Message boards : Rosetta@home Science : Rosetta & Parallelization (gaming consoles)
Previous · 1 · 2
Author | Message |
---|---|
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Apparently myself and other donors have been expecting too much in terms of the Project being willing to communicate with us... The silence remains deafening... I do wish the project people would communicate a bit better as far as optimizations go, though. Just giving us a status update would be much appreciated. Agreed. All I am suggesting is that at some point in the Cost Benefit Analysis, there H-A-S to be a point where the expenditure of resources is justified in porting Rosie to PS/3. What is the Project's answer to this question? Rosie is currently about 60 tflops. So, 10x that, 600 tflops? 1 pflop? F@H is already here. 60 pflops, 1000x where Rosie currently is at? I can't believe the Project would say the expenditure of resources could not be justified for a yottaflop !!! Spending time writing better scientific code can have more value than spending time optimizing for SSE or porting to GPU because it allows the code to do things that it couldn't otherwise do, even with more FLOPS, and thus can help discover medical breakthroughs. |
Luuklag Send message Joined: 13 Sep 07 Posts: 262 Credit: 4,171 RAC: 0 |
Apparently myself and other donors have been expecting too much in terms of the Project being willing to communicate with us... well guys if we have the code for rosetta, and i guess we could get some code or help from F@H or from Sony maybe a group of volenteers could port it over using linux.? |
![]() ![]() Send message Joined: 13 Mar 06 Posts: 158 Credit: 417,178 RAC: 0 |
Apparently myself and other donors have been expecting too much in terms of the Project being willing to communicate with us... I am a computer programmer. I do software development. I work for a medium sized business (appx. 70 employees, appx. $100 million in revenues). A recently completed scope document for a medium sized project consumed over 200 hours of the output of our developers. The task was completed with an excellent balance of speed and thoroughness, by individuals of high integrity and outstanding work ethic. We devoted this time to the scope document so that we could accurately and honestly "communicate with" upper management about the project in question. That communication had a measurable cost, which somebody had to pay. I do not presume to know the resources available to the development shop of the Rosetta@home project, but I do presume to know that even assessing the scope of the task of porting the code base to gaming consoles is non-trivial. Tossing out huge numbers for potential scaling benefits is a rhetorical device that completely ignores the technical questions involved. Who will pay the cost of answering those technical questions? Rosie, Rosie, she's our gal, If she can't do it, no one shall! |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Who paid the "costs" of answering the question if Rosetta could be adapted to the Boinc platform? Were the benefits (60 tflops) worth the costs? How much "cost" is honestly involved in answering the following questions, in terms of a generalized, order-of-magnitude, response (not a response that requires precision to 1 x 10^-10, or to the penny)? Does not the University of Washington have a Computer Science department? Could not those resources also be utilized, for answering questions? Would it take 200 hours of "cost" to determine if the Rosetta code could / should be ported over to a Timex / Sinclair? A Commodore Vic 20? An Apple ][+ ? An intel 386? An intel 486? An intel pentium? An intel quadcore? A Cell BE? Sometimes a "quick-and-dirty" answer can be just as valuable, if not moreso, than a 200-hour answer. (1) Given that the Project, iirc, has already had preliminary discussions with Sony, is the size of the PS/3's on-board memory a limiting factor so as to prevent the Rosetta code from running on this hardware? Yes or no? (2) In a generalized, order-of-magnitude, response: What is the point at which the Project will consider the expenditure of resources justified in porting Rosie to PS/3? 60 tflops? 600 tflops? 6 pflops? (3) Given the below, in general terms (based upon the below experience), approximately, in an order of magnitude response, how much effort to port existing parallel Rosie code to spe's of the PS/3, given that Rosie sequential code was ALREADY converted to parallel for the IBM supercomputer? 20 hours? 200 hours? 2000 hours? For example, how was this possible? "When David Baker, who also serves as a principal investigator for Howard Hughes Medical Institute, originally developed the code, it had to be run in serial - broken into manageable amounts of data, with each portion calculated in series, one after another. Through a research collaboration, SDSC's expertise and supercomputing resources helped modify the Rosetta code to run in parallel on SDSC's massive supercomputers, dramatically speeding processing, and providing a testing ground for running the code on the world's fastest non-classified computer. The groundbreaking demonstration, part of the biennial Critical Assessment of Structure Prediction (CASP) competition, used UW professor David Baker's Rosetta Code and ran on more than 40.000 central processing units (CPUs) of IBM's Blue Gene Watson Supercomputer, using the experience gained on the Blue Gene Data system installed at SDSC." I do not presume to know the resources available to the development shop of the Rosetta@home project, but I do presume to know that even assessing the scope of the task of porting the code base to gaming consoles is non-trivial. |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
EDIT --> Duplicate post. |
Luuklag Send message Joined: 13 Sep 07 Posts: 262 Credit: 4,171 RAC: 0 |
so the code that they have written, could also be use to use all cpu's of a ps3 parallel. cause that code is already written. so just the porting over keeps as a problem. Who paid the "costs" of answering the question if Rosetta could be adapted to the Boinc platform? Were the benefits (60 tflops) worth the costs? |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
