Hi everyone, first of all it's great to see betas coming early in the science run, the days remaining should start to melt away.
Here are some results for my A64 3700 2.2GHz @ 2.5GHz with Windows XP Pro SP2 and Linux 2.6.15
All results are for small WUs:
Windows
app 4.02 - average for last 14 WUs: 3081.80s
app 4.24 - average for first 14 WUs: 2055.72s
Speed increase - 33.29%
Linux
app 4.01 - average for last 15 WUs: 2359.36s
app 4.17 - average for first 15 WUs: 1998.87s
Speed increase - 15.28%
Looks like the two beta apps are running neck and neck, great job guys!
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
Hi everyone, first of all it's great to see betas coming early in the science run, the days remaining should start to melt away.
Here are some results for my A64 3700 2.2GHz @ 2.5GHz with Windows XP Pro SP2 and Linux 2.6.15
All results are for small WUs:
Windows
app 4.02 - average for last 14 WUs: 3081.80s
app 4.24 - average for first 14 WUs: 2055.72s
Speed increase - 33.29%
Linux
app 4.01 - average for last 15 WUs: 2359.36s
app 4.17 - average for first 15 WUs: 1998.87s
Speed increase - 15.28%
Please correct your speed increase. ;)
You would get 0 sec in case a speed increase of 100%, the way you are calculating.
You better do it on the base of credits/time or just calculate the percentage on the time difference between the to results compared to the lower time.
Windows (4.02): 3081.80s
Windows (4.02): 2055.72s
Diff: 1026.08s
Speedup: (1026.08 / 2055.72) * 100 = 49.91%
That looks even better, doesn't it? :-)
Btw. this is much more speedup as with Intel CPUs(afaik ~33%)
I got simular results with my AMD X2 .
Credit/h and core rose from 23,48 to 35,85.
Compared to other projects this is way to much, ever if I take into account that my X2 is pretty fast. So if the beta-app becomes standard, the credits/WU should get changed to a lower value.
Ahh yes, thanks for pointing that out. The numbers I posted should read:
app 4.24 needs 33.29% less time to crunch a small WU
app 4.17 needs 15.28% less time to crunch a small WU
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
RenaudKener......I know I am off topic here, but I wanted to know if...
Am I reading your Latin phrase as;
Entia, they are not multipying beyond necessity....if wrong, please correct me
thks in advance...
Thi is known as Occam's Razor, and is the basis of all unified field theories, from Newton to Maxwell, Einstein, Salam etc.
Tullio
Tullio, thanks, you are right , this translates in English to : "entities" should not be multiplied beyond necessity which is, of course, Occam's Razor
It looks like the speedup of 50% is comparable on old PIII Intel boxes, too! My old, crusty dually box also got a tremendous speed boost, and all has validated so far. This should get changed to the released app just as soon as possible in order to start getting some of the backlog worked through so that it won't take over a year to work through the S5 data. Thanks to Bernd and Akos for squeezing the most out of our machines!!
This should get changed to the released app just as soon as possible in order to start getting some of the backlog worked through so that it won't take over a year to work through the S5 data.
Give me some time to take a closer look at the results from these Apps. I think the Apps will be made official ones some time at the end of next week (or shortly after the weekend), if nothing suspicious shows up. We will then adjust the credit shortly after that, too, to stay conformant to the other projects.
RE: The stderr out for this
)
please, read reports for v 4.16 for explanation about "small x".
It is a debugging feature :-) nothing to worry about.
"Entia non sunt multiplicandam praeter necessitatem"
(OKHAM)
RenaudKener......I know I am
)
RenaudKener......I know I am off topic here, but I wanted to know if...
Am I reading your Latin phrase as;
Entia, they are not multipying beyond necessity....if wrong, please correct me
thks in advance...
Hallo all, I compare my first
)
Hallo all, I compare my first result from application 4.24 with applications for Linux on the same PC:
OS - App - WU - credit - time [sec] - credit/hour
Linux - 4.01 - h1_1278.5 - 176,31 - ~ 29108 - 21,81
Linux - 4.17 - h1_1278.5 - 176,31 - ~ 24900 - 25,49
Win2K - 4.24 - h1_0801.5 - 162,29 - 25574 - 22,85
Hi everyone, first of all
)
Hi everyone, first of all it's great to see betas coming early in the science run, the days remaining should start to melt away.
Here are some results for my A64 3700 2.2GHz @ 2.5GHz with Windows XP Pro SP2 and Linux 2.6.15
All results are for small WUs:
Windows
app 4.02 - average for last 14 WUs: 3081.80s
app 4.24 - average for first 14 WUs: 2055.72s
Speed increase - 33.29%
Linux
app 4.01 - average for last 15 WUs: 2359.36s
app 4.17 - average for first 15 WUs: 1998.87s
Speed increase - 15.28%
Looks like the two beta apps are running neck and neck, great job guys!
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
RE: RenaudKener......I know
)
Thi is known as Occam's Razor, and is the basis of all unified field theories, from Newton to Maxwell, Einstein, Salam etc.
Tullio
RE: Hi everyone, first of
)
Please correct your speed increase. ;)
You would get 0 sec in case a speed increase of 100%, the way you are calculating.
You better do it on the base of credits/time or just calculate the percentage on the time difference between the to results compared to the lower time.
Windows (4.02): 3081.80s
Windows (4.02): 2055.72s
Diff: 1026.08s
Speedup: (1026.08 / 2055.72) * 100 = 49.91%
That looks even better, doesn't it? :-)
Btw. this is much more speedup as with Intel CPUs(afaik ~33%)
I got simular results with my AMD X2 .
Credit/h and core rose from 23,48 to 35,85.
Compared to other projects this is way to much, ever if I take into account that my X2 is pretty fast. So if the beta-app becomes standard, the credits/WU should get changed to a lower value.
cu,
Michael
Ahh yes, thanks for pointing
)
Ahh yes, thanks for pointing that out. The numbers I posted should read:
app 4.24 needs 33.29% less time to crunch a small WU
app 4.17 needs 15.28% less time to crunch a small WU
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national deficit! We used to call them astronomical numbers. Now we should call them economical numbers. - Richard Feynman
RE: RE: RenaudKener......
)
Tullio, thanks, you are right , this translates in English to : "entities" should not be multiplied beyond necessity which is, of course, Occam's Razor
It looks like the speedup of
)
It looks like the speedup of 50% is comparable on old PIII Intel boxes, too! My old, crusty dually box also got a tremendous speed boost, and all has validated so far. This should get changed to the released app just as soon as possible in order to start getting some of the backlog worked through so that it won't take over a year to work through the S5 data. Thanks to Bernd and Akos for squeezing the most out of our machines!!
RE: This should get changed
)
Give me some time to take a closer look at the results from these Apps. I think the Apps will be made official ones some time at the end of next week (or shortly after the weekend), if nothing suspicious shows up. We will then adjust the credit shortly after that, too, to stay conformant to the other projects.
BM
BM