View Full Version : Project on Schedule?
Auritania
04-25-2002, 08:40 PM
I was reading through the DF site and took particular notice of the schedule for the phases. If I am reading things correctly, The project ultimately is aimed at conformers for CASP 5 in early 2003. Between now and then, the plan is to progress through:
Phase Ia : 1 Billion ea. of 5 knowns - Complete by Mar2002 - DONE
Phase Ib: 10 Billion ea. of 10 - Complete by June 2002 - Current
Phase II : 10 Billion of 10 CASP 4 structures - Complete by Sept 2002
Phase III: 10 Billion of unknowns - Complete by early 2003
If I understand this all correctly and we are in fact on protein 1 of 10 in phase Ib, we have some 200,000,000,000+ conformers to complete before phase III can start. At the current rate of 62Mil/day (Den's stats) that means there are some 3278 days=8.9 years before completion? That would mean that the first True Blind would be completed sometime in the year 2011. Thats a little late for CASP 5, isn't it? Am I totally missing something here? Was there/Is there really an expectation of 800+ million conformers per day?
Could I just be that bad at math?
Paratima
04-25-2002, 11:48 PM
We were hoping that you would buy a few more boxen to put on the project! :p
IronBits
04-26-2002, 12:22 AM
More like you forgot to take into consideration that processors get faster all the time, and more people join up to help out ;)
If all was static, then I suppose the math might be correct.
I'm sure the Hammers being released this fall, will provide a substantial performance gain over what we use now.
Auritania
04-26-2002, 08:21 AM
Originally posted by Paratima
We were hoping that you would buy a few more boxen to put on the project! :p
Umm.. I have been ! :p Scratching up some more cash on eBay to buy box #16 to put into my minifarm.
I know things get faster and people join but I guess my real point was that I don't think that the time frames will hold. Especially in the near term. It sounds like there is a deadline in CASP 5. It would appear, on the surface, that there is no way short of the project increasing output more than 10 fold in the next week or so to meet it.
Since it looks like at least one of the goals of the project is to submit unknowns for CASP 5, then there would need to be some sort of adjustment in the plan to make that possible. Looking forward from there one also must wonder if the DF project can/will exist past there. Is this a project with upper level support to last 2, 3 or more years?
What is the goal of the project if the schedule can't be satisfied? Will DF continue to exist for however long it takes, or is this a means to an end? Without being able to reach that end will it terminate before the plan plays out? I'm on this train and plan on staying on it, but like every little kid I just want to know "are we there yet?" and if not, when will we get there. Where is there anyways?
Brian the Fist
04-26-2002, 11:04 AM
Do not fear. The schedule posted was only tentative and was made up before we even begun. CASP is indeed a scheduled event and so will take precedence over the other phases when the time comes. Not all CASP targets will be predicted through distributed folding though, some will be done by other methods not requiring distributed computing (basically, the 'harder' ones will be reserved for DF). The project will continue beyond this though. And it is not out of the question that our user base will multiply ten-fold in the near future... :smoking: We are in the process of modifying the back end to allow it to scale to more users reliably and so have avoided massive recruitment of users until we are sure we can handle the influx ;) So far the response has been great though.
MAD-ness
04-29-2002, 06:28 PM
I think that adding some additional stability/features/stats to the client (not talking about anything specific here, just a general evolution of the client to be more friendly and user friendly) and then ramping up advertising/recruiting for the project could allow an increase of...well...a lot.
Up to this point, it has still been beta testing and there hasn't been active recruitment. A few of the DC teams have decided to put a lot of energy into the project, but far from the majority of said teams. Team Stir Fry from Ars is essentially a team composed of members who are newer to Ars or who specifically focus on Distributed Folding, rather than the mammoth producers like Ars has in projects like G@H and SETI. I know that Free-DC could ramp up a lot if some of thier larger farmers were motivated to install the client on thier farms and keep it running. DSLReports (Team Endeavor) and DPC both have HUGE distributed computing communities and could bring hundreds of Ghz to this project in a very short amount of time if they so desired.
I guess my point is that Distributed Folding is still pretty unknown and is not yet considered a "major" distributed computing project by the major DC teams, nor is it something that has pushed on the 'masses' like Folding@home and SETI@home. As this project matures and things begin to ramp up, I think that this can definitely change..and quickly.
Digital Parasite
05-03-2002, 11:29 AM
Since the CASP5 prediciton season starts in May (any day now I guess) and goes until August does that mean all computing and results have to be finished by the end of August so the results can be discussed at the December conference/meeting?
Or when is the actual deadline for having the predictions finished for CASP5?
Do you have any idea when we might start computing CASP5 stuff?
Thanks.
Digital Parasite
05-06-2002, 10:56 AM
Howard? Anyone else out there that can reply to my quesiton ^^^
Thanks.
Brian the Fist
05-06-2002, 01:42 PM
There is a link to the CASP 5 web site on the About section of the web site, you can read more about CASP there (or just search google)
Digital Parasite
05-06-2002, 01:56 PM
Originally posted by Brian the Fist
There is a link to the CASP 5 web site on the About section of the web site, you can read more about CASP there (or just search google)
Thanks, I had already visited the CASP5 web site linked from your page before I posted the question. At that time there was very little information, I see it has more now. I had a few questions I was hoping you could clarify from the information on their web page.
On the target list, they will show an expiry date. I'm guessing that all the prediction results have to be submitted for the protein prior to the expiry date. In the past do the expiry dates have a wide difference from each other ie: some expire weeks/months after others?
Since the target list is currently being created/posted and the DistributedFolding project is only working on the first protein in Phase 1b, are we going to start working on the CASP5 proteins as soon as they are available or are you going to wait predicting how long it will take to compute so we can do as much of the earlier phases as possible first?
Thanks.
Powered by vBulletin® Version 4.2.4 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.