If you needed a dedicated server (linux) I'd be willing to host if possible.
I currently run 2, 1 vhost web account and 1 linode Linux server (which OICS is currently running).
The machine is under utilized and I barely his 1% bandwidth allocation so would gladly share for such a project.
Again, distributed perft
Moderator: Ras
-
jshriver
- Posts: 1371
- Joined: Wed Mar 08, 2006 9:41 pm
- Location: Morgantown, WV, USA
-
gladius
- Posts: 568
- Joined: Tue Dec 12, 2006 10:10 am
- Full name: Gary Linscott
Re: Co-location server requirements
Or, you could start with a cost of $0 with EC2, and migrate to dedicated boxes if the project takes off? Even then, I'd use a hosted box, not a colo. For example, for $50/mo you get a quad core with 16Gb ram - http://www.hetzner.de/en/hosting/produk ... server/ex4.sje wrote:The history of distributed computing, and SETI@home in particular, shows that server requirements have been repeatedly and severely underestimated.
How many people do you expect to be running the perft client, and how many updates/second? Right now, running the micro ec2 instance, the fishtest server easily handles hundreds of updates a minute, logging all the results to a database, and serving the webpage from the same machine.
That's at a utilization of about 3-5% average. So, you'd need thousands of updates a minute before it would start becoming a problem.
-
sje
- Posts: 4675
- Joined: Mon Mar 13, 2006 7:43 pm
Re: Again, distributed perft
While I'm sure that Perft@home would succeed if properly staffed and funded, I'm also sure that I can't do much in either aspect, at least at present.
What I could do is modify my CookieCat chess program, which already has several different perft routines, and turn it into a Perft@home client. This would mean adding little more than a command line argument parser and a reporting facility.
Using CookieCat to build a client is attractive because:
1. It's open source.
2. It's been very well tested.
3. It uses bitboards, and so is reasonably fast on a 64 bit machine (but can run on a 32 bit box).
4. It's written in Free Pascal which has itself been ported to many different platforms. Having a portable client is very important to attract a large number of volunteer end users.
CookieCat could also be used to build the Perft@home server software, but hooks for a database, the network, and other BOINC requirements would be needed. This would take same time, for sure.
https://dl.dropboxusercontent.com/u/316 ... kieCat.tar
What I could do is modify my CookieCat chess program, which already has several different perft routines, and turn it into a Perft@home client. This would mean adding little more than a command line argument parser and a reporting facility.
Using CookieCat to build a client is attractive because:
1. It's open source.
2. It's been very well tested.
3. It uses bitboards, and so is reasonably fast on a 64 bit machine (but can run on a 32 bit box).
4. It's written in Free Pascal which has itself been ported to many different platforms. Having a portable client is very important to attract a large number of volunteer end users.
CookieCat could also be used to build the Perft@home server software, but hooks for a database, the network, and other BOINC requirements would be needed. This would take same time, for sure.
https://dl.dropboxusercontent.com/u/316 ... kieCat.tar