• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Faster folding

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

res0r9lm

Member
Joined
Dec 14, 2001
Location
florida
ok I have noticed that if you run foldmonitor seprately on each comuter instead of using one computer setup with all your clients it is faster at folding. at least it is for me can't explain why maybe it has something to do with winME and home networking. Can anybody verify this or is it just my computer?
 

David

Forums Super Moderator
Joined
Feb 20, 2001
res0r9lm said:
ok I have noticed that if you run foldmonitor seprately on each comuter instead of using one computer setup with all your clients it is faster at folding. at least it is for me can't explain why maybe it has something to do with winME and home networking. Can anybody verify this or is it just my computer?

Perhaps the time on each computer is different? Even being 5 mins out might affect the system.
 

Cluster

Member
Joined
Dec 5, 2001
Location
Canuckistan
I'm just curious, why leave fold monitor running? Sure it tells you how fast your folding, but once you've seen it a couple of times, why not just get leave it be.
 

David

Forums Super Moderator
Joined
Feb 20, 2001
I usually have it in the systray all the time - never had a problem
 

doer

Member
Joined
Dec 9, 2001
Location
EU, Finland
well... I ran 3 clients on my TB 1.4 and all the clients got a bit different frame times on the same protein. I monitored them from my main rig. If client writes log every 6 mins and the Foldmonitor reads it every 6 mins it might affect times... But IMHO foldmonitor decreases performance cause it takes some (little amount) of cpu time. If your NICs need cpu time then its maybe them making folding slower cause sending results to the foldmonitor comp.
 

JetMech

Member
doer said:
well... I ran 3 clients on my TB 1.4 and all the clients got a bit different frame times on the same protein. I monitored them from my main rig. If client writes log every 6 mins and the Foldmonitor reads it every 6 mins it might affect times... But IMHO foldmonitor decreases performance cause it takes some (little amount) of cpu time. If your NICs need cpu time then its maybe them making folding slower cause sending results to the foldmonitor comp.
It's possible the different times are due unequal distribution of cpu cycles. Also each client sends and recieves work at a different rate dependent on the servers responses. Over time this can become a sizeable gap in average time. While using two clients I had as much as 20 frames difference sometimes and then at times they were equal. Just the nature of the beast. Some servers are/were notorious for making you wait for work. :beer: :burn:
 
OP
R

res0r9lm

Member
Joined
Dec 14, 2001
Location
florida
I think it might have something to do with not all protiens fold at the same speed even if it's the same protien. I'm getting a average of 92 sec per a frame and best of 80 sec per a frame doing native bba5's these times are for based on whole wu and computer not in use. that's a 10% difference:eek:
 

RickShelton

Registered
Joined
Mar 7, 2002
Location
Wilkesboro, NC
res0r9lm,

I will agree to what you have found, as I looked at the logs from my two athlon systems and noticed the same thing. Times vary as much 10-15 secs/frame on the smaller proteins, even though the systems are only folding and not doing anything else at all.

Cheers,
Rick
 

nil_esh

Member
Joined
Nov 7, 2001
Location
Tampa, FL
Cluster said:
I'm just curious, why leave fold monitor running? Sure it tells you how fast your folding, but once you've seen it a couple of times, why not just get leave it be.

I run it to make sure my PCs are folding and everything is running right. Sometimes someone will turn off a machine and without FoldMonitor I won't know until I see a drop in production or walk into the room and see that the computer is off. Its also nice to see what proteins are being worked on.

I can't really verify that thing about FoldMonitor slowing things down over the network, but it makes sense-- a network access will take a good few CPU cycles on both sides of the connection, most likely more than a hard drive access, but is that really enough to make a significant increase in your times? Try increasing the time between FoldMonitor updates.