Message boards : Number crunching : Projects Computing Power
Message board moderation

To post messages, you must log in.

AuthorMessage
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 7744 - Posted: 18 May 2005, 2:23:25 UTC
Last modified: 18 May 2005, 2:23:58 UTC

Hey Markku,
May I ask how LHC's computing power compares to what you had originally planned in house for the sixtrack program?I know we have a only a bit over 5000 users and 15000+ hosts at any given time at present. Estimates of a million work unit run with 5 million results are that it would last us about a month or so as posted on these boards seem generally accurate. May main question is how does that compare timwise crunching compared to supercomputers you might have used or clustering with same amount of workunits?
ID: 7744 · Report as offensive     Reply Quote
Profile Markku Degerholm

Send message
Joined: 3 Sep 04
Posts: 212
Credit: 4,545
RAC: 0
Message 7753 - Posted: 19 May 2005, 11:12:05 UTC - in response to Message 7744.  

> Hey Markku,
> May I ask how LHC's computing power compares to what you had originally
> planned in house for the sixtrack program?I know we have a only a bit over
> 5000 users and 15000+ hosts at any given time at present. Estimates of a
> million work unit run with 5 million results are that it would last us about a
> month or so as posted on these boards seem generally accurate. May main
> question is how does that compare timwise crunching compared to supercomputers
> you might have used or clustering with same amount of workunits?

We have currently about 2500 active users and 5000 active hosts. It's more than with any other computation platform we can use for Sixtrack, but I'm not sure about the details.

Markku Degerholm
LHC@home admin
ID: 7753 · Report as offensive     Reply Quote
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 7758 - Posted: 19 May 2005, 22:24:05 UTC

Thank you for your answer Markku :) If details ever become available to you on this it would be nice to know how our presently small community stacks up to say a supercomputer you might use.
ID: 7758 · Report as offensive     Reply Quote
Profile Skip Da Shu

Send message
Joined: 2 Sep 04
Posts: 33
Credit: 2,057,517
RAC: 0
Message 7759 - Posted: 20 May 2005, 0:09:52 UTC - in response to Message 7758.  
Last modified: 20 May 2005, 0:13:49 UTC

> Thank you for your answer Markku :) If details ever become available to you on
> this it would be nice to know how our presently small community stacks up to
> say a supercomputer you might use.

There's an interesting Scientific American article on the subject of Distributed Computing and BOINC. David P. Anderson gives some t-flops numbers and the author compares that to IBM's BlueGene/L (biggest super computer) t-flops numbers. SETI CPDN and LHC are all mentioned in the article.

Here's a link... http://www.sciam.com/article.cfm?articleID=0001BDCA-77CB-1264-B1DB83414B7F0000

PS: Well LHC was in the original print version in a sidebar, I don't see it in the on-line version!


- da shu @ HeliOS,
"Free software” is a matter of liberty, not price. To understand the concept, you should think of “free” as in “free speech”, not as in “free beer".
ID: 7759 · Report as offensive     Reply Quote
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 7763 - Posted: 20 May 2005, 2:30:38 UTC

Nice article Skip Da Shu -thanks.... now let me see Dr Anderson says 500k hosts (must be including classic) outperform fastest supercomputer by about 30%. LHCs resources are about 1% of that... howevever obviously sixtrack would not have been using that computer... so I guess I have to defer to Markku's comment that we outperform any platform they could use and it has to suffice for now for an answer hmmm my math tells me that our output is approx 1 terraflop sound right? What do university's have available to them?
ID: 7763 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 7766 - Posted: 20 May 2005, 7:12:51 UTC - in response to Message 7763.  

> Nice article Skip Da Shu -thanks.... now let me see Dr Anderson says 500k
> hosts (must be including classic) outperform fastest supercomputer by about
> 30%. LHCs resources are about 1% of that... howevever obviously sixtrack would
> not have been using that computer... so I guess I have to defer to Markku's
> comment that we outperform any platform they could use and it has to suffice
> for now for an answer hmmm my math tells me that our output is approx 1
> terraflop sound right? What do university's have available to them?

Anything they can afford to rent time on. That is the real rub.

