there are big and small 4.24 wus before and after the 8% drop
problem : this machine doesn't have a constant non-boinc load, so time variates
constant : small wus still grant 13.29 on this host, big ones dropped from 122.25 to 121.37
I'm currently running einstien and cpdn on my a64x2-2.55, cpdn is giving 225 credits/12.6hrs or ir 17.85c/h. I haven't gotten any new WUs in several days, but if the credit's being cut by another 8%, I'll be getting 110c/5.5hrs or 20.08c/h which is still a slight nod in einstien's favor.
Guess what ?
Active hosts on E@H are 21% more powerfull than on S@H.
S@H is just an example. You can try with the rest of the projects.
I repeat : credit is not fair on E@H
Bruce wants (or has to) make it fair regarding other projects, but hosts are not the same !
As a consequence, attach you P3's to E@H and your top boxes to other projects if you want to grant credit. Stupid !
The total floating point speed of a projekt is calculated from the sum of the Recent Average Credit (RAC) for all users!
(A doubled average credit/hour would result in a doubled total floating point speed of a projekt at BoincStats, but without any increase in REAL computing power.)
The (RAC-calculated) floating point speed of E@H decreased in last 3 days (due to the last credit drop)
The real computing power = work done per day (Percent per day) is steady.
E@H : 0.956 GigaFLOPS / Host
S@H : 0.756 GigaFLOPS / Host
A result of more powerfull Hosts, or a result of higher average credit/hour on E@H?
Guess what ?
Active hosts on E@H are 21% more powerfull than on S@H.
S@H is just an example. You can try with the rest of the projects.
I repeat : credit is not fair on E@H
Hmm, very good point. Regarding to this it would also be normal if the average Credits are 21% higher than on other projects....
@Siegfried Niklas
Also a good point..hmmmm
But i still think the average PC on Einstein is faster... Seti is very old and i think many people are crunching with every pc they have, Pentium I's or whatever.. of course there are also some on einstein but not soo many i suppose.. but no clue...
Hmm, very good point. Regarding to this it would also be normal if the average Credits are 21% higher than on other projects....
@Siegfried Niklas
Also a good point..hmmmm
But i still think the average PC on Einstein is faster... Seti is very old and i think many people are crunching with every pc they have, Pentium I's or whatever.. of course there are also some on einstein but not soo many i suppose.. but no clue...
I really don't think that there are more "extreme" hosts here at E@H. I'd say that here are more "optimized" boinc applications and that's the reason for higher total Gigaflops etc. over at S@H let's say 80& of all users there are using a stock boinc client and here at E@H this number is way lower.(I'd guess 40 % at a minimum are using optimized boinc clients)
So you can't really take global Flops into account for the STEALING of credits that was done by the project.
And while we're at it ther is NO way of comparing the the calculations done by S@H to E@H or any other project!!! It's way to different and because of that it IS impossible to get a calibration between all projects !
Hmm, very good point. Regarding to this it would also be normal if the average Credits are 21% higher than on other projects....
@Siegfried Niklas
Also a good point..hmmmm
But i still think the average PC on Einstein is faster... Seti is very old and i think many people are crunching with every pc they have, Pentium I's or whatever.. of course there are also some on einstein but not soo many i suppose.. but no clue...
I really don't think that there are more "extreme" hosts here at E@H. I'd say that here are more "optimized" boinc applications and that's the reason for higher total Gigaflops etc. over at S@H let's say 80& of all users there are using a stock boinc client and here at E@H this number is way lower.(I'd guess 40 % at a minimum are using optimized boinc clients)
So you can't really take global Flops into account for the STEALING of credits that was done by the project.
And while we're at it ther is NO way of comparing the the calculations done by S@H to E@H or any other project!!! It's way to different and because of that it IS impossible to get a calibration between all projects !
I'm not sure to agree with you regarding the optimized clients. I think everybody uses the 4.24 optimized E@H client (as its is automatically sent to all users, and it has been released more than 15 days ago, wixh is over the maximum cache allowed by projects). I agree that only a few S@H users run the optimized client (stock is, but still 15% behind Simon's).
