Why E@M tasks are so "huge"?
Anonymous
28 Apr 2018 8:51:57 UTC
Topic 214712
(moderation:
I was wondering why these tasks have a huge estimated remaining time, with strict deadline (typically I see 15days).
I see that other projects (World Grid, SETI@Home) tipically have tasks that run in about 1-2 hour of computing time.
My problem is that I run BOINC on my work pc, which has great computational capabilites. However since I travel often I know for sure that some tasks can never be ended by the deadline. Therefore I abort them, without providing any help to this project.
My question is: is it possible to create smaller tasks for this project?
Estimated runtime is a
)
Estimated runtime is a function of BOINC, which I generally ignore, actual runtime depends upon E@H & your host. The deadline is a way of propelling E@H ( secondarily a courtesy to other users in the same quorum ). They converge with regular use, while aborting tasks averts this. Either, just let it run for a while without intervention or perhaps E@H is not for you. :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Stefano Collovati wrote:I was
)
I'm wondering ... what is your definition of huge? The deadline is 14 days so I guess a multi-day crunch time would be 'huge' :-). Your computers are 'hidden' so it's not possible for others to see what searches you are contributing to or what your crunch times really are.
The problems being investigated by different projects are computationally quite different. The Einstein staff have already limited the task 'size' as much as is physically possible to do. The two searches that take the most CPU time are the gravitational wave (GW) search and the Gamma-ray pulsar (FGRP5) search. A relatively modern desktop machine with 'great computational capabilities' should be able to process a GW task in say around 9-12 hours and a FGRP5 task in maybe 6 hours or less. I don't think either of these could really be categorised as 'huge' :-). Of course, machines with lesser capabilities will take correspondingly longer but still well within the deadline. It really can't be helped that the 'signal' being searched for is extremely computationally difficult to tease out of the background noise.
Without having more information about all sorts of variables here, it's impossible to make suggestions, other than to simply say, "Limit your task downloads to what can be crunched in the time your machine will be in 'crunching mode'.
In short, the answer is unfortunately, 'No'. For a longer answer, just add - 'so that the project can actually work with a degree of efficiency and a hope that something can be found (or not found as the case may be) in a realistic time frame' :-). It's not like the project deliberately sets out to antagonise its volunteers :-).
Cheers,
Gary.
Stefano Collovati
)
There is a chance to run smaller tasks if your CPU has got an iGPU. You could then, in the E@H web account, choose YES in the Beta settings plus select Arecibo in the application settings. However, I don't know what the official E@H policy is as to the Beta settings.
Did you know that you can run E@H on mobile devices, too?
Thank you all for the gentle
)
Thank you all for the gentle answers.
I was simply trying to understand what is the best way for me to contribute to this (and other) projects.
My "problem" is that the only computer I have is a Macbook Pro, and sometimes I know that I have to keep it powered off.
However thanks to your suggestions I found a good compromise: two CPUs dedicated for E@H and two for other projects (like Seti@Home) that usually have smaller tasks. Tipically I use 50% of the CPU time, in this way I can use the computer for normal tasks without lag problems.
I'm glad you found a workable
)
I'm glad you found a workable compromise.
Thanks for letting us know!
Cheers,
Gary.
In the last 1 to 2 months the
)
In the last 1 to 2 months the average completion time for my cpu units has nearly doubled, and gpu units have gone from 12 t0 18 minutes to an average of around 30 with occasional outliers as long as 90 minutes. My cpu is running as fast as ever (3.68 GHZ turbo), and I have 64 GHZ of memory. If this is simply a case of bigger or more complex units, I'm fine with that, but if it's on my end I'll need to find a solution. Comments appreciated.