When I got a tour of the San Diego Super-Computer center they had a Cray there (XMP?) and they had room for another (I don't know if they ever got it). Anyway, one of the factoids was that their time was usually booked out 2-5 years in advance. The real problem is capacity and availability.

With BOINC, we have continuous overall availability and effectively unlimited capacity. BOINCoids like me are just as likely to add more computers if we have more interesting projects to subscribe to so that we can participate. I mean, though SETI@Home is not one of my priorities it is distressing (lord only knows why as it isn't a priority) to see my "standings" number "falling", heck, I am two or three tics from falling out of the top 1,000 ... :(

Then again, I am hoping I can talk one of my daughters into letting me run BOINC on her machine ... and I have hopes to buy a new G5 this year (or whatever Apple calls the hoped for dual-dual-dual machine (dual-threaded, dual-core, dual processor workstation; oh, and running at 3.0 GHz).

Anyway, with BOINC turning into a community, we are tying more participants together, with more available projects that can interest a wider variety of people... Heck, maybe we should ask Dr. Anderson to call Dell and have them include an icon on the desktop to BOINC like they do for AOL ... :)
ID: 7766 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 7767 - Posted: 20 May 2005, 7:31:33 UTC - in response to Message 7766.  

> Then again, I am hoping I can talk one of my daughters into letting me run
> BOINC on her machine

You seem to have many daughters ;-)

> Heck, maybe we should ask Dr. Anderson to call Dell and have them
> include an icon on the desktop to BOINC like they do for AOL ... :)

Or directly in the operating systems from Microsoft and Apple ;-)



Supporting BOINC, a great concept !
ID: 7767 · Report as offensive     Reply Quote
Profile Chrulle

Send message
Joined: 27 Jul 04
Posts: 182
Credit: 1,880
RAC: 0
Message 7772 - Posted: 20 May 2005, 9:53:24 UTC

Well let's see.

Actually knowing what kind of compute power is available at Cern is not easy, as each experiment is its own unit. Normally they will buy computing power in the computer centre, because there are people with the needed expertise there 24/7, but sometimes the experiments set up their own servers.

I can only comment on what is available in the computer centre:

The main computational resource at Cern is the LXBatch cluster. It consist of about 1500-2000 server type pcs(dual Xeons with plenty of ram). This number is being increased and it is planned that it should reach around 6000 when the LHC comes online.

Furthermore Cern is experimenting with other new resources. These are used to run some of the computing on, but they are not yet fully integrated:

The Grid
Cern is creating a Grid called LCG (Lhc Computing Grid), the software is still being developed but it is being used for some computation. At the moment it is on the order of 10'000 machines. For more information on grids see:
http://gridcafe.org/

OpenCluster
The OpenCluster is a collaboration between industry(IBM,HP,Enterasys and others) and Cern it consists of about a 100 dual Itanium 2 1.5 GHz machines from HP and an IBM storage tank for disk space and a couple of entrasys 10 Gbit/s switches.
More information here:
http://proj-openlab-datagrid-public.web.cern.ch/proj-openlab-datagrid-public/



When LHC@home was really running during the 50th anniversary the users delivered the same amount of compute power as about 350 dedicated machines(P4 2,4 GHz) could. While this is a lot less than what the computer centre can deliver, it is a lot more than the share that is allocated to Sixtrack. As far as i know the division that is responsible for sixtrack has bought 64 dedicated machines for running beam studies. So LHC@home has facilitated a number of studies that would not have been possible otherwise. Normally for example only a few starting angles are used(around 5). It is believed that this will give an adequate sampling of the phasespace. In one of the studies we checked this assumption by running a study with a much more fine grained sampling of the angles. This study would probably not have been done without LHC@home.

I hope this answered your question.



Chrulle
Research Assistant & Ex-LHC@home developer
Niels Bohr Institute
ID: 7772 · Report as offensive     Reply Quote
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 7775 - Posted: 20 May 2005, 10:27:32 UTC

Yes Chrulle it answered my question as best as possible as it was a somewhat generalzed question (went fishing).It is something I have been curious about and am glad to know just how much this research is helping Cern. Thank-you for your reply
ID: 7775 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 7779 - Posted: 20 May 2005, 14:00:49 UTC - in response to Message 7767.  

