Collected experimental results in the CK format from the ReQuEST@ASPLOS'18 tournament on reproducible SW/HW co-design of Pareto-efficient deep learning for the following submission:
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

compatibility License: CC BY 4.0

This repository contains raw experimental results in the CK format for the image classification workflow from the ReQuEST tournament at ASPLOS'18 on reproducible SW/HW co-design of deep learning (speed, accuracy, energy, costs). The live ReQuEST scoreboard shows a subset of these results.



Minimal CK installation

The minimal installation requires:

  • Python 2.7 or 3.3+ (limitation is mainly due to unitests)
  • Git command line client.

You can install CK in your local user space as follows:

$ git clone
$ export PATH=$PWD/ck/bin:$PATH

You can also install CK via PIP with sudo to avoid setting up environment variables yourself:

$ sudo pip install ck

Install this CK repository and dependencies

$ ck pull repo:ck-request-asplos18-results-caffe-intel

List available experiments

$ ck ls ck-request-asplos18-results-caffe-intel:experiment:*

Replay experiment

$ ck replay experiment:{name from above list}

Note that CK will try to automatically rebuild experimental setup by detecting already installed software dependencies and installing missing ones using shared CK packages.

If you want to have a software setup as close to the original one as possible, install packages before running replay as described in the ReadMe of the related CK workflow.

Start dashboard to visualize/compare results

$ ck dashboard request.asplos18 --results=ck-request-asplos18-results-caffe-intel

See the live scoreboard


Further discussions