PDA

View Full Version : Candidated Getting Larger?

Acrazar
10-28-2005, 07:58 PM
Anyone else noticing that the size of the canditates is getting alot bigger?

They seem to be getting bigger and way more complex.
[XC] DragonOrta
10-28-2005, 08:00 PM
I was thinking that same thing. I left my 2.8 P4 laptop crunching for three hours, and it only had 10 cands done, which seems really long.
Disposibleteen
10-28-2005, 08:02 PM
Not me, my rig has been getting a lot of 7 and 8 rotatable bond units which it flys through.
Vapor
10-28-2005, 08:02 PM
My X2 is still cruising at 310 a day.....except when MATLAB leaks and uses the CPU fully randomly. I've noticed a decline in Mafia output though....:shrug:
trakslacker
10-28-2005, 08:09 PM
I'm still crunching full on, no slowdown here.
mad mikee
10-28-2005, 08:23 PM
43 atoms / 14 bonds - takes 2 mins+ / conformer
Magnj
10-28-2005, 08:59 PM
I'm buffering...i sent you a PM...I'll having a 1000 cand present next week
shmaa
10-28-2005, 09:09 PM
When I checked Dspy just for the heck of it at random last week I saw one that had 60 atoms and 22 rotatable bonds.. it was taking 4:30 per conformer on a 2.7ghz A64
bullet2urbrain
10-28-2005, 09:15 PM
i've run into several 3:00 min conformers on my x2.. i wish there was a way to evenly distirbute the core's to equalize.. like if one core is doing a 5 bond structure and the other has a 21 bond more CPU would be allocated..

ROD!!!!! get started on this

Magnj: good to know.. let me know via PM which day that will be please.
Acrazar
10-28-2005, 10:02 PM
yeah im getting stuctures like these all the time now...
Vapor
10-28-2005, 10:26 PM
Magnj....no worries, I know.

I can't say I look at DSpy enough to know when I get a beast of a cand, all I know is that I see a lot of 4-13 sec conformers and very few >100sec conformers (never seen one over 120sec, TBH). I guess I'm lucky? :shrug:
ex2cib
10-28-2005, 10:53 PM
i had a candidate that took about 30 minutes to complete one ngiht on my 2.7 venice. i think my computer just wanted a restart. worked fine (and faster) afterwards
Entity_Razer
10-29-2005, 02:35 AM
I actually noticed the same thing. Can't say if it's just a bundle I downloaded with a lot of high atoms/time consuming work (all queue sizes at 2000 ) but I to noticed the output dropping on this end. My Xeons still do a good 130-140 & day but still.....

If you have a small buffer try flushing your task and see if you get smaller tasks? Might be that they where "giving out" large tasks or something....