Hi Tulpas, welcome to Betterhumans and the Rosetta@home team ! It's great you decided to sign up for both Betterhumans and the team. It's also good to have one of the (almost) 'locals' in the team. ;-) With six members, this is starting to become a respectable, little team. So, who wants to be number seven ...and eight ??

Go to the fist post of this thread for more info on Rosetta@home and details on how to join.

Previous posts:

What is Rosetta@home ?
Lecture on Rosetta@home
Letter to Rosetta@home participants: CASP
David Baker, 2004 Feynman prize
Rosetta@home is not for profit
Rosetta@home vs. Folding@home comparison
AP/CNN article on Rosetta@home
Design proteins and enzymes on your computer

Welcome to Rosetta@home and the Betterhumans team, McCoy, great to have another fellow-European in the team - and even one with three computers crunching for Rosetta ! :-)

PS: You seem to have had one or two errors among your first Rosetta WUs. It might perhaps help to set "Leave application in memory while preempted" to "yes" in your general Rosetta preferences - should be no problem in your case since you seem to have lots of memory.

Hey Hoelder1in,

I think the problems were because my post powerful computer is overclocked a lot, and thus sometimes it throws out windows error messages during intense calculations. Although I tested it with Prime during a couple of hours and I didn't get any errors, and have played quite a lot of games with it with no erros also, it seems it's not 100% rock stable, because I've seen a couple of times a general exception fault with Rosetta@home (which is maybe the cause of the errors).

Are those occassional errors a big issue for the project? I don't want to clock down my processor because it works perfect with everything else (I use that computer mainly for gaming and web design and so far I have had 0 problems). It would be also a shame to not be able to use this computer with Rosetta@home as it's the most powerful one in my house.


I also have a PowerMac G4 with Ubuntu installed, so I want to install Rosetta@home there too, but first I need to get another router for this room (the Mac doesn't has internet access unless I borrow the cable from the overclocked computer).

Hi McCoy,

Good to hear that you have even one more computer to run Rosetta on. I guess this means that my days on place one of the team are numbered. ;-) So, good luck with your Ubuntu-Mac...

Are occasional errors an issue ? Well, the WUs that error out won't do any harm, other than wasting your computer time. So, I guess your question is: will the science be compromised if some of the other WUs, that don't produce exception faults, return numerically incorrect results ? I think the answer to that is no ! The way Rosetta works, only the protein structures with the lowest energies are useful and the team re-calculates the energies of all the returned lowest-energy structures (which can be done very quickly). So if one of your results erroneously claimed a very low energy, it would be caught at that stage. And due to the random nature of the search algorithm, I'd assume that an occasional numerical error won't even effect the end result in a negative way.

There is one other thing about your returned results, though: result 24572039, 24560201, 24543954, and 24528355 all didn't create any protein structures "0 (nstruct)", and so are useless for the project. They all contain the message "Too many restarts with no progress. Keep application in memory while preempted". So it looks like you suspend BOINC/Rosetta a lot, perhaps while using your computer for other things ? The problem with this is, if you don't leave the application in memory, each time you suspend Rosetta, you loose all the work done since the last checkpoint (they aren't all that often - I think every 20 min or so) and it may happen that the calculation doesn't manage to advance because it doesn't reach its next checkpoint before being suspended again. Anyway, I guess you really should set "leave in memory" in your preferences as I suggested in the previous post.

So, good luck with your Rosetta activities. Perhaps we finally will manage to get some competitive spirit going in this team - I might upgrade one of my computers, for example... ;-)

-H.

Ok, I've activated that option, let's see if it helps. I'm glad that the occassional errors doesn't mean problems :-).

McCoy:

Ok, I've activated that option, let's see if it helps. I'm glad that the occassional errors doesn't mean problems :-).

Good news, McCoy ! I looks like the errors on your fast computer may not be due to overclocking after all: I searched the Rosetta number crunching forum and found a number of recent posts discussing the exit code 0xc000000d errors that you seem to have. See here, here, here, and here. The error only seems to happen on windows machines with certain graphics cards when the screen saver is running and the Rosetta team knows about this and is working to fix it. So it looks like for the time being, turning off the screen saver should prevent the errors. Thought you might want to know... -H.

Well my Dual Pentium with 4GB of ram is very stable but it just gave me this error 1073747819 (0xc0000005))

?

Russell:

Well my Dual Pentium with 4GB of ram is very stable but it just gave me this error 1073747819 (0xc0000005)) ?

