Boinc pentathalon race

[biodoc]

Official race thread.

I've registered our TeAm and submitted our votes for the following projects:

Astronomy: Milkyway@home
Physics & Chemistry: Einstein@home
Biology & Medicine: Rosetta@home
Mathematics: PrimeGrid
Alternative: Yoyo@home

 

[biodoc]

The first project has been announced here.

snippet:

"We herewith announce the 1st project in the discipline of Physics & Chemistry.

The relevant project is Einstein@Home. All credits granted between
05/05/2010, 0.00 (UTC) and
05/10/2010, 0.00 (UTC) will be taken into the BOINC Pentathlon validation. Please note that with a lot of projects it usually takes a while before credits are granted!"

 

[petrusbroder]

OK, I'll put most of my comps on Einstein@home directly after the PrimeGrid race is done.

 

[filibusterman]

will gpu stats be counted?

 

[petrusbroder]

All creds will be counted - if the project supports GPUs then they count ...

 

[Ken g6]

All creds will be counted - if the project supports GPUs then they count ...

And Einstein doesn't appear to support GPUs. :hmm:

 

[filibusterman]

And Einstein doesn't appear to support GPUs. :hmm:

I was referring to milkyway, sorry.

 

[Rattledagger]

And Einstein doesn't appear to support GPUs. :hmm:

Hmm, it doesn't look like there's any gpu-apps, according to that page...

... But, they do have a "use CUDA-card"-selection in the project-specific preferences, and there is some posts about CUDA and so on... Among other things that only one part of the Einstein-application uses the GPU, while most of the time is still used by the cpu, so you'll tie-up a cpu-core for a "small" increase in production...

In any case, I've never had a nvidia, so someone else will have to test-out if they're getting any cuda-work from Einstein or not...

 

[biodoc]

More info by email:

The BOINC Pentathlon will start on 05/05/2010. Thus, the first project was announced. The anncouncements can be retrieved by

Feed: http://www.setigermany.de/blog/category/boinc-pentathlon/feed
Twitter: http://twitter.com/SETIGermany
Facebook: http://www.facebook.com/pages/SETIGermany/177588328907 (with an average delay of 3 hours; update every 6 hours)
All announcements are available in english, french and german.

The BOINC Pentathlon stats are here:
http://www.setigermany.de/boinc_pentathlon/statistiken/pentathlon.php
Your team stats are available at:
http://www.setigermany.de/boinc_pentathlon/statistiken/teamstats_en_81_TeAm+AnandTech.html
Please control for each project (projects will appear a week before it will start), if the address of your team XML file is correct:
http://www.setigermany.de/boinc_pentathlon/teamxml.php

We wish you lots of fun crunching the projects.
crille
SETI.Germany

 

[biodoc]

Hmm, it doesn't look like there's any gpu-apps, according to that page...

... But, they do have a "use CUDA-card"-selection in the project-specific preferences, and there is some posts about CUDA and so on... Among other things that only one part of the Einstein-application uses the GPU, while most of the time is still used by the cpu, so you'll tie-up a cpu-core for a "small" increase in production...

In any case, I've never had a nvidia, so someone else will have to test-out if they're getting any cuda-work from Einstein or not...

I did crunch a few cuda WUs some time back and you're right, there was only a small increase in production. As soon as the PrimeGrid race is over, I'll test it again on my GTX295 to see if there is a steady supply of WUs.

 

[Ken g6]

Please control for each project (projects will appear a week before it will start), if the address of your team XML file is correct:
http://www.setigermany.de/boinc_pentathlon/teamxml.php

Well, that's interesting. Our XML file is "not well-formed"!

 

[biodoc]

Well, that's interesting. Our XML file is "not well-formed"!

I sent an email to get this problem resolved.

 

[Ken g6]

The second project has been announced here:

We herewith announce the 2nd project in the discipline of Astronomy.

