Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 2
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 747 times and has 1 reply Next Thread
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
HCMD2 - each WU is different...

Hello everybody,

sorry to start a thread for this but I checked a bit and couldn't find the answer.

Why the WU's are so different in CPU time the one from the other? Now I have one with 7h10m and 80% and I reported one of 8 minutes total CPU time!

This is driving BOINC crazy when you buffer it: I am always not sure if I buffered enough or too much.

Thanks.
L.
[Oct 28, 2010 9:54:17 PM]   Link   Report threatening or abusive post: please login first  Go to top 
gb009761
Master Cruncher
Scotland
Joined: Apr 6, 2005
Post Count: 2977
Status: Recently Active
Project Badges:
Reply to this Post  Reply with Quote 
Re: HCMD2 - each WU is different...

Hi latakia, I know your frustrations - as I (and I'm sure many others) also wish that there could be some sort of 'standardisation' with HCMD2 WU's. Unfortunately though, due to the unpredicatability of the science, this isn't possible. Basically, until a WU is actually crunched, it's impossible to tell exactly as to how long it'll run for - be it 8 mins or 12 hours.

There are some checks in place that either cut the WU off at 6 hours if it hasn't reached a certain percentage (I think, 60%) and at 12 hours tops - no matter how far it's got. Then, what ever's left, is cut down into a smaller WU and processed again to these limits (at which point, the process is repeated time and time again, until it's finished). Thus, the total randomness of it all...

Another factor which the WCG techs have attempted to eliminate as much as possible, is the pairing of WU's to similar computers (which isn't perfect, but it's better than what it was). So, computer A may process the whole WU in under 6 hrs, but computer B may take longer (and hence, be stopped at 6 hours) - therefore, only the part of the WU which both computers processed can be validated. The rest, is chopped up into another WU and sent back out to be processed (hopefully) to completion in that 1 go.

What I do though to compensate this total randomness, is have my cache setting at 1 day, which in practice, pretty much covers any long running WU's from being returned later than 2 days after they were sent to my computer.
----------------------------------------

[Oct 28, 2010 10:39:37 PM]   Link   Report threatening or abusive post: please login first  Go to top 
[ Jump to Last Post ]
Post new Thread