Hi Russell,

first of all, welcome to the team - great to have another Betterhumans member with lots of "crunching power" in the team ! ;-) Now about your first error: I am not sure why that happend - however, as in the case of McCoy's initial errors, my general advice would be to go to your "general preferences" page and set Leave applications in memory while preempted? to "yes". See the message Too many restarts with no progress. Keep application in memory while preempted. on the result page of your second WU and the explanation in my response to McCoy (third paragraph). You also seem to have restarted your first WU several times before it crashed, so I suspect leaving the application in memory might perhaps have prevented that error...

-H.

K done that, seems more stable.

Having a look round at the statistics and benchmarks my old 2.4GHz Althlon with 1GB ram would beat my new dual Pentium with 4GB ram? Very odd way this program is optimized.

Russell:

Having a look round at the statistics and benchmarks my old 2.4GHz Althlon with 1GB ram would beat my new dual Pentium with 4GB ram? Very odd way this program is optimized.

The benchmarks are per thread. So you need to multiply yours by two. This makes your computer 47% faster than Joelstephensons's Athlon XP 3200+ and still slightly faster than McCoy's Athlon 64 3200+, which he says is heavily overclocked. Running 100% Rosetta 24 hours a day on your computer would give you about 380 credits/day. ;-) See here if you wonder about some of the benchmarks on the stats pages. The Rosetta team has designed their own benchmark to be run inside their Rosetta application to create a more level playing field. Current plans are to implement the new system after CASP ends in August...

Well I'll not be earning 380 credits a day, my computer is for high end rendering and photoshop work. I've knocked it down Rosetta using one thread 80% efficiency some of the time so it will be patchy progress. Every little helps though.

Do the joint Boincs projects have days when they try to break the Flops record for distributed processing? I mean everyone push for a few hours to benchmark the entire grid? It would be cool to beat the mainstream super computers to the petaflop record!

I'm interested in the individual projects like folding but see the entire distributed project as more important, so many computers out there waiting to be tapped, better software more projects like AI simulation etc.

Russell:

Do the joint Boincs projects have days when they try to break the Flops record for distributed processing? I mean everyone push for a few hours to benchmark the entire grid? It would be cool to beat the mainstream super computers to the petaflop record!

I'm interested in the individual projects like folding but see the entire distributed project as more important, so many computers out there waiting to be tapped, better software more projects like AI simulation etc.

With currently about 400 tereflop for all of BOINC, a petaflop would be quite an ambitious goal, even if only over a short period of time - perhaps in a year or two with some help from Moore's Law. While a DC project that does AI simulations would be a cool thing, I suspect AI projects would require much more tightly coupled processors than what is currently possible over the Internet, at any rate some kind of P2P BOINC would be required where the processors talk to each other directly instead of to a central server. If the BOINC team developed something like that it probably would also be useful for other projects, like protein folding - which I happen to think is just as cool as AI. ;-)

Great, another new team member - hi Florin ! The team currently seems to attract new members at a rate of about one per week. Lets hope we will be able to keep this up for a while. The next one will be number 10...

This week the daily production of the team for the first time hit the 1000 credits per day mark and I think we are now firmly established among the top 50 international teams !

Thanks to all who contributed to this - plus a special thanks to McCoy who currently provides almost half of our daily production !

Follow these instructions if you want to contribute your spare computing cycles to Rosetta@home. To join the Betterhumans team, go to our team page and click on join.

Previous posts:

What is Rosetta@home ?
Lecture on Rosetta@home
Letter to Rosetta@home participants: CASP
David Baker, 2004 Feynman prize
Rosetta@home is not for profit
Rosetta@home vs. Folding@home comparison
AP/CNN article on Rosetta@home
Design proteins and enzymes on your computer

Imagine having the power to create a brand new protein - a biosensor for a small molecule, say, or a novel enzyme - on demand. It's not pure fantasy. Computational structural biology is poised to put this power into our hands.

This is the first paragraph of the title story of the July issue of The Scientist in which Rosetta@home lead scientist David Baker outlines his group's recent advances in protein structure prediction and design and what lies ahead for Rosetta@home after the completion of the CASP structure prediction trials in early August. The full article is available here (PDF).

The article closes with the sentence: With the power to create unheard of catalysts, improved biomolecular interactions, and genuinely useful new proteins, I consider myself privileged to be part of a field limited merely by imagination and computing power.

It is this last part where our help is required to make sure the scientists have access to the computing resources they need to work towards these goals. Follow these instructions if you want to donate your spare computing power to Rosetta@home. To join the Betterhumans team, go to our team page and click on join.

Hey, I'm glad that the team is doing good and that my contribution has weight on it. Since the start of my  holidays I've been using my computers less often so maybe that's why my contribution has grown.

Also, disabling the screensaver solved the problem I had with my Athlon 64 throwing errors. Now it works like a charm, although it's less fancy :P

Unfortunately, I'm leaving to Russia this friday (14th July) and won't return until 4th August, so my computers will be turned off during that period.