Skip to content

Latest commit

 

History

History
44 lines (24 loc) · 1.48 KB

server.md

File metadata and controls

44 lines (24 loc) · 1.48 KB

Getting started: access our servers for artifact evaluation

Steps

Step 1: Connect to HKU CS gatekeeper

For security reasons, our department only allow access to the gatekeeper using SSH keys.

  1. Create a new file in your computer, name it as soterkey. Copy the private key from our HotCRP response into soterkey.

  2. chmod 400 ./soterkey.

  3. ssh -i ./soterkey atc22@gatekeeper3.cs.hku.hk.

Step 2: Connect to our cluster

How to connect to the GPU server

ssh [server_account_1]@202.45.128.185

How to connect to the SGX virtual machine (using 3 commands below)

ssh [server_account_1]@202.45.128.185

ssh [server_account_2]@10.22.1.16

ssh ubuntu@192.168.122.158

server_account_1, server_account_2 and their corresponding passwords are listed in our HotCRP response. Step 3 does not need a password.

Note that, since the environment is not completely clean, the experimental results may have some outliers if other people happen to run tasks on the cluster at the same time. You may re-run the outlier data points or simply ignore them when such contention happens.

After accessing the cluster

We have helped to finish the setup for meeting prerequisite requirements and creating the working environment for GPU computation and Graphene-SGX, so you can skip these steps in the instructions.

After cloning the codebase, you can skip all Deployment steps. We have marked the starting point for running this artifact as Kick-off Functional.