Artificial Intelligence System

Grizz · **Posted:** Mon Feb 04, 2008 2:15 pm

12th by RAC, come on Linux users!
This is a Canadian project, Lets get to #1 :#1:

user312 · **Posted:** Mon Feb 04, 2008 8:49 pm

My only somewhat powerful Linux box was sharing MindModeling and BURP - I've now set them both to "no new tasks" so my output should increase in the next couple of days. My only concern is that whenever I check the server status page it always has "0 results ready to send", so I'm not quite fully willing to commit to it as a sole project on any given box. I may go back to sharing with other projects if the work flow isn't available.

Grizz · **Posted:** Mon Feb 04, 2008 8:51 pm

You could set some projects to 10% and increase the AI to 1000% then when they run out of work the other projects take over.

user312 · **Posted:** Tue Feb 05, 2008 8:56 pm

Grizz wrote:

You could set some projects to 10% and increase the AI to 1000% then when they run out of work the other projects take over.

I'm well aware of that - but that only works on a project basis. Since the projects I'm crunching are being crunched on 50 or so computers (haven't counted lately, but it's somewhere around there) that would screw up my resource share for my secondary/tertiary projects on the rest of my farm. If I could do this on a per-computer basis I would. Before you mention it - yes I know about editing XML files to set the preferences, however since I use BAM this either a)resets the resource share every time it re-synchronizes or b) screws up BOINC if I set the file to "read only" to prevent this. If you know a way around this, let me know...

Pawly · **Posted:** Thu Feb 21, 2008 11:00 am

Windows Apps Now available!! Pawly has his first few credits already!! WOOHOO!

user312 · **Posted:** Thu Feb 21, 2008 12:15 pm

Pawly wrote:

Windows Apps Now available!! Pawly has his first few credits already!! WOOHOO!

I don't know about the rest of you, but I've been seeing incredibly slow downloads on their latest batch of wus...

Grizz · **Posted:** Thu Feb 21, 2008 12:19 pm

I have been getting slow downloads as well, and when I visit the site, it is slow to.

JPLiz · **Posted:** Thu Feb 21, 2008 12:37 pm

It takes longer to get to your account then to crunch a wu.

Pawly · **Posted:** Thu Feb 21, 2008 7:13 pm

HHHHHHMMMMM?? Yeah it was super slow earlier and now it seems intermittent. Fast, slow, fast, slow. But ya gotta love Ovidiu leading the way for our team so far! The forum seems fast but with a few bugs here and there.

user312 · **Posted:** Thu Feb 21, 2008 9:49 pm

My Windoze host was a little screwed and ended up having nothing but failed downloads (x100 - I guess that's why there's a daily limit!). A Detatch/Re-attach seems to have cleared it up, but now I have to wait until tomorrow to get any work. My Linux server is crunching nicely though...

Grizz · **Posted:** Thu Feb 21, 2008 10:29 pm

I had the same issue with Windows and had to detach and reattach. Linux is doing good.

It looks like they increased the length of the WU's quite a bit.

Buffalo Bill · **Posted:** Fri Feb 22, 2008 4:58 am

I created an account and got 10 wu's and it was only a little slow. I attached a Celeron Win XP 2.9 GHz for now. Anyone know if the applications run faster on Linux or Windows?

BB

user312 · **Posted:** Fri Feb 22, 2008 7:10 am

Grizz wrote:

I had the same issue with Windows and had to detach and reattach. Linux is doing good.

It looks like they increased the length of the WU's quite a bit.

Supposedly in the 20hr range now. Luckily there's checkpointing (although the status doesn't increment yet!)...

Grizz · **Posted:** Fri Feb 22, 2008 1:21 pm

Buffalo Bill wrote:

I created an account and got 10 wu's and it was only a little slow. I attached a Celeron Win XP 2.9 GHz for now. Anyone know if the applications run faster on Linux or Windows?

BB

I run on both but I don't see a significant difference in speed, except for the fact that my Linux computer is alot slower than the windows one I run it on.

Buffalo Bill · **Posted:** Sat Feb 23, 2008 4:49 am

Celeron did first WU in 8 hours 49 min.