maybe. the Project isn't commenting. so the code that they have written, could also be use to use all cpu's of a ps3 parallel. cause that code is already written. so just the porting over keeps as a problem. |
Klimax Send message Joined: 27 Apr 07 Posts: 38 Credit: 2,524,105 RAC: 0 |
But they did comment,in various threads,thay said why it was not possible. And for them(Admins,mods,...),I SUPPOSE(!),it is for now closed matter...until the code is more stable... |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Would you kindly link to any posts from the Project staff that explains why a port to the ps/3 is not possible? Thank-you. But they did comment,in various threads,thay said why it was not possible. |
Klimax Send message Joined: 27 Apr 07 Posts: 38 Credit: 2,524,105 RAC: 0 |
A list,which I suppose you know: https://boinc.bakerlab.org/rosetta/forum_thread.php?id=3234&nowrap=true#48331 It was before app 5.90 After larger dig-op I found this... :-| : https://boinc.bakerlab.org/rosetta/forum_thread.php?id=3234&nowrap=true#48031 Most probably i missed post,where it was stated quite clearly,but I do not have time to search for that. And then with such amount of changes to app,it is not possible to properly maintain ports.See EAH,where optimization is missing as well,altough the code has been crafted and is present ONLY in linux version and only as experimental.Why?Because of rounding errors and remaining bugs. Just take a look in "Problem with..." threads... |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Thanx for your response, but I note the following: (1) Post #1 is from "Admin" in the capacity of a Message Board Moderator cleaning-up a thread that was getting off-topic. (2) Post #2 is from DB, and only mentions the code being made smaller, and now potentially ready for sse optimization. No discussion about parallelization, or potential memory limits of the ps/3. (3) Neither of these threads addressed the question that was asked: Would you kindly link to any posts from the Project staff that explains why a port to the ps/3 is not possible? (4) You likely "missed" the post where it was allegedly "stated quite clearly" because iirc no such post exists. (5) Even if you did "have time to search for that", imho you would be searching for something that does not exist. (6) Ergo, I respectfully disagree with your original statement: "But they did comment,in various threads,thay said why it was not possible." (7) I would be very happy for someone to prove me wrong. I've searched, again, and can't locate what you claim exists. Most probably i missed post,where it was stated quite clearly,but I do not have time to search for that. This may, or may not, be true for Rosetta code. Let's first determine if it is even theoretically possible to run Rosie code on ps/3 and gpu. Then, and only then, we can start discussing potential issues of code maintainence. And then with such amount of changes to app,it is not possible to properly maintain ports. |
Klimax Send message Joined: 27 Apr 07 Posts: 38 Credit: 2,524,105 RAC: 0 |
Thanx for your response, but I note the following: General: May be may be not,searching is not always of help(in responses the "keyword" is missing) and number of posts between now and then is huge,so even manual way is not possible. And I did not include those,which I supposed you read. Perpoints: 1) That was for start and I knew that,was only reference point. (2)-(6) addressed above 7)You likely used same or similar keywords as I.That is why we easly could missed that. And than what is more important:Fixing problems and include more targets or "freezing" code for optimization with possibility of future making code unnecessary or porting and having another load of problems.Only projects with huge staff and money or with relativly slowly changing code can afford porting. And have you seen any "problems with" threads?I "monitor" them and would say that there are still some nice problems to solve... And finally have you tried to post in thread "Moderators Contact Point"?Don't remember seen any post by you there... |
The_Bad_Penguin![]() Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
The burden of proof is upon the person making the claim. You made the claim that: "But they did comment,in various threads,thay said why it was not possible". And I could make the claim that the moon is made out of green cheese. If you can't support your claim with evidence, I suggest that you withdraw your claim. You likely used same or similar keywords as I.That is why we easly could missed that. Why don't we learn the comment to that statement from Doc Baker, when he commented over 0.75 years ago "But they did comment,in various threads,thay said why it was not possible." Why don't we learn the comment to that statement from Doc Baker, when he commented over 1.25 years ago Only projects with huge staff and money or with relativly slowly changing code can afford porting. WOW, TWO POSTS FROM THE DOC HIMSELF, STATING THAT A PORT TO A GAMING CONSOLE IS POSSIBLE. TOO BAD THE MOST RECENT OF THE TWO IS NEARLY 9 MONTHS OLD. Why would I bother posting there? And finally have you tried to post in thread "Moderators Contact Point"?Don't remember seen any post by you there... |
Klimax Send message Joined: 27 Apr 07 Posts: 38 Credit: 2,524,105 RAC: 0 |
The burden of proof is upon the person making the claim. You made the claim that: "But they did comment,in various threads,thay said why it was not possible". And I could make the claim that the moon is made out of green cheese. If you can't support your claim with evidence, I suggest that you withdraw your claim. Burden of proof-clear... But why do bother to post in that forum??? When you claim that project people or even mods won't say anything then that is the best way to get their attention even if only to get it relayed futher. If you do want to know ,then contact them,otherwise they are occupied by people having problems with current version...And they(or mods not sure) somewhere said they do not mostly monitor threads except those with "Problems with..." and such.(I claim,but no time to search for post...) P.S.:I do no longer have time to search and most probably to reply,because I have to learn... |
Message boards :
Rosetta@home Science :
Rosetta & Parallelization (gaming consoles)
©2024 University of Washington
https://www.bakerlab.org