> > Then again, I am hoping I can talk one of my daughters into letting me
> run
> > BOINC on her machine
>
> You seem to have many daughters ;-)

Only two. :)

And all I can say is that even with my, um, rather inept parenting they turned out not too bad. But I will also say that two was plenty.

> Or directly in the operating systems from Microsoft and Apple ;-)

Well, some might consider that they constiture a virus ... :)

ID: 7779 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 7781 - Posted: 20 May 2005, 16:01:09 UTC - in response to Message 7779.  

> Well, some might consider that they constiture a virus ... :)

DC is a different kind, but it is really a virus ;-)



Supporting BOINC, a great concept !
ID: 7781 · Report as offensive     Reply Quote
ric

Send message
Joined: 17 Sep 04
Posts: 190
Credit: 649,637
RAC: 0
Message 7782 - Posted: 20 May 2005, 16:27:57 UTC - in response to Message 7781.  
Last modified: 20 May 2005, 16:29:41 UTC

>I hope this answered your question.
Thanks for this informative stuff!




> > Well, some might consider that they constiture a virus ... :)
>
> DC is a different kind, but it is really a virus ;-)
>
a virus ? a Worm ? or a trojan horse?

according a good book over this theme from Chaos Computer Club (Germany)
one possible definition of a virus is

- self reproduction
- functionality


it happen, when opening the taskmanager, there is more than the "usual" number of "crunching application" running self reproduction

it happen the Boinc client is not acting like estimated, for example, some
(unrecommended) Versions are not getting "work", older version crashed in some circumstances, had ul/dl saturations or generaly not doing what the user was asking for functionality

depending the point of view, if both is given..

;)
ID: 7782 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 7783 - Posted: 20 May 2005, 17:32:38 UTC

I know that BOINC is not either, but an application running at 100% CPU, if you don't know what is going on, well, most people are not that computer literate.

As a matter of fact, it is a sad commentary that computers are still so difficult to operate that it takes significant amounts of skill and knowledge to work with them effectively. I have been doing computers since 1975 both professionally and personnally (as a hobby) meaning that almost all my waking moments have been about computers, and my lack of knowledge and skills are astounding.

As an example, the last decade of my professional life was about the logical design of relational database systems and the analysis of data to extract business rules and I get a lot of people telling me that I don't know what I am talking about ... :)

Anyway, BOINC does appear to have a few problems where two copies get started, I see on the Macintosh extra processes (Science Applications) running attempting to do 3 work units at the same time on a two CPU system. What makes this interesting is that I have not seen it on any Windows machine (which does not mean that it is not happening) nor have I heard of it on a Linux machine.

ID: 7783 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 7788 - Posted: 21 May 2005, 11:10:52 UTC

Guys, sorry, but you missunderstood my post.

I wanted to say, that a human can get infected "like a virus" with the fun of running DC-projects.

If I see, how much of my time I have spent with boinc, I'm infected. Many other people too




Supporting BOINC, a great concept !
ID: 7788 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 7790 - Posted: 21 May 2005, 14:03:33 UTC - in response to Message 7788.  

> Guys, sorry, but you missunderstood my post.
>
> I wanted to say, that a human can get infected "like a virus" with the fun of
> running DC-projects.
>
> If I see, how much of my time I have spent with boinc, I'm infected. Many
> other people too

There is *NO* way that I am infected!

I can walk away from BOINC at any time!

An I never spend more than 23.5 hours a day checking my stats, writing on my site, checking production of my farm, etc.

The very IDEA that I might be infected, absurd!

Hey! Have you checked out those new computers, imagine what they would do in my farm ...
ID: 7790 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 7792 - Posted: 21 May 2005, 15:14:44 UTC - in response to Message 7790.  

> There is *NO* way that I am infected!
>
> I can walk away from BOINC at any time!
>
> An I never spend more than 23.5 hours a day checking my stats, writing
> on my site, checking production of my farm, etc.
>
> The very IDEA that I might be infected, absurd!
>
> Hey! Have you checked out those new computers, imagine what they would do in
> my farm ...
>

LOLOLOLOLOLOLOLOL

Okay, you are not infected ...

... you are part of the virus ;-)




Supporting BOINC, a great concept !
ID: 7792 · Report as offensive     Reply Quote

Message boards : Number crunching : Projects Computing Power


©2024 CERN