in https://github.com/apache/singa/blob/dev/examples/cifar_distributed_cnn/train_cnn.py

Jun 13 '24 14:06 lzjpaul