| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: How to set cashing of tasks and what projects to concentrate on per processor? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 4
|
|
| Author |
|
|
wolfman1360
Senior Cruncher Canada Joined: Jan 17, 2016 Post Count: 176 Status: Offline Project Badges: 
|
Hi,
----------------------------------------So, I've been a lurker here for quite some time. Only recently (in the last few months, really) have I actually started really taking an interest in WCG. I like that there's a myriad of projects that I can concentrate on. However I'm not sure if I'm overthinking. I have several classes of processors crunching at the moment. I'm not sure what is visible to everyone - WCG doesn't seem to list actual CPU type - but I have several Intel atom (d2550, z3740/3735G, x5 z8500). core i5 430M, 2415M, 3317U, 4210Y, 2300. Core i7 3610QM, 3720QM and 6700U. As well as an older core 2 duo hammering away in the basement that just won't quit ;). Several AMD's an older neo athlon, APU in a samsung laptop with bad hinges and my fx8350 that is a daily driver. Oh, and of course my android phones ;). Anyway... Now, I realize that's probably overkill for something like this - that's probably pretty close to 60 threads/cores. My questions are as follows and again, I might be overthinking. What is the best way to cash work on these machines? For instance, I notice they're all crunching pretty quickly compared to their deadlines at 0.5 days of work. Should I increase this on any of the higher end processors or is this just unnecessary? Second, what is the best project to run on which processor? For instance is MCM going to net as many credits and do as much good as, for instance, fight aids, which, even on a lower end CPU, takes maybe an hour or two? Especially on the atom processors where mapping cancer markers seems to take a solid 24 hours plus. Again, please let me know if I'm overthinking this in any way.
Crunching for the betterment of human kind and the canines who will always be our best friends.
AWOU! |
||
|
|
mmonnin
Advanced Cruncher Joined: Jul 20, 2016 Post Count: 148 Status: Offline Project Badges: 
|
The cache is up to you. WCG is a pretty stable project and doesn't go down very often. It's not like SETI that goes down every Tuesday for half the day. So a 1-2 day cache I don't think is necessary for computers that are always connected. It also might depend on how close you monitor BOINC. I usually have 1/4 to 1/5 day cache on my systems.
----------------------------------------The OS plays a large roll in run times. Linux is typically faster especially for VINA based work. An example I went from run times of ~90-95min to 55-60min just by switching from Win10 to Linux on my AMD 1950x when running ZIKA. The OS might play just as large of a role as the app. MCM is a bit lower PPD than ZIKA. I've heard some certain FAAH tasks are good. Some others probably have better input than myself.  |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7847 Status: Offline Project Badges: 
|
You have a lot going on there, but don't blow your electric bill out of the water. Mmonnin is correct in saying any project which uses VINA (currently OET, ZIKAand SCC) run more efficiently on Linux than windows. I can not say for the other projects, it might be about the same. For the most part if you run with a cache of .5 to 1 day you should be fine. No need to overdo it as outages are rare, but they do happen occasionally. Do not fret about the TB project as those work units are rare. I run about 60 to 70 threads and I only see maybe 1 or 2 a week.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
wolfman1360
Senior Cruncher Canada Joined: Jan 17, 2016 Post Count: 176 Status: Offline Project Badges:
|
Utilities are included in rent so far, anyway.
----------------------------------------Next question - is there a tool for windows to manage all machines on the same network? Anything in particular I should be monitoring/paying attention to? Right now I have the mean CPUs running either FAAH or HSTB with priority and run other projects if no work is available from those two. Everything else is set to the defaults although the atom z8500 has come close to not completing FAAH a few times because of underclocking, so set it to use 60% CPU and consistently staying at 2.2 ghz. Excellent on the cashing. Have it set to 0.5 days. I wasn't sure if there was any advantage or anything in changing that - although if HSTB was the only thing I focused on it might be good to get a few days in.
Crunching for the betterment of human kind and the canines who will always be our best friends.
AWOU! |
||
|
|
|