Don't worry about the english to much, We understood what you were saying, and besides its not spoken everywhere in the world.
When someone starts speaking Klingon I'll start to worry, as that would mean that our Earth DC Projects are expanding beyond your recognised borders. :)
Für ausführliche Erklärung danke. Nicht für dein Englisch bitte sich entschuldigen. Es ist weit besser als mein Deutscher.
I had to use Google to translate the above. German was not an option when and where I went to School.
Regards
Franz
No Problem, it is easier to my to read English than to write or speak but after this fabuluos timeout of the servers you will understand why people with fast machines like to have some more workunits. blöd_brot won't answer to this thread weekend, he is on a marriage to Liverpool: http://www.heise.de/foren/go.shtml?read=1&msg_id=10523898&forum_id=74338
I’ve been a new member for just over 2 weeks, during this time I have noticed a few oddities. From reading the posts and looking at some of the results obtained by fellow crunchers, I noticed that there seems to be 3 versions of Einstein. They are as follows:
1. Standard Einstein 4.37 that gets results as shown by the 1st example below.
2. Enhanced Einstein by Akosf that gets results as shown by the 2nd and 3rd examples below.
3. Super Charged Einstein by Akosf that gets results as demonstrated by the 4th and 5th examples below.
The computer in the 2nd and 3rd example should be faster than the last 2 but they aren’t. My question is why???
Can anyone please explain the difference or trick? I’ve tried using the 5.4.9 Client on the 3rd example but reverted to Truxoft’s 5.3.12tx36 Client since there was no noticeable difference.
Examples:
Intel Pentium 4 CPU 3.40 GHz, 1 GB RAM, ? MB L2, 800 MHz FSB
5.4.6 core client and standard 4.37 (average of 59 work units)
CPU time = 13,923.19 sec., Claimed credit = 8.65 CS/hr., Granted credit = 8.69 CS/hr.
Intel Pentium 4 CPU 3.40 GHz, 1 GB RAM, ? MB L2, 800 MHz FSB
5.3.12.tx36 core client with Akosf Ver U41.04 (average of 76 work units)
CPU time = 4,737.50 sec., Claimed credit = 24.62 CS/hr., Granted credit = 25.05 CS/hr.
Pentium 4, 650 3.4 GHz, 2 GB RAM, 2 MB L2, 800 MHz FSB
5.3.12.tx36 core client with Akosf Ver U41.04 (average of 25 work units)
CPU time = 5,568.02 sec., Claimed credit = 13.79 CS/hr., Granted credit = 21.13 CS/hr.
Intel Pentium 4 640(1) CPU 3.20GHz, 1 GB RAM, 2 MB L2, 800 MHz FSB
5.4.9 core client with Akosf Ver S41.07 (average of 50 work units)
CPU time = 3,096.05 sec., Claimed credit = 66.45 CS/hr., Granted credit = 55.71 CS/hr.
Intel Pentium D 805 CPU 2.66 GHz, 1 GB RAM, 2 x 1 MB L2, 533 MHz FSB
5.4.9 core client with Akosf Ver U41.04 (average of 34 work units)
CPU time = 2,479.36 sec., Claimed credit = 71.80 CS/hr., Granted credit = 61.37 CS/hr.
I just noticed that your XP1700+ box that you commented on in the Milestones thread has just had 5 unhandled exception errors, all from the same data file z1_1212.5, quite recently. Any further thoughts on what might be causing these errors? I haven't seen any more of these errors on any of my boxes although I would only notice if there were a whole string of them like the ones I reported previously. I haven't seen that behaviour occur since that time.
While I've (hopefully) got your attention, I'd like to ask you a question that I'm sure others would be interested it. When the new standard app is released with the S5 science run, we are all expecting to see considerable performance improvement over the current standard app, thanks to many improvements, some of which we can undoubtedly thank you for. However, from the beta apps posted recently by Bernd, it would appear that your techniques will still provide a further performance enhancement on top of what is going to be in the official app. So my question is this - Do you intend to take the official app when released and further optimise it for specific platforms, like you did previously? Could we perhaps expect to see another drop-in replacement that would perhaps give a 50-100% speedup over the official app?
