If I recall correctly, you have refused to set the "Connect Every X" and
"Extra Work" settings to what we have advised to allow your computers to
work. My memory is that "Connect Every X" is set to 0, and "Extra Work" is
set to something like 10. If you would change "Connect Every X" to a
value between 1 and 2, I expect that your woes about not enough work to
make it through the day would be reduced.
jm7
Mikus Grinbergs
<[email protected]>
Sent by: To
<boinc_dev-bounce "Lynn W. Taylor" <[email protected]>
[email protected] cc
u> [email protected]
Subject
Re: [boinc_dev] BOINC 6.10 release
10/27/2009 01:59
AM
> More to the point, by simply declaring that he's staying on 6.4.5, Mikus
> is no longer helping develop new versions of BOINC, he's hindering it.
It is absolutely correct that I've given up on developing new
versions of BOINC. My reason is that BOINC development is no longer
helping me in my participation in distributed computing projects.
My "big" system is running the 5.10.45 client. Why? Because none
of the newer clients fetch as much work as I need. I typically
connect once a day, and "squirt" work up/down. I then disconnect
until the next day. What happens when I run with a newer client is
that I manually wipe out all history, and enough work gets
downloaded. But the next day, *less* work gets downloaded than was
crunched. And the same thing the day after. Very soon, the new
work being downloaded does not add enough to the ready queue to keep
the system crunching for 24 hours. I must perform an "unscheduled"
download to keep the system from going idle, and have to manually
wipe history to restore download volume. [With 5.10.45, what gets
downloaded each time is about the same as what got crunched since
the last download.] [My guess is that the newer clients are
refusing to download work from "overworked" projects that have work
available, so as to supply crunching capacity to the "underworked"
projects -- which unfortunately happen *not* to have work to do !!]
It seems to me that BOINC development in the "work fetch" area is
solving server throughput problems for large projects by adding
constraints to the clients. I have enough trouble with occasional
retrying of transfers -- when they miss a "squirt", they miss a day
-- if a result upload misses several "squirts", the workunit may end
up missing its deadline. Now BOINC development has come up with
"project deferrals" -- I'm fairly sure that's incompatible with how
I participate. If BOINC development were willing to provide a way
to "opt out" from some of the new "work fetch" constraints, I'd be
willing to continue to help. As it is, things which used to work
with old versions of BOINC no longer work for me with new versions.
mikus
p.s. I laugh at one particular project. They limit the amount of
work given out at one time, and they set a short deadline. The
result: I connect for a "squirt"; a few WUs get downloaded, and
immediately go into EDF; no more WUs get downloaded until the next
"squirt" - which is tomorrow. If they allowed a larger amount to be
downloaded, my system has the capability to crunch them all in less
than 24 hours -- but the "supply" policy that project has chosen
ends up limiting the very benefit (have-others-do-the-crunching)
that distributed computing provides.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