The relevant project is MilkyWay@home. All credits granted between
05/09/2010, 0.00 (UTC) and
05/14/2010, 0.00 (UTC) will be taken into the BOINC Pentathlon validation.

I'm currently working on Einstein WUs. I'm downloading enough for 4 days, I'll disable networking once they're in (egad, that's a lot of data!), and I'll re-enable it on 5/5. If their WUs have a long enough timeout, people with ATI GPUs might want to do something similar with Milkyway.

 

[filibusterman]

so for this race they dont care if you download wu early and just dump them during the race?

 

[Philippart]

it seems so

 

[rabrittain]

OK ... (wow, I'm really foggy this morning) ... I think that I understand that we cruch Eistein starting on the 5th, and then Milkyway starting on the 9th ... I will refer to this thread often so that I can stay straight ... for some reason I was thinking that the Pentathalon would be over by the 15th so that I could concentrate on CASP ... It's been a while since I was so confused about what I'm doing 2 weeks out. I can usually see 6 months down the road. Maybe it's Altzheimer's.

 

[Ken g6]

We flush Einstein starting on the 5th GMT (that sounds like a nasty thing to do to him! ), and then flush Milkyway starting on the 9th. I'm already crunching Einstein.

Edit: By the way, this is a very confusing race. I didn't figure the above out until yesterday.

 

[Philippart]

yes it is indeed quite strange to run 5 projects for 5 days each during a 2 week period...



They should do more math projects for this contest

 

[filibusterman]

yea how do you prevent uploads other than disconnecting from the internet. on higher lenght wu i guess you could suspend the task right before its finished but for milkyway my 5850 goes through them too fast to make that reasonable.

 

[Rattledagger]

The second project has been announced here:



I'm currently working on Einstein WUs. I'm downloading enough for 4 days, I'll disable networking once they're in (egad, that's a lot of data!), and I'll re-enable it on 5/5. If their WUs have a long enough timeout, people with ATI GPUs might want to do something similar with Milkyway.

This method won't really work with Milkyway, since Milkyway limits you to 48 tasks at once per GPU. On the 58xx-series of clients, this means less than 2 hours cached work.

As for Einstein@home, it seems the "Arecibo"-files finally has started to download, so while I won't start running Einstein@home before evening the 4th., I should atleast have the work cached beforehand.

 

[Philippart]

ouch we may run into trouble...

http://einstein.phys.uwm.edu/server_status.php

The regular einstein app will be out of work in approx 6 days, right in the middle of our race. Let's hope they have enough arecibo work for us

 

[Ken g6]

yea how do you prevent uploads other than disconnecting from the internet.

Pretty close to that - in BOINC Manager -> Activity -> Network Activity Suspended

Let's hope they have enough arecibo work for us

Let's see...less than 2 hours observation time processed per day...over 1000 hours left to process...I think it'll be OK.

 

[Ken g6]

Well, believe it or not...

We herewith announce the 3rd project in the discipline of Alternative.

The relevant project is FreeHAL@home. All credits granted between
05/10/2010, 0.00 (UTC) and
05/15/2010, 0.00 (UTC) will be taken into the BOINC Pentathlon validation.

Not only does this almost entirely overlap Milkyway, we also got out-voted! D:

I guess people with GPUs are supposed to do Milkyway, and the rest of us do this?

 

[biodoc]

In the Milkyway preferences, you can "uncheck" "use CPU", so I can crunch milkyway on my video cards and crunch FreeHal on my CPUs.

FreeHal has a 32-bit client for Windows and 32 & 64-bit clients for linux. I'll test 64-bit linux vs the 32-bit windows client to see if one is faster than the other. All the top rigs at FreeHal are windows boxes so I'm not too optimistic about their linux clients.

Another note, FreeHal has a minimum quorum of 1 on their WUs, so there's no dependency on a "wingman" to receive credit.

This will be interesting to say the least!

 

[biodoc]

Weird!! FreeHal uses next to nothing on CPU cycles. It looks like you can crunch it at the same time as a CPU intensive project.