21)
Message boards :
Number crunching :
Problems with rosetta 5.48
(Message 37464)
Posted 5 Mar 2007 by tralala Post: Hi all, the "HINGE" WUs are simulating a very large protein which has more than 800 residues ( versus less than 200 residues normally we have run on BOINC ) and thus requires much more memory than usual. We have put a higher memory requirement on these jobs. Also, high priority has been assigned because it is for a blind docking prediction with the deadline coming soon. That can explain why some low-memory clients can not receive jobs temporarily. I recommend telling in advance if WU with bigger memory requirement are to be sent out (preferable as a news item together with a warning that it might cause more errors and some problems). People don't complain if they know the cause and the need for (deadline, competition, etc.). So much for the social aspects of DC. |
22)
Message boards :
Rosetta@home Science :
amyloid fibril structure prediction
(Message 31143)
Posted 14 Nov 2006 by tralala Post: So you start with Abeta(10-40) which means 30 structures. When do you think you will have established those 30 structures and what's the plan afterwards? Do you have a definite set of amyloids you want to study or does it all depend on the progress you make? How accurate do you want to predict the structure? I understand high-resolutions means pretty accurate but is you goal a prediction as accurate as X-ray crystallography (<1.5 Angstrom)? |
23)
Message boards :
Rosetta@home Science :
CASP7 T0363 top predictions posted
(Message 28477)
Posted 25 Sep 2006 by tralala Post: Hi fluffy, The resolution of 1.65 Angstroms is pretty close to the native structure and almost within the error margin of a crystalized structure (which is about 1.5 Angstrom iirc). So the question shouldn't be why is it not exact but why is it so damn close. I don't know fur sure but I think with any structure being withn 2 Angstroms the project staff is pretty happy. |
24)
Message boards :
Number crunching :
Closed to all, but those with stinky feet
(Message 28299)
Posted 23 Sep 2006 by tralala Post:
I never thought I would agree on more than a sentence of one of your posts, but here I have to admit, that I agee to all of those. :-) Concerning the BOINC benchmark and it's unbalanced values across different platforms I just want to say that I'm happy Rosetta now has a work based credit system which equalizes such disparities. Concerning the optimized boinc clients I admit I used them as well (as I have stated on several occasions) and still don't view it as cheating although it certainly increased the disparities. I'm happy that now we have a credit system which is neutral to the version of the boinc client (and personally wouldn't mind to be backdated to "true" values). Concerning the discussion of backdating or zeroing out I agree that it is a dead issue, since as kevint pointed out the project made a decision and (hopefully) won't revoke it. However the discussion has some academic value. ;-) |
25)
Message boards :
Number crunching :
Closed to all, but those with stinky feet
(Message 28249)
Posted 23 Sep 2006 by tralala Post: @XS_VS I don't think Saenger qualifies as the average cruncher, he is sharing his rather slow computer between various projects. I calculated the average cruncher produces about 100 new_credits/day, which means 500 of them produce 50k new_credits/day.
UW-Madison was active on other projects for a long time. He included Rosetta in his mix just after the new credit system was invented. It is not unlikely that this is connected. Of course it's speculation, but I just wanted to point out, that new top-crunchers come occasionally.
Well that is the core of our disagreement. I just recall it radically different. IMHO the project did not follow a "vocal minority" but did what was the wish of the overwhelming majority (fair work based credit system). There was a debate whether to backdate and/or whether to grant more like the standard boinc client or like the optimized one. In the course of the debate some called users of 5.5.0 cheaters and some of XS and others flamed back. The project than listened to your complaints, forbid cheating accusations, retreated on the idea of backdating and Dr. Baker made a special appearance in your board, however you were angered to a level where you demanded the condemnation of specific posters which was inappropriate.
