Keith - I appreciate what you say, however I was merely giving JC a solution to his request to speed up his reporting.

Theres no need to apologise Keith, I always appreciate reading the good advice you give - I'm sorry I sounded a bit snappy in my reply - it was a long day at work

I must admit I didn't consider the possible impact on the server - I was under the impression that even though its set to communicate every 7 mins, it would only do so when required, ie. when a WU is nearing completion/ready to report etc.

In 'Messages' it doesn't show these 7 minute communications happening between these times - are you saying they do actually happen, but its not something we notice?

Ady

Thanks stares, although I was aware of that, I thought by reporting quickly it might help the others in the quorum with faster machines to complete it quicker.

Not that it really matters, we get them in the end - and its the results that matter

Do you have an answer to the other aspect I mentioned in my previous post?

Ady

Is there such a thing as a short day at work these days?

[/Diety] help me if I ever have to learn another language, English is tough enough for me!

I have to wonder that even if there's no actual communication with the server that the code still has to take some minute amount of time to check the various conditions and determine that there's no justification for communicating with the server yet. Clearly, probably a matter of milliseconds but add it up when you do it often enough and it may matter. I'm finding that BOINC seems to be at least designed well with respect to being something you can set, start and forget about. Things seem to be pretty well balanced out. If you want to tweak something, you can and there will be reasons to do it and reasons not to. The key is figuring out the importance/relevance of those reasons. EG, when you're going on vacation for a week and won't have network access but want to keep crunching, it's pretty easy to pull down enough WU to keep a machine busy for that time. However, there are ways to do that to minimize the extent to which you could cause those WUs to be sent to additional folks unnecessarily. I'm still trying to find the method to the madness BOINC uses to decide the "To completion" times for your WUs queued up.

The time to completions doesn't really hold any value on the WCG as it's not possible to predict although you can make an educated guess as to what it's likely to be which is what I think the times are at the start of the process

On other DC projects like SETI each work unit is of equal length to crunch so as BOINC is an all rounder time to completion in this case really is time to completion.

What does make me smile is it's stubborness to acknowledge that it might have got it a bit wrong. I've got one now on 53% and it's taken 3 and a bit hours but it's still predicting another 7 hours to completion

But as long as that work unit queue has got one lined up for when the other finishes I'm a happy bunny

Dave