S5 is optimized, and doing more work per hour than S4, so the increase in work is giving you better credit. A 4x increase is about what the optimization is doing. So you are right on track.
S5 is optimized, and doing more work per hour than S4, so the increase in work is giving you better credit. A 4x increase is about what the optimization is doing. So you are right on track.
Imho anything that comes straight from the projects is stock application, and thus deserves plain standard credits.
If some volunteers do some modifications for better performance than stock, it's optimized (and may deserve more credits).
If increasing the science output through better programming on the project side will increase the credits per hour for a puter for that project, I'd recommend for any new project to start with very sloppy programming to have an big increase in credits (and a big jump in user base afterwards) after some time the programming is "optimized".
S5 is optimized, and doing more work per hour than S4, so the increase in work is giving you better credit. A 4x increase is about what the optimization is doing.
seti_enhanced 5.x is also optimized compared to the old seti 4.x. But the new seti doesn't give you more credits per hour than the old seti (if you are comparing the stock applications).
Quote:
So you are right on track.
No. The basic idea of boinc is that the stock application of every project gives you approximately the same credits per hour on a given computer system. A new stock application must not break this idea.
To show you that something is wrong let's assume that we still have the old credit system based on time * benchmark method and we still need 3 crunchers to reach the quorum.
Let's say that the old application needs 4 hours to do a workunit on computer x. It is claiming 40 credits based on the time * benchmark method. Let's say that the other results of the quorum are claiming 30 credits and 50 credits. So, granted credits will be 40. The credit production rate of computer x is 10 credits/h.
Now let's turn to the new application. It is optimized and it is 4 times faster. Now, computer x needs only 1 hour instead of 4 hours to crunch a similar workunit. The benchmarks haven't changed. The time is a quarter of the old time. Therefore it is now claiming 10 credits based on the time * benchmark method. The other computers are also 4 times faster. Their results are also claiming a quarter of the old credit claims: 7.5 and 12.5. The granted credits will be 10. The credit production rate of computer x still is 10 credits/h. The new application didn't change the credit production rate which is in line with the idea of boinc.
To summarize: The increased credit production rate of the new application has nothing to do with optimization.
in my eyes for some reason almost all other projects started to give less and less and less credits and einstein just more or less stayed "fair". I dont see that here is too much credit and if einstein also decreased it i would need twice the time i needed to get the credits i now have to get the same credis again. that also wouldnt be too good.
And: in comparison i did not get such low credits for S4 like you did (not even with unoptimized app), maybe your pcs were just not optimal for that calculations?
and even if its a bit too high, your "enormous amount of credit" is exagerating.
in my eyes for some reason almost all other projects started to give less and less and less credits and einstein just more or less stayed "fair". I dont see that here is too much credit and if einstein also decreased it i would need twice the time i needed to get the credits i now have to get the same credis again. that also wouldnt be too good.
And: in comparison i did not get such low credits for S4 like you did (not even with unoptimized app), maybe your pcs were just not optimal for that calculations?
and even if its a bit too high, your "enormous amount of credit" is exagerating.
I crunch since October 2004 for BOINC, and always got about the same credits. There was no "less and less and less credits", it's a bit less now on those with the old benchmarking system, because I changed to Linux, but that's not the difference I'm talking about.
There was a real huge inflation when the optimised apps were introduced and with them the "optimised" clients, that fiddled with the benchmarks. But after the incorporation of optimisations in the stock everything should have got back to normal levels.
Concerning my low credit rate: it's an old puter, Athlon 2200+, but that's as well no issue here, as credits for one puter with the same setup and stock-only apps has to get the same from every project per hour of CPU-use. Faster puters should of course get more, but as well the same from every project.
hm, i dont crunch there but seti enchanced for me seems to give much less credits than "normal" seti before, doesnt it?
The only Seti I've crunched since the move to anhanced got me 6.7 credits/h, so it was a bit too high. I don't do it more often, because I like the other science far better. I've just started another one, and I wait to see it's outcome. Just look it up in my Seti account.
Hmm, cant find it now, but shouldnt be ~20 credits per hour the standard in Boinc with the hypothetical referencecomputer? or was it ~12?
Quote:
Concerning my low credit rate: it's an old puter, Athlon 2200+, but that's as well no issue here, as credits for one puter with the same setup and stock-only apps has to get the same from every project per hour of CPU-use. Faster puters should of course get more, but as well the same from every project.
Your running Linux. There are several posts in this board that mentioned a relevant difference in the benchmarks of Linux and Windowsclients, which leads to heavy underscoring of the Linuxclients in the old creditsystem(time*bench) - so you got extra low Cr/h in S4.
Quote:
From Bruce in the toppost:
We are switching to a new uniform system for awarding credits. All users on all platforms will claim the same credit for each workunit, with an amount of credit proportional to the length of the workunit: 'equal credit for equal work'.
Now that the credits in E@H are not more depending on flawed benchmarks Linuxusers gets the same credits as Winusers, they see a extra "Boost" in credits.
At the time of transition to S5 there was a post from Bruce or Gary about the amount of average cranted credits for S5, maybe someone with better memory ca give us a link.
Now that the credits in E@H are not more depending on flawed benchmarks Linuxusers gets the same credits as Winusers, they see a extra "Boost" in credits.
