> Now - I'm wondering if my participation in Einstein@Home is worthwhile.
A reasonable question.
The formal limit is that you turn round WU in 7 days, but to be on the safe side I'd suggest limiting it at around 5days. For comparison both my 700MHz machines crunch just under 7 WU/week running E@H 24/7 (Win-ME and Win-XP).
As a very rough rule of thumb, anything down to around a 150MHz
machine might worth trying if running 24/7 and running no other BOINC project.
If running only during a 42-hr working week, make that a 600MHz machine, and so on.
Proportionally faster again if the machine runs other BOINC projects as wells as E@H. I see you have Macs - perhaps you have a very rough feel for how your Mac compares to a Win PC of given speed?
If the machine seems like it might meet that rule of thumb I'd peronally say give E@H a go.
Watch the first couple of work units. If they take more than 5ish so days to crunch then there is a risk that some of your WU will turn in late, which just wastes your machine's time and delays other people's results. In that case, and in the nicest possible way, your machine's contribution would be better appreciated elsewhere.
But please, while you do those first couple of WU, keep the 'connect every' interval set to 0.1 so your box does not download too many WU till you know what it can crunch. If you do increase this value later, step it up gradually in steps of 0.2 or 0.3 at a time with a day or so between changes. There are unexpected side effects that mean if you increase it too fast you will get more data than you can crunch, and as the previous postings show this gets people *really* upset!
Equally nobody should be rude to someone who tries a couple of WU and then ducks out again when they take too long. We should be willing to absorb a couple of mistakes from any given volunteer without being rude.
If you find can regularly turn round E@H WU in <~5 days there is no reason not to contribute - every little helps, you meet the criteria set by the project team, and you should not be put off by those who can turn round dozens of WU each week.
Gravywavy - thanks for your long and informative reply!
Watch the first couple of work units. If they take more than 5ish so days to crunch then there is a risk that some of your WU will turn in late, which just wastes your machine's time and delays other people's results. In that case, and in the nicest possible way, your machine's contribution would be better appreciated elsewhere.
Will do.
I've also noted the deadlines for the WUs.
It all seems reasonable to me. The idea is to get results, not impede the process. I'll gladly step aside if necessary.
> Now - I'm wondering if my
)
> Now - I'm wondering if my participation in Einstein@Home is worthwhile.
A reasonable question.
The formal limit is that you turn round WU in 7 days, but to be on the safe side I'd suggest limiting it at around 5days. For comparison both my 700MHz machines crunch just under 7 WU/week running E@H 24/7 (Win-ME and Win-XP).
As a very rough rule of thumb, anything down to around a 150MHz
machine might worth trying if running 24/7 and running no other BOINC project.
If running only during a 42-hr working week, make that a 600MHz machine, and so on.
Proportionally faster again if the machine runs other BOINC projects as wells as E@H. I see you have Macs - perhaps you have a very rough feel for how your Mac compares to a Win PC of given speed?
If the machine seems like it might meet that rule of thumb I'd peronally say give E@H a go.
Watch the first couple of work units. If they take more than 5ish so days to crunch then there is a risk that some of your WU will turn in late, which just wastes your machine's time and delays other people's results. In that case, and in the nicest possible way, your machine's contribution would be better appreciated elsewhere.
But please, while you do those first couple of WU, keep the 'connect every' interval set to 0.1 so your box does not download too many WU till you know what it can crunch. If you do increase this value later, step it up gradually in steps of 0.2 or 0.3 at a time with a day or so between changes. There are unexpected side effects that mean if you increase it too fast you will get more data than you can crunch, and as the previous postings show this gets people *really* upset!
Equally nobody should be rude to someone who tries a couple of WU and then ducks out again when they take too long. We should be willing to absorb a couple of mistakes from any given volunteer without being rude.
If you find can regularly turn round E@H WU in <~5 days there is no reason not to contribute - every little helps, you meet the criteria set by the project team, and you should not be put off by those who can turn round dozens of WU each week.
~~gravywavy
Gravywavy - thanks for your
)
Gravywavy - thanks for your long and informative reply!
Watch the first couple of work units. If they take more than 5ish so days to crunch then there is a risk that some of your WU will turn in late, which just wastes your machine's time and delays other people's results. In that case, and in the nicest possible way, your machine's contribution would be better appreciated elsewhere.
Will do.
I've also noted the deadlines for the WUs.
It all seems reasonable to me. The idea is to get results, not impede the process. I'll gladly step aside if necessary.
Cheers.