Good for Rosetta@home?

Message boards : Number crunching : Good for Rosetta@home?

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
The_Bad_Penguin
Avatar

Send message
Joined: 5 Jun 06
Posts: 2751
Credit: 4,271,025
RAC: 0
Message 48223 - Posted: 1 Nov 2007, 11:32:44 UTC - in response to Message 48216.  
Last modified: 1 Nov 2007, 12:24:54 UTC

(1) PS/3 is open platform. Good for Rosie.

(2) PS/3 price has decreased from $599 to $399. Good for Rosie.

(3) PS/3 performance (at F@H) has increased from 20 Gflops to 35 Gflops (and may yet increase again). Good for Rosie.



And now with the new 65nm Cell BE...

(4) PS/3 power usage decreases from 200 watts to 135 watts. Good for Rosie.

(5) PS/3 can handle double precision calculations in the 128 bit registers, reaching double precision 100 GFLOPs. Good for Rosie.



However, I believe Rosetta changes more rapidly than the folding software, and the cell each have a limited amount of memory which might restrict rosetta.


OK, I can accept these two issues as valid concerns, which may or may not be overcome.

So long as we know what the concerns are, we can have an intelligent discussion about them.


Let's talk about these two issues.

What other potential "cons" exist (with the new 65nm architecture) ?




Now that F@H has got 'the market', as it were, it might be hard for R@H to capture the gains f@h did... perhaps better to aim for xbox or something, especially when to the casual observer the two projects seem quite similar.


I'm not so certain that this is a valid reason for Rosie to not pursue the PS/3, or to go for the xBox 360 instead.

(A) I know the xBox has "overheating" issues (at least one billion dollars worth). I don't know enough about the xBox that it's cpu brains ( a xeon?) is (i) open platform for running alternative software, or (ii) is capable of double precision.

From wikipedia:

the Xbox 360 uses the triple-core IBM designed Xenon as its CPU


and

The Xbox 360 can be subject to a number of possible technical problems. Since the console's release in 2005, the console's reliability gained a bad reputation by the general public, and in the press through articles portraying its poor reliability and relatively high failure rates.



(B) And the "marketing" issues involved of using an xBox instead of the PS/3:

"there's no way MS are gonna release something that maxes the CPU when there's already so many overheating problems! Also, it would look to the uninformed observer that the xbox isn't anywhere near as powerful as the PS3 when compared to the number of PCs it replaces."



(C) The argument could also be made that one is BOINC and one isn't, and that is "separation" enough.

There are (will be) PS/3's in the millions. F@H currently has about 40,000 active units.

And some of those may prefer to crunch for Rosie instead, if such an option was available.

(I'm purchasing the new PS/3, and going to F@H 'cause its the only game in town. If Rosie goes PS/3, I'd switch.)


MS were talking with the bakerlab about the 360 but there's no way MS are gonna release something that maxes the CPU when there's already so many overheating problems! Also, it would look to the uninformed observer that the xbox isn't anywhere near as powerful as the PS3 when compared to the number of PCs it replaces.


In spite of Bill G's "donations" to Rosie, the xBox (imho) just isn't up to snuff. Heat issues, etc.

The Cell BE's are octocores (actually 9 cores) being used by IBM in bladeservers and supercomputers.

ID: 48223 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Michael G.R.

Send message
Joined: 11 Nov 05
Posts: 264
Credit: 11,247,510
RAC: 0
Message 48231 - Posted: 1 Nov 2007, 18:06:47 UTC

I can imagine a future (next generation of consoles, maybe?) where all consoles come from the factory with a menu for distributed computing and users just have to pick a project and some parameters (of course there would be info about potential heat issues and electricity usage so that people make informed decisions).

Reducing the barrier to entry into distributed computing is extremely important if we want it to scale well.
ID: 48231 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 112,523,185
RAC: 43,827
Message 48234 - Posted: 1 Nov 2007, 18:24:11 UTC - in response to Message 48216.  
Last modified: 1 Nov 2007, 18:29:37 UTC

MS were talking with the bakerlab about the 360 but there's no way MS are gonna release something that maxes the CPU when there's already so many overheating problems! Also, it would look to the uninformed observer that the xbox isn't anywhere near as powerful as the PS3 when compared to the number of PCs it replaces. Of course Cell is responsible for graphics duties in the PS3 so the xenos would need to be maxed to make a fair (but pointless!) comparison, and I imagine coding for xenon would be as difficult as for any GPU...

i got it the wrong way round:
Xenon = CPU (3 core PPC based)
Xenos = GPU (unified shader)

The xenos has access to more memory than the PS3's Cells and I believe it's related to the current ATI GPUs so might be easier to program for.
ID: 48234 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
The_Bad_Penguin
Avatar

Send message
Joined: 5 Jun 06
Posts: 2751
Credit: 4,271,025
RAC: 0
Message 48242 - Posted: 1 Nov 2007, 19:27:40 UTC - in response to Message 48231.  

I'm not convinced that the xBox 360, given reported heat and reliability issues, is ready to take on DC projects.

Don't get me wrong, anything to expand DC is good. I just don't think MS consoles are up to it at present.

Sony PS/3's makes DC with F@H as easy as a few clicks, uses the Cell BE (used in IBM bladeservers and supercomputers), and (as of yet) does not have the same heat and reliability concerns that the xBox 360 has.

Let's concentrate on the hardware issues (not software / programability) for the moment.

How can we learn what Rosie's memory requirements are?