My boost in credits is about 90% on my AMD Athlon XP 2000+. My OS is windows ...
If you are looking at the overall credit production of Einstein you will see that a lot of users are getting a "boost". The credit production of the entire project has increased significantly since the change from S4 to S5.
Saenger, S5 is optimized,
)
Saenger,
S5 is optimized, and doing more work per hour than S4, so the increase in work is giving you better credit. A 4x increase is about what the optimization is doing. So you are right on track.
RE: Saenger, S5 is
)
Imho anything that comes straight from the projects is stock application, and thus deserves plain standard credits.
If some volunteers do some modifications for better performance than stock, it's optimized (and may deserve more credits).
If increasing the science output through better programming on the project side will increase the credits per hour for a puter for that project, I'd recommend for any new project to start with very sloppy programming to have an big increase in credits (and a big jump in user base afterwards) after some time the programming is "optimized".
Grüße vom Sänger
Hi, RE: S5 is optimized,
)
Hi,
seti_enhanced 5.x is also optimized compared to the old seti 4.x. But the new seti doesn't give you more credits per hour than the old seti (if you are comparing the stock applications).
No. The basic idea of boinc is that the stock application of every project gives you approximately the same credits per hour on a given computer system. A new stock application must not break this idea.
To show you that something is wrong let's assume that we still have the old credit system based on time * benchmark method and we still need 3 crunchers to reach the quorum.
Let's say that the old application needs 4 hours to do a workunit on computer x. It is claiming 40 credits based on the time * benchmark method. Let's say that the other results of the quorum are claiming 30 credits and 50 credits. So, granted credits will be 40. The credit production rate of computer x is 10 credits/h.
Now let's turn to the new application. It is optimized and it is 4 times faster. Now, computer x needs only 1 hour instead of 4 hours to crunch a similar workunit. The benchmarks haven't changed. The time is a quarter of the old time. Therefore it is now claiming 10 credits based on the time * benchmark method. The other computers are also 4 times faster. Their results are also claiming a quarter of the old credit claims: 7.5 and 12.5. The granted credits will be 10. The credit production rate of computer x still is 10 credits/h. The new application didn't change the credit production rate which is in line with the idea of boinc.
To summarize: The increased credit production rate of the new application has nothing to do with optimization.
Regards,
Carsten
Would anybody from the
)
Would anybody from the project please explain why the stock application for the S5 run is demanding (and getting) such enormous amount of credit?
And what will be done to stop this unfair behaviour?
Grüße vom Sänger
in my eyes for some reason
)
in my eyes for some reason almost all other projects started to give less and less and less credits and einstein just more or less stayed "fair". I dont see that here is too much credit and if einstein also decreased it i would need twice the time i needed to get the credits i now have to get the same credis again. that also wouldnt be too good.
And: in comparison i did not get such low credits for S4 like you did (not even with unoptimized app), maybe your pcs were just not optimal for that calculations?
and even if its a bit too high, your "enormous amount of credit" is exagerating.
RE: in my eyes for some
)
I crunch since October 2004 for BOINC, and always got about the same credits. There was no "less and less and less credits", it's a bit less now on those with the old benchmarking system, because I changed to Linux, but that's not the difference I'm talking about.
There was a real huge inflation when the optimised apps were introduced and with them the "optimised" clients, that fiddled with the benchmarks. But after the incorporation of optimisations in the stock everything should have got back to normal levels.
Concerning my low credit rate: it's an old puter, Athlon 2200+, but that's as well no issue here, as credits for one puter with the same setup and stock-only apps has to get the same from every project per hour of CPU-use. Faster puters should of course get more, but as well the same from every project.
Grüße vom Sänger
hm, i dont crunch there but
)
hm, i dont crunch there but seti enchanced for me seems to give much less credits than "normal" seti before, doesnt it?
RE: hm, i dont crunch there
)
The only Seti I've crunched since the move to anhanced got me 6.7 credits/h, so it was a bit too high. I don't do it more often, because I like the other science far better. I've just started another one, and I wait to see it's outcome. Just look it up in my Seti account.
Grüße vom Sänger
Hmm, cant find it now, but
)
Hmm, cant find it now, but shouldnt be ~20 credits per hour the standard in Boinc with the hypothetical referencecomputer? or was it ~12?
Your running Linux. There are several posts in this board that mentioned a relevant difference in the benchmarks of Linux and Windowsclients, which leads to heavy underscoring of the Linuxclients in the old creditsystem(time*bench) - so you got extra low Cr/h in S4.
Now that the credits in E@H are not more depending on flawed benchmarks Linuxusers gets the same credits as Winusers, they see a extra "Boost" in credits.
At the time of transition to S5 there was a post from Bruce or Gary about the amount of average cranted credits for S5, maybe someone with better memory ca give us a link.
Hi, RE: Now that the
)
Hi,
My boost in credits is about 90% on my AMD Athlon XP 2000+. My OS is windows ...
If you are looking at the overall credit production of Einstein you will see that a lot of users are getting a "boost". The credit production of the entire project has increased significantly since the change from S4 to S5.
http://boincstats.com/charts/chart_uk_einstein_project_new_credits.gif
Regards,
Carsten