Script(s) for testing execution time of Keras-TF ANN training in CPU and GPU Benchmark_v2.py is the functional script for
Before using the script, make sure you have installed numpy, platform, psutil, cpuinfo, and tensorflow
benchmark_v2.py [scan, run, show, save] Parameters: scan: Scan the hardware's device, return the device name, the processor name(cpu gen), processor speed [ Hz ], device's total RAM, operating system and the GPU if it has one. No argument needed. help: -h, displays the help of the program. run: Runs the program, it has default values for each of the arguments, the values in the argument can be changed - H: (default 1) The device in which the model is running, 1=CPU, 2=GPU. - d: (default 1) Select the dataset used for training 1=fashion_mnist, 2=cifar10. - l: (default 3) The number of hidden layers used in the model. - i: (default 1000) The epochs of training for the model. show: Print all the devices information saved. No argument needed. save: Save the current benchmark. No argument needed.