... and many other teams have upped their contribution by varying amounts. In fact it can't be denied that the project lost a huge resource but I think it'll pay off long-term. If the new credit system would not have been invented, or Dr. Baker would have taken sides in favor of XS or the project would now grant 3 times the credit it does now it would have created more harm than to loose XS (and Teddies). Between various bad the project choose the best option (but did many mistakes in the course of the events). |
26)
Message boards :
Number crunching :
Closed to all, but those with stinky feet
(Message 28243)
Posted 23 Sep 2006 by tralala Post: @XS_VS While I agree that not all crunchers contribute equally to the work done and top-crunchers are more valuable to the project from a strict crnching-time-perspective I disagree with your numbers: a) XS got about 8-10% of the granted credits - in real work done it was probably closer to 3%. b) To replace a top-cruncher like PY222 or DDTung you don't need about 33.000 average cruncher but more likely about 500-1000 or one other top-cruncher like UW-Madison CAE who joined AFTER the change to the new credit system. Teams and top-crunchers are valuable to a project but I came to the conclusion not as much as one could think at first. Teams and top-crunchers often (not always) change the project and if one team reduces his output another one ups it. There are many teams which upped their production AFTER the credit system change (Czech-National, TSC! Russia, Planet3D etc.). I don't want to imply with what I said that a project should not care for his top-teams and top-crunchers and for those mostly interested in the discussion but it should never subordinate to the demands of any team or cruncher even when he thinks he is irreplaceable. |
27)
Message boards :
Number crunching :
Closed to all, but those with stinky feet
(Message 28176)
Posted 22 Sep 2006 by tralala Post: OK I'm awake now. [ ] You have read the post. Tony speaks about the 24th August when the new credit system went into effect and not about February 2006 (which was the possible starting date to backdate). He also does not suggest recalculating the credits (backdating) but to seperate the credits until August, 24th 2006 and after. This is possible for sure but I don't think it will help the project. |
28)
Message boards :
Rosetta@home Science :
Quantum Leap in Protein Folding
(Message 28160)
Posted 22 Sep 2006 by tralala Post: It seems they are describing the actual folding process and how it happens according to the rules of physics. That is more what folding@home is doing - simulating the whole folding process of a protein. However they claim to reach the final, native structure with very few steps. I wonder if one can develop an algorithm for Rosetta based on their work, but I doubt. I participated in programmers dicussions about chess engines and know that most scientific papers of computer chess were of little worth for real engines because e.g. the claims of reducing the search space relied heavily on the actual engine used in the experiment and are most of the time not applicable to other engines. So it might well be that the reduction of steps achieved by Pederiva et al. are relative to just a random search and would not speed up Rosetta (which already uses some kind of restriction of the search space iirc). Ah, here is the snag: "Simulations using more sophisticated all-atom models are in progress and will clarify whether these are general features or are biases of the topology-based model adopted in this work." edit: In the paper Hoelderl1n linked to they studied one tiny protein, with just 36 amino acids. Whether there findings work for other proteins is not clear. |
29)
Message boards :
Number crunching :
No annonymous moderators
(Message 28068)
Posted 21 Sep 2006 by tralala Post: Mod.De is Parsec so is Parsec for Mod.DE! :-))) |
30)
Message boards :
Number crunching :
No annonymous moderators
(Message 28058)
Posted 21 Sep 2006 by tralala Post: Mod.De is Parsec As I have some insider information I'd like to comment on that. IIRC correctly you know from a mod that Mod.Tymbrini is neither tralala nor MattDavis. That is correct, but that does not say much about tralala being a mod or not. ;-) @carl.h From your list only one name is correct but it's attributed to the wrong mod. So now it's gonna really easy isn't it? |
31)
Message boards :
Number crunching :
problems whit Account Key
(Message 28035)
Posted 21 Sep 2006 by tralala Post: Yes, the key is there, but isn´t the key for the mail madog333 , the key that appears is the key for jfrf1978 (thats my new one). Hotmail obviously marks the mail as Junk. So ask for the key to be sent by email again and then check your junk folder (make sure you have not selected to delete junk immediately if that is an option at hotmail). |
32)
Message boards :
Number crunching :
Take the pledge:
(Message 28020)
Posted 21 Sep 2006 by tralala Post: I sure am not a mosderator. ;-) |
33)
Message boards :
Number crunching :
Another discussion on the New Credit System
(Message 28017)
Posted 21 Sep 2006 by tralala Post: The only way to solve the dilemma between backdating and not would have been to implement the new credit system in such a way that it would grant approximately what the 5.5.0 client claimed. That would have left old credits untouched but would have not curtailed the possibilities for teams to overtake each other. As a nice sideeffect we would produce about 120 Tflops - at least on paper. |
34)
Message boards :
Number crunching :
The best moderating tool this board can have is...
(Message 28014)
Posted 21 Sep 2006 by tralala Post: Yes that would be fine. But there is no private messaging - what now? We agree! :-) but there is no easy way to install one. For the boinc MB it does not yet exist and switching to another forum software is a) much work for the devs b) requires every poster to reregister and c) requires a lot of work to migrate some important info like the FAQ and interesting threads. |
35)
Message boards :
Number crunching :
Another discussion on the New Credit System
(Message 28003)
Posted 21 Sep 2006 by tralala Post: S That does not make Sense. ;-) |
36)
Message boards :
Number crunching :
No annonymous moderators
(Message 27999)
Posted 21 Sep 2006 by tralala Post: Are you stating you are not ? ;-) No I'm not (stating that), I just wonder how I became suspect. |
37)
Message boards :
Number crunching :
No annonymous moderators
(Message 27991)
Posted 21 Sep 2006 by tralala Post: Hey, why do you all think I'm a mod? Did I hint on that somewhere? |
38)
Message boards :
Number crunching :
Another discussion on the New Credit System
(Message 27987)
Posted 21 Sep 2006 by tralala Post: Even as a mod Lala you cannot prove or disprove who on what team ran opti can you ? Or from when !! No I can't, but I can state which statements I'm going to believe and which not. |
39)
Message boards :
Number crunching :
The best moderating tool this board can have is...
(Message 27973)
Posted 21 Sep 2006 by tralala Post: Yes that would be fine. But there is no private messaging - what now? |
40)
Message boards :
Number crunching :
Another discussion on the New Credit System
(Message 27971)
Posted 21 Sep 2006 by tralala Post:
Come on you don't believe this yourself. As biggles has pointed out there were instructions for the optimized client on your board since december 2005. Free-DC and DPC did use the optimized boinc clients all the time as well. |
©2024 University of Washington
https://www.bakerlab.org