Who in the project can make a definitive statement, one way or the other, that given Rosie's uniqueness, Sony PS/3 and its new 65nm Cell BE either can or can't be used, strictly on the basis of hardware (memory) potential / limitations?
ID: 48242 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 112,523,185
RAC: 43,827
Message 48244 - Posted: 1 Nov 2007, 19:50:51 UTC - in response to Message 48242.  

I'm not convinced that the xBox 360, given reported heat and reliability issues, is ready to take on DC projects.
The 360 has just gone 65nm too - the new ones should be fine ;)

ID: 48244 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
The_Bad_Penguin
Avatar

Send message
Joined: 5 Jun 06
Posts: 2751
Credit: 4,271,025
RAC: 0
Message 48246 - Posted: 1 Nov 2007, 20:05:31 UTC - in response to Message 48244.  
Last modified: 1 Nov 2007, 20:07:09 UTC

How can we learn what Rosie's memory requirements are?

Who in the project can make a definitive statement, one way or the other, that given Rosie's uniqueness, Sony PS/3 and its new 65nm Cell BE OR MS xBox360 and its new 65nm cpu either can or can't be used, strictly on the basis of hardware (memory) potential / limitations?

Don't necessarily want to turn this into a PS/3 vs xBox360 thread, but...


What about being an open platform? Sony's PS/3 is:

http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9043942&source=rss_news50

http://gravity.phy.umassd.edu/ps3.html

http://moss.csc.ncsu.edu/~mueller/cluster/ps3/

http://gametomorrow.com/blog/index.php/2007/04/03/ps3-clusters-2/


And I think MS xBox isn't:

This makes the XBOX an unfit solution as a replacement for a personal computer or a cluster node.

After several days of configuration and set up, we finally got our cluster up and running. Costs of the cluster were a little higher than we had originally anticipated; we bought a hard drive blaster, a switch and various cabling, duct tape and shelving. The practicality of our cluster turned out to be fairly negligible; had we seen some really outstanding performance, we probably would have been able to justify the hours of work and configuration.

I'm not convinced that the xBox 360, given reported heat and reliability issues, is ready to take on DC projects.

The 360 has just gone 65nm too - the new ones should be fine ;)

ID: 48246 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 112,523,185
RAC: 43,827
Message 48247 - Posted: 1 Nov 2007, 20:16:50 UTC - in response to Message 48246.  

And I think MS xBox isn't:

This makes the XBOX an unfit solution as a replacement for a personal computer or a cluster node.

After several days of configuration and set up, we finally got our cluster up and running. Costs of the cluster were a little higher than we had originally anticipated; we bought a hard drive blaster, a switch and various cabling, duct tape and shelving. The practicality of our cluster turned out to be fairly negligible; had we seen some really outstanding performance, we probably would have been able to justify the hours of work and configuration.

That article/quote is referring to original xboxes I believe. You're right that you'd need MS's permission/help to port to the 360 though.

DB mentioned that the Rosetta code has recently been cleaned up and the memory requirements have been reduced (although that might mean they just want to run bigger models in the same memory!)... I guess the minimum memory requirements at the moment are quite clear - around 300MB (inc VM) per core for current jobs. I never got a definitive answer as to whether parallel processing is possible/beneficial as three cores running on the same model might have a lower memory requirement than three separate models.
ID: 48247 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
FluffyChicken
Avatar

Send message
Joined: 1 Nov 05
Posts: 1260
Credit: 369,635
RAC: 0
Message 48248 - Posted: 1 Nov 2007, 20:18:48 UTC
Last modified: 1 Nov 2007, 20:20:39 UTC

360 unfit, nah... if we can get them, then all the better.. We have a wealth of new computers. It dhoulf be capable of running at 100% no problem else there is a major design flaw ;)

As for PS3, so if F@H has the forfront, what's wrong with jumping on the bandwagon and grabbing 5% of that... 5% more than we had before.


For me though, if we could have a Wii version and interactive screensaver/game that would be great. I would buy on tommorow... Folding with the Wii-mote sound good to me. It would even beat Twister :D
Team mauisun.org
ID: 48248 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 112,523,185
RAC: 43,827
Message 48250 - Posted: 1 Nov 2007, 20:27:03 UTC - in response to Message 48248.  

360 unfit, nah... if we can get them, then all the better.. We have a wealth of new computers. It dhoulf be capable of running at 100% no problem else there is a major design flaw ;)

They should be able to, but probably can't all handle it! The 65nm ones are a different story though.

For me though, if we could have a Wii version and interactive screensaver/game that would be great. I would buy on tommorow... Folding with the Wii-mote sound good to me. It would even beat Twister :D

I think the Wii would be a great target - they're not that powerful compared to the PS3 and 360 but there's loads of them and it might be easier to port to...
ID: 48250 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
The_Bad_Penguin
Avatar

Send message
Joined: 5 Jun 06
Posts: 2751
Credit: 4,271,025
RAC: 0
Message 48252 - Posted: 1 Nov 2007, 20:37:24 UTC - in response to Message 48247.  
Last modified: 1 Nov 2007, 21:34:12 UTC

Who is the Rosie guru and sage who could provide such an answer?

I think with the switch to 65nm, it is now to the point where Rosie can no longer ignore the potential offered by gaming consoles and their installed base.

If Rosie won't parallel (or still requires "large" amounts of memory), that's fine. But lets somehow get an informed, reliable, answer.

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."




And why wouldn't such a strategy work on gaming consoles (PS/3, xBox360, Wii) ???




I never got a definitive answer as to whether parallel processing is possible/beneficial as three cores running on the same model might have a lower memory requirement than three separate models.

ID: 48252 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 · 2

Message boards : Number crunching : Good for Rosetta@home?



©2024 University of Washington
https://www.bakerlab.org