Vistax64 WU?

Message boards : Number crunching : Vistax64 WU?

To post messages, you must log in.

AuthorMessage
Starsystem017

Send message
Joined: 6 Jul 06
Posts: 4
Credit: 11,699
RAC: 0
Message 47307 - Posted: 2 Oct 2007, 6:55:53 UTC

When will you start to send out Vista 64 bit workunits?
ID: 47307 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1832
Credit: 119,676,482
RAC: 10,619
Message 47311 - Posted: 2 Oct 2007, 9:21:03 UTC

i doubt they ever will, unless they need to access more memory...
ID: 47311 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Starsystem017

Send message
Joined: 6 Jul 06
Posts: 4
Credit: 11,699
RAC: 0
Message 47312 - Posted: 2 Oct 2007, 9:44:03 UTC
Last modified: 2 Oct 2007, 9:44:55 UTC

That´s too bad!

They should think it over!

I´m doing the same amount of work with half the cpu power! (50% / core)

:-D
ID: 47312 · 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 47337 - Posted: 2 Oct 2007, 20:01:13 UTC - in response to Message 47312.  

That´s too bad!

They should think it over!

I´m doing the same amount of work with half the cpu power! (50% / core)

:-D


eh?

Get the 64bit client or 32bit client of BOINC and run it. It will max out all your cores either way.

P.S. They have thought it over, search the forum ;)
Team mauisun.org
ID: 47337 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1832
Credit: 119,676,482
RAC: 10,619
Message 47349 - Posted: 3 Oct 2007, 7:34:43 UTC

Bubben,

64-bit isn't necessarily quicker than 32-bit - in some cases it's slower, and wouldn't be of much benefit to the project at the moment. That's why 64-bit OS's aren't the mainstream yet. The use of SSE extensions would probably be much more beneficial, but this isn't a trivial process and would require reviewing much of the code.

HTH
Danny
ID: 47349 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Rabinovitch
Avatar

Send message
Joined: 28 Apr 07
Posts: 28
Credit: 5,439,728
RAC: 0
Message 47750 - Posted: 15 Oct 2007, 8:53:00 UTC - in response to Message 47349.  

Bubben,

The use of SSE extensions would probably be much more beneficial, but this isn't a trivial process and would require reviewing much of the code.

HTH
Danny


Does they (I mean scientists) need to perform faster calculations in R@h? If they does, then it's already time to include at least SSE optimization in Rosetta applications, IMHO. 'Cause AFAIK in present Rosetta's cores doesn't support it yet. It means my CPU can perform more calculations per WU processing time, but an application doesn't allows it? :-(
ID: 47750 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1832
Credit: 119,676,482
RAC: 10,619
Message 47756 - Posted: 15 Oct 2007, 12:07:50 UTC

SSE isn't trivial to implement SSE unfortunately. There were two guys (Who! from Intel and Mats Petersson from AMD) looking into it a while ago but I'm not sure if there's currently any movement on that. If there are any coders with the time and experience to optimise ~1m lines of protein-modelling code then the project team would probably accept the help, but there are other considerations too...
ID: 47756 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Vistax64 WU?



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