Sorry if you've already commented on this previously and I've not seen it.
I just noticed that your XP1700+ box that you commented on in the Milestones thread has just had 5 unhandled exception errors, all from the same data file z1_1212.5, quite recently. Any further thoughts on what might be causing these errors? I haven't seen any more of these errors on any of my boxes although I would only notice if there were a whole string of them like the ones I reported previously. I haven't seen that behaviour occur since that time.
I don't know its ground, but i will look after.
Quote:
So my question is this - Do you intend to take the official app when released and further optimise it for specific platforms, like you did previously? Could we perhaps expect to see another drop-in replacement that would perhaps give a 50-100% speedup over the official app?
I changed only a small ( but important ) part of the whole code. I presume that there are some new interesting opportunities and i have some untried ideas. I think that i will try out them because Einstein@Home is my favourite distributed computing project, although each new idea needs more time than previous and gives smaller increment. So, the time is my only problem.
.... I think that i will try out them because Einstein@Home is my favourite distributed computing project, although each new idea needs more time than previous and gives smaller increment. So, the time is my only problem.
Best,
Akos
Hi Akos,
Thanks very much for the reply. I'm sure there will be a large bunch of people cheering you on from the sidelines :). EAH is my favourite project too and I can't overstate how much I appreciate what you have done to speed up the potential rate at which the science can be done. A long time ago I remember Bruce commenting on how the servers had been designed to cope with a much larger user base that would be needed to get the work done. Your speedups must be having a very positive benefit even now when most participants are probably still using the standard app. I'm really looking forward to the changeover to S5 when every participant will be using a significantly faster app. The production rate should really take off then :).
I just noticed that your XP1700+ box that you commented on in the Milestones thread has just had 5 unhandled exception errors, all from the same data file z1_1212.5, quite recently. Any further thoughts on what might be causing these errors? I haven't seen any more of these errors on any of my boxes although I would only notice if there were a whole string of them like the ones I reported previously. I haven't seen that behaviour occur since that time.
I don't know its ground, but i will look after.
Probably i found the reason of this fault. Really stupid and simple thing.
I will modify the 41.xx codes for this 10 days... :)
RE: Sorry about my bad
)
Hallo Faeshn:
Für ausführliche Erklärung danke. Nicht für dein Englisch bitte sich entschuldigen. Es ist weit besser als mein Deutscher.
I had to use Google to translate the above. German was not an option when and where I went to School.
Regards
Franz
RE: Sorry about my bad
)
Don't worry about the english to much, We understood what you were saying, and besides its not spoken everywhere in the world.
When someone starts speaking Klingon I'll start to worry, as that would mean that our Earth DC Projects are expanding beyond your recognised borders. :)
RE: RE: Hallo
)
No Problem, it is easier to my to read English than to write or speak but after this fabuluos timeout of the servers you will understand why people with fast machines like to have some more workunits. blöd_brot won't answer to this thread weekend, he is on a marriage to Liverpool:
http://www.heise.de/foren/go.shtml?read=1&msg_id=10523898&forum_id=74338
Keep on crunching Franz!
faeshn
I’ve been a new member for
)
I’ve been a new member for just over 2 weeks, during this time I have noticed a few oddities. From reading the posts and looking at some of the results obtained by fellow crunchers, I noticed that there seems to be 3 versions of Einstein. They are as follows:
1. Standard Einstein 4.37 that gets results as shown by the 1st example below.
2. Enhanced Einstein by Akosf that gets results as shown by the 2nd and 3rd examples below.
3. Super Charged Einstein by Akosf that gets results as demonstrated by the 4th and 5th examples below.
The computer in the 2nd and 3rd example should be faster than the last 2 but they aren’t. My question is why???
Can anyone please explain the difference or trick? I’ve tried using the 5.4.9 Client on the 3rd example but reverted to Truxoft’s 5.3.12tx36 Client since there was no noticeable difference.