I notice some heavy disk thrashing on this project. Anyone else notice this. Slows the whole machine down and happens every few minutes. I have 1 Gig RAM on this machine.

BB

user312 · **Posted:** Sat Feb 23, 2008 5:22 am

Buffalo Bill wrote:

Celeron did first WU in 8 hours 49 min.

I notice some heavy disk thrashing on this project. Anyone else notice this. Slows the whole machine down and happens every few minutes. I have 1 Gig RAM on this machine.

BB

Yeah, I/O reads are massive on this app (writes are pretty high as well).

Pawly · **Posted:** Sat Feb 23, 2008 6:13 am

I was mailing back and forth with Crunch3r today (he's a mod at SETI as well) and thanked him for helping with the windows app. He says that there are still some "growing pains" to be expected.

Crunch3r wrote:

...The app v 1.03 are working for a few seconds and after that idling out … however Ovidiu got all the stuff to trace back the errors. I’ve send him all that I could figure out...

Crunch3r is EVERYWHERE these days. Milkyway, TSP, DECs Helping out all the Project admins solve their problems. If you bump into him online, let him know we appreciate all his efforts.

Buffalo Bill · **Posted:** Sun Feb 24, 2008 1:11 am

Pawly wrote:

I was mailing back and forth with Crunch3r today (he's a mod at SETI as well) and thanked him for helping with the windows app. He says that there are still some "growing pains" to be expected.

Crunch3r wrote:

...The app v 1.03 are working for a few seconds and after that idling out … however Ovidiu got all the stuff to trace back the errors. I’ve send him all that I could figure out...

Crunch3r is EVERYWHERE these days. Milkyway, TSP, DECs Helping out all the Project admins solve their problems. If you bump into him online, let him know we appreciate all his efforts.

Thanks for the update Pawly. I also attached my Linux box just now to see how it runs AI. I can live the disk activity as it's still less than Docking@home was.

BB

user312 · **Posted:** Wed Apr 09, 2008 4:40 am

FYI - AI has had fairly steady work for the last couple of weeks in case anyone else is interested in crunching a few units for a Canadian based project.

They are still in "debug" mode so stability is only OK - I'm experiencing about a 15-20% Compute error rate on my Windows hosts (Linux has been fine). My gut feeling is that it has something to do with interupting the work unit in the middle of a calculation (i.e. not leaving it in memory when suspended, or stopping boinc (reboot, restart service, etc)) as it occurs more frequently on hosts that get rebooted or don't leave wus in memory on suspension. It also will pop up an error message if you reboot when a wu is partially complete (not sure if it happens all the time, but I've seen it enough to take notice). According to a post on their forums by Ovidiu, they will be releasing a new version in a while with better debug capabilities (apparently in the next month or so according to their development schedule).

Pawly · **Posted:** Wed Jul 09, 2008 11:57 pm

Newsletter

Quote:

Dear Participant,

We have reached today the 100 billion simulated neurons level. I would like to congratulate you for making this a reality and also to thank you for your participation and patience during these couple of months. The beginning was especially
challenging, with the application being unstable and raising lots of error. This was a collective effort.

Proving that there is enough computer power to simulate brain size biophysical neural networks at the cellular level
removes a large stumbling block from the path of Artificial Intelligence. That in turn will lead to increase
levels of interest in this field and will result in additional applications that will eventually benefit
the entire society.

The significance of this event can be summarized with the following image:

If you are interested to find out more about the future project steps please use the following link:
http://www.intelligencerealm.com/aisyst ... php?id=205

In order to give something back in return for all this effort, I found that there is nothing more useful than the following learning method:

==========================
= Constant Inquiry =
==========================
= WHY? =
= WHAT? =
= HOW? =
= WHEN? =
= IF? =
==========================

Inquiry leads to understanding. Constant inquiry leads to constant understanding.
This has long been a proven method of analysis and learning, and it was under employment by Socrates, an ancient greek philosopher.

Thank you.

We appreciate your support.

Best regards,
Ovidiu Anghelidi

Artificial Intelligence System

Who is online