I have a few machines working on 5 (for the time being) projects. I'm just wondering how you guys who have a large number of machines tend to spread the load.
I realise that it's best to keep any one machine doing more than a single project incase there is an outage, but say you have 10 machines and 15 projects in total, do all your machines run all projects at the same percentages, or do you tend to have project specific machines, each with a backup project running at 10%...just incase.
Hope I've asked my question clearly enough, not exactly critical but I'm curious to see how the hardcore crunchers do it :)
Andrew
Spreading the workload
Forum rules
Welcome to The Scottish Boinc Team boards. See forum rules in pinned post. If you can't be bothered then try not to be too naughty as I have a delete button to press and a ban hammer to swing.
Welcome to The Scottish Boinc Team boards. See forum rules in pinned post. If you can't be bothered then try not to be too naughty as I have a delete button to press and a ban hammer to swing.
-
Temujin
#2 Re: Spreading the workload
Hi Andrew
I don't know about anyone else but I tend to run all my 40+ machines on just a single project, seti.
They all run with a 5 day cache which gets them over all but the longest outages.
On the rare occasions they do run out of seti work I normally switch them over to a short period project like milkyway.
I don't know about anyone else but I tend to run all my 40+ machines on just a single project, seti.
They all run with a 5 day cache which gets them over all but the longest outages.
On the rare occasions they do run out of seti work I normally switch them over to a short period project like milkyway.
#3
Your probably best seeing what suits each machine. I run everything on my main rig and whatever (short duration) project takes my fancy on the laptops - when they're on!
The machine at work can't upload/download so it gets things with distant deadlines. This used to mean projects like SETI and SZTAKI which had 4-6 week deadlines but for the last few months (year? more?) has been CPDN as each workunit takes months and they have no 'deadline' as such.
I would personally recomend 4-5 projects as a minimum to cover down time; but this can be reduced if you are attached to 'stable' projects - if you can find any!
I'm sure others will let you know their preferences...
The machine at work can't upload/download so it gets things with distant deadlines. This used to mean projects like SETI and SZTAKI which had 4-6 week deadlines but for the last few months (year? more?) has been CPDN as each workunit takes months and they have no 'deadline' as such.
I would personally recomend 4-5 projects as a minimum to cover down time; but this can be reduced if you are attached to 'stable' projects - if you can find any!
I'm sure others will let you know their preferences...
-
Buster Gunn
- Boinc Second Lieutenant
- Posts: 513
- Joined: Sun Aug 13, 2006 7:20 am
- Location: Wilmington Delaware USA
#4
I tend to run only 1 project per machine but I do load up the queues with 10 days worth of work. I have a few projects that I use for fill-in, in case of a hiccup. Prime Grid and Riesel Sieve always have work, their run times are short and easy to start up. I do have 2 machines that run as many as 10 projects at a time. I use these to work on my anal goal of getting ALL my projects over the 10K mark. It also helps me keep my CPID's in line.
Buster is my dog. Sadly, Buster is gone.
Buster is the Malamute, Obi is the Golden.
Obi is the dog of the house now.
Buster is the Malamute, Obi is the Golden.
Obi is the dog of the house now.
#6
Don't know about the others, but I definitely don't think I count as an expert. Slightly experienced user maybe...Thanks for the replies guys, just thought it would be interesting to see how the experts did things :)