LAMP Computing

Learning and Machine Perception team (LAMP) computing resources documentation. To know more about the team, check out the official page.

If you’re looking how to use the tunnel, check out Remote access.

Note

This documentation is under development.

A lot of inspiration has been taken from great cluster documentation out there, e.g., Princeton Research Computing or the beloved Aalto Scicomp.

Contents

Connecting to the cluster

The main way of connecting to a cluster or any server is through Secure Shell (ssh), which is executed via a terminal. Basic terminal skills are assumed here. A couple more complex options have been put in place.

Method	Description	From where?
SSH	Standard way of connecting via command line.	Connections only from University, proxy or VPN networks.
Tunnel	Proxy jump server to make ssh (see above) available from outside again.	Whole internet.
VDI	“Virtual desktop interface”, (Guacamole), from there you can access your in-house machine – connect to servers and run graphical programs.	Whole internet
Web portal	Web-based interface to the cluster. Includes shell access.	Whole internet
VPN	Institutional VPN that will introduce you to the university network. You will be able to connect to servers as if you were inside the center. Third-party client needed.	Whole internet

Getting an account

The cluster obeys CVC account system. In most cases, to use the cluster you must obtain an institutional account by contacting the IT department.

Connecting via SSH

A regular ssh command looks like this

ssh server.cvc.es

# Using a certain username
ssh username@server.cvc.es

# Using a certain username and port
ssh username@server.cvc.es -p 12345

CVC configuration

In the CVC, the default port for ssh connections is 22345, so don’t forget to specify it in your command.

If you’re inside the university network, domain names can also be used. I.e., if I were to connect to a server I could use the IP address xxx.xxx.xxx.115 or simply:

ssh username@cudahpc15 -p 22345

Check the assigned name of a server in Resources.

You can avoid retyping your password by setting up your private keys.

Even more, with OpenSSH you can make use of your ~/ssh/config file for a more seamless connection. Check it here or search online for ssh config file.

Remote access

If you are outside of the university network and want to connect to a cluster or desktop computer in the CVC, connections are not longer available.

But worry not, an SSH tunnel has been set to enable regular work again. First of all, mail IT services or Héctor (hlaria@cvc...) for an account.

After that, only one more flag is needed in your ssh command

ssh -J tunnel_user@tunnel_ip:22345 username@cudahcp15 -p 22345

and you should be able to work normally.

Guacamole portal

…

Web Service portal

…

VPN

…

Data

GPU computing

Resources

Node name	Node type	CPU type	Memory Configuration	Ethernet	GPUs	Disks
cudahpc03	Supermicro SYS-7048GR-TR	2x6 cores Xeon E5 2620 v3 @ 2.40GHz	128GB 2133	1Gbps	4x RTX 3090 24G	SSD
cudahpc06	Supermicro SYS-4029GP-TRT	2x10 cores Xeon Silver 4114 @ 2.20GHz	192GB DDR4-2666	20Gbps	8x A40 45G	SSD
cudahpc15	Supermicro SYS-4029GP-TRT2	2x16 cores Xeon Silver 4216 @ 2.10GHz	384GB DDR4-2933	1Gbps	10x RTX 6000 24G	SSD
cudahpc38	ASUS X99-E WS/USB 3.1	1x6 cores Xeon E5 1650 v4 @ 3.60GHz	64GB DDR4-2400	1Gbps	4x GTX 1080 Ti 11G	SSD
cudahpc39	ASUS X99-E WS	1x6 cores i7 6850K @ 3.60GHz	64GB DDR4-2133	1Gbps	4x GTX 1080 Ti 11G	SSD
cudahpc05 (audi)	Supermicro SYS-4028GR-TRT2	2x12 cores Xeon E5 2650 v4 @ 2.20GHz	128GB DDR4-2400	1Gbps	10x TITAN Xp 12G	SSD
cudahpc12 (audi)	Supermicro SYS-4029GP-TRT2	2x16 cores Xeon Silver 4216 @ 2.10GHz	384GB DDR4-2933	1Gbps	5x RTX 6000 24G, 5x TITAN Xp 12G	SSD