I think EAH should send out four Albert work units, like SAH, that may reduce the credit pending problem.
Einstein used to send out 4, just like Seti. It was only recently they changed to sending out only 3 initial WU's. It slows down validation, but they feel that it increases their computational power.
how do / did they compute the score- i love it wheni crunch with 2 big clusters like physics and s* speaking for erlin- i report 30, they report 8 and 9 and i et 9... wish the clusters only crunched with other clusters scorewise cause their low score is inversly proportionate (sp?) to the computing power sorta; seems like more power lower score till you get to some of the p3's that'll take 10k secs to report 15 credits. dunno
Quote:
Quote:
I think EAH should send out four Albert work units, like SAH, that may reduce the credit pending problem.
Einstein used to send out 4, just like Seti. It was only recently they changed to sending out only 3 initial WU's. It slows down validation, but they feel that it increases their computational power.
how do / did they compute the score- i love it wheni crunch with 2 big clusters like physics and s* speaking for erlin- i report 30, they report 8 and 9 and i et 9... wish the clusters only crunched with other clusters scorewise cause their low score is inversly proportionate (sp?) to the computing power sorta; seems like more power lower score till you get to some of the p3's that'll take 10k secs to report 15 credits. dunno
There are problems with Linux machines having low benchmark scores. So, you have been linked up with them for some of the work and this hurts you in the great cobblestone race.
The later generation BOINC Client Software has a new mechanism and we have hopes that this will eliminate this problem once and for all. The first test will be with SETI@Home Enhanced. Bruce Allen said he would look into it for Einstein@Home. BUt, it will have to wait for another update to the application before we would see a change.
mac's too cause thats really a linux core too ive noticed. it's be great to get all the os's on the same plane but thats about impossible to do
Quote:
Quote:
There are problems with Linux machines having low benchmark scores. So, you have been linked up with them for some of the work and this hurts you in the great cobblestone race.
The later generation BOINC Client Software has a new mechanism and we have hopes that this will eliminate this problem once and for all. The first test will be with SETI@Home Enhanced. Bruce Allen said he would look into it for Einstein@Home. BUt, it will have to wait for another update to the application before we would see a change.
mac's too cause thats really a linux core too ive noticed. it's be great to get all the os's on the same plane but thats about impossible to do
Problem's kind of bad because there are so many more WUs in a datafile now, so the pairings in co-crunchers go on for more WUs. You could be in a much worse situation, do you realize that? That could be you, crunching on a linux box and hoping for a couple of Windows machines to come along and pull your low scores up to normal.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
There are problems with Linux machines having low benchmark scores. So, you have been linked up with them for some of the work and this hurts you in the great cobblestone race.
Easily cured by running the optimized Linux client found here:
RE: .... [snip]..... ...
)
Bloody Hell!!! I was half way through signing up before I'd finished reading your post ;-)
Wonder what I was actually signing up to?!
I think EAH should send out four Albert work units, like SAH, that may reduce the credit pending problem.
And I hope we're still giving Jim Fekete some of the answers he was after
Join the #1 Aussie Alliance on Einstein
RE: I think EAH should send
)
Einstein used to send out 4, just like Seti. It was only recently they changed to sending out only 3 initial WU's. It slows down validation, but they feel that it increases their computational power.
Jim
Jim
how do / did they compute the
)
how do / did they compute the score- i love it wheni crunch with 2 big clusters like physics and s* speaking for erlin- i report 30, they report 8 and 9 and i et 9... wish the clusters only crunched with other clusters scorewise cause their low score is inversly proportionate (sp?) to the computing power sorta; seems like more power lower score till you get to some of the p3's that'll take 10k secs to report 15 credits. dunno
RE: how do / did they
)
There are problems with Linux machines having low benchmark scores. So, you have been linked up with them for some of the work and this hurts you in the great cobblestone race.
The later generation BOINC Client Software has a new mechanism and we have hopes that this will eliminate this problem once and for all. The first test will be with SETI@Home Enhanced. Bruce Allen said he would look into it for Einstein@Home. BUt, it will have to wait for another update to the application before we would see a change.
mac's too cause thats really
)
mac's too cause thats really a linux core too ive noticed. it's be great to get all the os's on the same plane but thats about impossible to do
RE: mac's too cause thats
)
Problem's kind of bad because there are so many more WUs in a datafile now, so the pairings in co-crunchers go on for more WUs. You could be in a much worse situation, do you realize that? That could be you, crunching on a linux box and hoping for a couple of Windows machines to come along and pull your low scores up to normal.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: There are problems with
)
Easily cured by running the optimized Linux client found here:
http://einsteinathome.org/host/249812/tasks
My claimed is actually better than most unoptimized Win clients. Check it out:
http://einsteinathome.org/host/249812/tasks
how would one optimise a
)
how would one optimise a windows client?
RE: how would one optimise
)
reformat the hard drive and install linux! ROFL
sorry, just had to say it.
Jim
RE: how would one optimise
)
Try looking here.