And I TOTALLY agree with you concerning comparison between calculations among projects, and especially your last sentence "It's way to different and because of that it IS impossible to get a calibration between all projects !"
So what is that damned average credit/hour/host that Boinc enforces between projects ? Nonsense
I'm not sure to agree with you regarding the optimized clients. I think everybody uses the 4.24 optimized E@H client (as its is automatically sent to all users, and it has been released more than 15 days ago, wixh is over the maximum cache allowed by projects). I agree that only a few S@H users run the optimized client (stock is, but still 15% behind Simon's).
I'm talking about the client ---> BOINC
Application ---> einstein_S5R1_4.24_windows_intelx86.exe
Quote:
And I TOTALLY agree with you concerning comparison between calculations among projects, and especially your last sentence "It's way to different and because of that it IS impossible to get a calibration between all projects !"
So what is that damned average credit/hour/host that Boinc enforces between projects ? Nonsense
Yes cross project calibration is useless.
But we can argue as much as we like, the "projects" won't listen anyway to their users ;-(
I don't see any reason to complain about the Einstein@home RAC. For example my Athlon 64 3000+, running 24/7 gives me:
SIMAP:
resource share: 100
RAC: 26.38
Einstein:
resource share: 200
RAC: 97.88
SAP:
resource share: 400
RAC: 126.02
Your can't compare any projects to each other because the calculations done are so extremly different that it's allmost like comparing apples and oranges...
Here's some info from the Mac side. Info below is over the past week, so it's five days at the previous credit allocation, and 2 days at the newest lower rate. The machine is set 100/100 SETI/Einstein
MacBookPro 2.16 GHZ Core Duo SETI 139 cpusec/cred Einstein 156 cpusec/cred Einstein gives 89% of the credit SETI gives. SETI std dev = 22.9; Einstein std dev = 2. Interestingly, I have been averaging 156 cupsec/credit on the newest app, and now that the credits have been reduced, this machine will now be taking 178 cpusec/cred, which is 78% of the credit given by SETI for the same amount of computer time.
SETI has WU that vary greatly in run time and credits earned. The MBP lowest cpusec/cred was 115 and the highest was 198 (one of those that awarded 12.87 cred for 2547 cpusec of work)
My PowerBook 1.67GHZ G4 has a SETI average of 278 with a std dev of 63.8. Einstein has been changing a lot: When the large WU average 122 credits, my PB was taking 406 cpusec/credit, which is only 68% of SETI. Then a couple of days ago, I switched to the newest Einstein app, and my time decreased on Einstein down to 300 cpusec/credit (2 WU, both large); then, Einstein decreases the credit again (third time!!) for parity with other projects, and my times increased again to 332 cpusec/credit, which is 84% of SETI.
So, Bruce, how can you say that Einstein is now the same as other apps for awarding credit for the use of my computers? Way back when I earned my degree in Math, 78% and 84% are not even close to 100%.
RE: RE: RE: yes... the
)
Have a look at my host results : http://einsteinathome.org/host/393047/tasks&offset=20
there are big and small 4.24 wus before and after the 8% drop
problem : this machine doesn't have a constant non-boinc load, so time variates
constant : small wus still grant 13.29 on this host, big ones dropped from 122.25 to 121.37
I'm currently running
)
I'm currently running einstien and cpdn on my a64x2-2.55, cpdn is giving 225 credits/12.6hrs or ir 17.85c/h. I haven't gotten any new WUs in several days, but if the credit's being cut by another 8%, I'll be getting 110c/5.5hrs or 20.08c/h which is still a slight nod in einstien's favor.
RE: From
)
The total floating point speed of a projekt is calculated from the sum of the Recent Average Credit (RAC) for all users!
(A doubled average credit/hour would result in a doubled total floating point speed of a projekt at BoincStats, but without any increase in REAL computing power.)
The (RAC-calculated) floating point speed of E@H decreased in last 3 days (due to the last credit drop)
The real computing power = work done per day (Percent per day) is steady.
