Guys, you *don't need* more than 2 day cache for Einstein@Home (if you have 24/7 internets connection on your side).
Riding out the rare connection glitch is my primary reason for a big cache. My default of 4 days covers a long weekend + repair the next working day. Prior to going on vacation I normally bump it up enough to last a day past when I'm back to hedge against a major project outage.
My default of 4 days covers a long weekend + repair the next working day.
That makes a fair amount of sense--once things have stabilized enough so that four days really means something like four days.
I think there is, however, a good argument for using really small numbers--maybe 0.2 days, when one has made a major change which could either trash the entire cache contents (such as a new non-standard application, or use of app_info), or cause the completion time estimates to be wildly wrong (such as joining a new project).
My personal practice at such boundaries is to start at something like a quarter of a day, then once things seem at least a little well behaved, to bump it up in increments of a half day or so.
Sadly, my personal track record at new app_info is something like a 50% probability of trashing the full current cache. Even knowing that and trying to be careful, I've not gotten much better. It is so nice that on Einstein I currently don't feel a need to use an app_info.
Juan: VERY nice lot of computers you have there. Wow.
Thanks. My "fleet" is allways ready to do it´s job, but they are actualy split in 3 diferent places, it´s safer an to hot to handle in one with no AC.
Well, it seems that Seti is having another somewhat extended outage following a power failure that took everything down this morning.
Sooooooooooo......
The kitties will be sending most of their rigs over here again to keep the kibble bowls full of fine crunchables.
Meow meow meow!
Meow indeed, or maybe fffffft! Even my slow old machines have finally come over to Einstein again for some work to do. Well, at least it gives my Einstein and overall Boinc stats a good shot in the arm, and I'm not losing any ground at Seti.
I find it highly interesting that Seti has the up/download servers on even while all other external communication is cut off.
David
David
Miserable old git
Patiently waiting for the asteroid with my name on it.
big box now moved to E@H from
)
big box now moved to E@H from SETI. another one of the GPUUG crew!
Andy
RE: big box now moved to
)
Nice rig......
That should be able to do well here!
RE: RE: big box now moved
)
Welcome aboard the E@H GPUUG cause.
Take a look in the GPUUG SETI-forum for some information about whow to configure your ring for both SETI/E@H usage.
Team, now we have the firepower to cross the 2MM dialy mark!
RE: Argh,
)
Riding out the rare connection glitch is my primary reason for a big cache. My default of 4 days covers a long weekend + repair the next working day. Prior to going on vacation I normally bump it up enough to last a day past when I'm back to hedge against a major project outage.
RE: My default of 4 days
)
That makes a fair amount of sense--once things have stabilized enough so that four days really means something like four days.
I think there is, however, a good argument for using really small numbers--maybe 0.2 days, when one has made a major change which could either trash the entire cache contents (such as a new non-standard application, or use of app_info), or cause the completion time estimates to be wildly wrong (such as joining a new project).
My personal practice at such boundaries is to start at something like a quarter of a day, then once things seem at least a little well behaved, to bump it up in increments of a half day or so.
Sadly, my personal track record at new app_info is something like a 50% probability of trashing the full current cache. Even knowing that and trying to be careful, I've not gotten much better. It is so nice that on Einstein I currently don't feel a need to use an app_info.
Juan: VERY nice lot of
)
Juan: VERY nice lot of computers you have there. Wow.
RE: Juan: VERY nice lot of
)
Thanks. My "fleet" is allways ready to do it´s job, but they are actualy split in 3 diferent places, it´s safer an to hot to handle in one with no AC.
I think Einstein rather likes
)
I think Einstein rather likes Hungry hosts :¬)
In the News forum,
Einstein@Home has passed 800 Teraflops!
http://einsteinathome.org/node/196647
Well, it seems that Seti is
)
Well, it seems that Seti is having another somewhat extended outage following a power failure that took everything down this morning.
Sooooooooooo......
The kitties will be sending most of their rigs over here again to keep the kibble bowls full of fine crunchables.
Meow meow meow!
RE: Well, it seems that
)
Meow indeed, or maybe fffffft! Even my slow old machines have finally come over to Einstein again for some work to do. Well, at least it gives my Einstein and overall Boinc stats a good shot in the arm, and I'm not losing any ground at Seti.
I find it highly interesting that Seti has the up/download servers on even while all other external communication is cut off.
David
David
Miserable old git
Patiently waiting for the asteroid with my name on it.