Giters

enyac-group / NeuralPower

The code for paper: Neuralpower: Predict and deploy energy-efficient convolutional neural networks

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Run modified Paleo on new device architecture

Ashutosh1995 opened this issue · comments

commented

If I want to run the software on an embedded device let's say Raspberry Pi, would giving the specs in device.py file suffice?

commented

I believe it would not be an issue since this device list is only applied to Paleo's model. They utilize this configuration information to have some theoretical method.

Our NeuralPower is learning based method, which would be fine as long as we one can profile the runtime/power data for embedded devices. For example, we already applied NeuralPower into embedded devices, such as Nvidia Jetson TX1. The relevant paper can be found in Power/Performance Modeling and Optimization: Using and Characterizing Machine Learning Applications, Section 6.3.