E@H : 0.956 GigaFLOPS / Host
S@H : 0.756 GigaFLOPS / Host
A result of more powerfull Hosts, or a result of higher average credit/hour on E@H?
Sigi
RE: From
)
Hmm, very good point. Regarding to this it would also be normal if the average Credits are 21% higher than on other projects....
@Siegfried Niklas
Also a good point..hmmmm
But i still think the average PC on Einstein is faster... Seti is very old and i think many people are crunching with every pc they have, Pentium I's or whatever.. of course there are also some on einstein but not soo many i suppose.. but no clue...
RE: Hmm, very good point.
)
I really don't think that there are more "extreme" hosts here at E@H. I'd say that here are more "optimized" boinc applications and that's the reason for higher total Gigaflops etc. over at S@H let's say 80& of all users there are using a stock boinc client and here at E@H this number is way lower.(I'd guess 40 % at a minimum are using optimized boinc clients)
So you can't really take global Flops into account for the STEALING of credits that was done by the project.
And while we're at it ther is NO way of comparing the the calculations done by S@H to E@H or any other project!!! It's way to different and because of that it IS impossible to get a calibration between all projects !
RE: RE: Hmm, very good
)
I'm not sure to agree with you regarding the optimized clients. I think everybody uses the 4.24 optimized E@H client (as its is automatically sent to all users, and it has been released more than 15 days ago, wixh is over the maximum cache allowed by projects). I agree that only a few S@H users run the optimized client (stock is, but still 15% behind Simon's).
And I TOTALLY agree with you concerning comparison between calculations among projects, and especially your last sentence "It's way to different and because of that it IS impossible to get a calibration between all projects !"
So what is that damned average credit/hour/host that Boinc enforces between projects ? Nonsense
I don't see any reason to
)
I don't see any reason to complain about the Einstein@home RAC. For example my Athlon 64 3000+, running 24/7 gives me:
SIMAP:
resource share: 100
RAC: 26.38
Einstein:
resource share: 200
RAC: 97.88
SAP:
resource share: 400
RAC: 126.02
RE: I'm not sure to agree
)
I'm talking about the client ---> BOINC
Application ---> einstein_S5R1_4.24_windows_intelx86.exe
Yes cross project calibration is useless.
But we can argue as much as we like, the "projects" won't listen anyway to their users ;-(
RE: I don't see any reason
)
Your can't compare any projects to each other because the calculations done are so extremly different that it's allmost like comparing apples and oranges...
It's impossible.
Here's some info from the Mac
)
Here's some info from the Mac side. Info below is over the past week, so it's five days at the previous credit allocation, and 2 days at the newest lower rate. The machine is set 100/100 SETI/Einstein
MacBookPro 2.16 GHZ Core Duo SETI 139 cpusec/cred Einstein 156 cpusec/cred Einstein gives 89% of the credit SETI gives. SETI std dev = 22.9; Einstein std dev = 2. Interestingly, I have been averaging 156 cupsec/credit on the newest app, and now that the credits have been reduced, this machine will now be taking 178 cpusec/cred, which is 78% of the credit given by SETI for the same amount of computer time.
SETI has WU that vary greatly in run time and credits earned. The MBP lowest cpusec/cred was 115 and the highest was 198 (one of those that awarded 12.87 cred for 2547 cpusec of work)
My PowerBook 1.67GHZ G4 has a SETI average of 278 with a std dev of 63.8. Einstein has been changing a lot: When the large WU average 122 credits, my PB was taking 406 cpusec/credit, which is only 68% of SETI. Then a couple of days ago, I switched to the newest Einstein app, and my time decreased on Einstein down to 300 cpusec/credit (2 WU, both large); then, Einstein decreases the credit again (third time!!) for parity with other projects, and my times increased again to 332 cpusec/credit, which is 84% of SETI.
So, Bruce, how can you say that Einstein is now the same as other apps for awarding credit for the use of my computers? Way back when I earned my degree in Math, 78% and 84% are not even close to 100%.
C
Join Team MacNN