Examples:
Intel Pentium 4 CPU 3.40 GHz, 1 GB RAM, ? MB L2, 800 MHz FSB
5.4.6 core client and standard 4.37 (average of 59 work units)
CPU time = 13,923.19 sec., Claimed credit = 8.65 CS/hr., Granted credit = 8.69 CS/hr.
Intel Pentium 4 CPU 3.40 GHz, 1 GB RAM, ? MB L2, 800 MHz FSB
5.3.12.tx36 core client with Akosf Ver U41.04 (average of 76 work units)
CPU time = 4,737.50 sec., Claimed credit = 24.62 CS/hr., Granted credit = 25.05 CS/hr.
Pentium 4, 650 3.4 GHz, 2 GB RAM, 2 MB L2, 800 MHz FSB
5.3.12.tx36 core client with Akosf Ver U41.04 (average of 25 work units)
CPU time = 5,568.02 sec., Claimed credit = 13.79 CS/hr., Granted credit = 21.13 CS/hr.
Intel Pentium 4 640(1) CPU 3.20GHz, 1 GB RAM, 2 MB L2, 800 MHz FSB
5.4.9 core client with Akosf Ver S41.07 (average of 50 work units)
CPU time = 3,096.05 sec., Claimed credit = 66.45 CS/hr., Granted credit = 55.71 CS/hr.
Intel Pentium D 805 CPU 2.66 GHz, 1 GB RAM, 2 x 1 MB L2, 533 MHz FSB
5.4.9 core client with Akosf Ver U41.04 (average of 34 work units)
CPU time = 2,479.36 sec., Claimed credit = 71.80 CS/hr., Granted credit = 61.37 CS/hr.
Regards
Franz
at a guess machines 2 and 3
)
at a guess machines 2 and 3 are running a 2nd project and not 100% einstein.
RE: 2)Intel Pentium 4 CPU
)
Probably 2nd and 3rd run with HT.
Hi Akosf, I just noticed
)
Hi Akosf,
I just noticed that your XP1700+ box that you commented on in the Milestones thread has just had 5 unhandled exception errors, all from the same data file z1_1212.5, quite recently. Any further thoughts on what might be causing these errors? I haven't seen any more of these errors on any of my boxes although I would only notice if there were a whole string of them like the ones I reported previously. I haven't seen that behaviour occur since that time.
While I've (hopefully) got your attention, I'd like to ask you a question that I'm sure others would be interested it. When the new standard app is released with the S5 science run, we are all expecting to see considerable performance improvement over the current standard app, thanks to many improvements, some of which we can undoubtedly thank you for. However, from the beta apps posted recently by Bernd, it would appear that your techniques will still provide a further performance enhancement on top of what is going to be in the official app. So my question is this - Do you intend to take the official app when released and further optimise it for specific platforms, like you did previously? Could we perhaps expect to see another drop-in replacement that would perhaps give a 50-100% speedup over the official app?
Sorry if you've already commented on this previously and I've not seen it.
Cheers,
Cheers,
Gary.
Hi Gary, RE: I just
)
Hi Gary,
I don't know its ground, but i will look after.
I changed only a small ( but important ) part of the whole code. I presume that there are some new interesting opportunities and i have some untried ideas. I think that i will try out them because Einstein@Home is my favourite distributed computing project, although each new idea needs more time than previous and gives smaller increment. So, the time is my only problem.
Best,
Akos
RE: .... I think that i
)
Hi Akos,
Thanks very much for the reply. I'm sure there will be a large bunch of people cheering you on from the sidelines :). EAH is my favourite project too and I can't overstate how much I appreciate what you have done to speed up the potential rate at which the science can be done. A long time ago I remember Bruce commenting on how the servers had been designed to cope with a much larger user base that would be needed to get the work done. Your speedups must be having a very positive benefit even now when most participants are probably still using the standard app. I'm really looking forward to the changeover to S5 when every participant will be using a significantly faster app. The production rate should really take off then :).
Cheers,
Cheers,
Gary.
RE: RE: I just noticed
)
Probably i found the reason of this fault. Really stupid and simple thing.
I will modify the 41.xx codes for this 10 days... :)