4 computers for one WU - is waste of resources
13 Mar 2005 21:11:15 UTC
Topic 188510
(moderation:
isn't it?
why do not assign 3 computers for the work?
and only if the results are different from anyone of these 3 - assign same work for additional workers?
is this aproach - assign work for 4 computers - have some statistical/mathematical background?
do you have statistics from existed results - how different they are?
4 computers for one WU - is waste of resources
)
I agree four is a bit much. But the whole idea of having several like results to validate a work unit goes back to the rampant cheating by some scumbag users at the original pre-Boinc Seti@Home. I like the number 3.
Freedom of Speech is a cherished right in every democracy and democratic institution.
I like 4; it's the square of
)
I like 4; it's the square of 2 and the square root of 16. Ahh.. happiness. :)
---
"Chance is irrelevant. We will succeed."
- Seven of Nine
I'm just guessing here, but
)
I'm just guessing here, but maybe 4 was chosen for a good reason. I mean, it's not like they haven't had the opportunity to see that 3 is a possible choice on boinc, as seti is using the number 3 ;)
I'm not sure what are you
)
I'm not sure what are you talkin about? Is it that one unit is processed by 4 users? If so, I also agree that 4 units are too much.
It really depends onhow fast
)
It really depends onhow fast yu need a verified result. Three is enough to weed out bad results, but the 4th can be looked at as insurance. If the results have to be verified quickly then the 4th is useful. If there is no real necessity for fast verification, then 3 is enough.
BOINC WIKI
Well, there is hopefully a
)
Well, there is hopefully a better explanation for this than fast verification. Because if 3 computers suffice there is one quarter of the computation power wasted. (Read it this way: The Einstein project could be one third faster!)
HI I still have 2 unverified
)
HI
I still have 2 unverified results which were returned on the 5th of March. (This was written on the 16th of march) The other client who reported did so on the 3rd. Two of the four clients failed to report. So even with 4 there can be a significant time lapse before a result is confirmed. The organisers seem to know what they are doing. We are not privy to the detailed research that was carried out before this system was set in place.
I consider this a redundant thread.
you can lead a horse to water but a pencil must be led
Since the goal is to only
)
Since the goal is to only declare a result valid if 3 computers have checked it, the purpose of having a WU sent to 4 computers is to prevent more time being wasted if one computer doesn't return a result. Remember that the extra computer is the slowest of the bunch or one which has had an error so the wasted computer effort is one which is contributing the least.
------------------------------------
There's a thin line between Genius and Insanity. That's where I live, baby!
THE SPACEPORT - The Other Side of Space