I can only beg to fix that Windows/AMD compiler bug... 25% in my case is almost 14h wasted!!! This is more than 3 SETI WU`s wasted per one E@H WU!
I have just stopped taking new wu from E@H until the problem is fixed. I will let the "QUE" work itself out. I will also suspend E@H after the last one uploads so the computer will take on additional work from the other projects. All 3 of my computers are AMD/window units.
Anyone else have any better idea's?
I'm pulling my AMD's from this project, It's just not a good use of resources. I'll leave the Intels running it for now.
Wow.... the Floating Point Speed is taking a steep dive...!!!
Any ideas why?
acording to the server stats page, the flops count is based off of RAC, and with a 25-40% decrease in credits/hour for higher end CPUs this is a rather large hit.
That's what I've noticed. I'm only getting about 65% in points over time on my Athlon 64 XP 3200, but at the same time some slower 32bit Athlon machines running in the 1.1 -1.6 range have shown little change. May have to switch to Rosetta on the faster machines and let the latest dual core AMD chips as well as their slower above meantioned 32bit chips run Einstein. At least that's the way it looks from here. But tomorrow's a new day.
F. Prefect
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams
I've finished the first two S5R2 units and am half way through the third without any trouble so far. I'm running an old AMD 850 and taking about 50 hours per unit time. Is it just the newer machines that are overclocked that are having trouble?
(edit) is running multiple projects not helping, I run Einstein only.
I know... that`s why i`m running E@H exclusively, right now. I dont have any problems with errors, but the fact, that i hit the same time to complete an s5r2 WU as AMD 850? I got twice the juice in my barton (1.88ghz).
Could you pls give me a link for those workunits of yours? How many credits did you got for them?
My question is why has Linux now become so much slower than Windows?
Is part of the 'optimising' that is to come going to level these OS's out?
The Linux app is much faster on AMD than the Windows app. -> Example.
The differences you found, are caused by the different size of the units. You can not compare just the time for a unit, you must take the different credits into account.
Well, what was said beforehand was that this version contains only bugfixes. I haven't tried it myself, it will be some time until I need new WUs despite a one-day cache- those WUs are just so large ;-)
I have just stopped taking new wu from E@H until the problem is fixed. I will let the "QUE" work itself out. I will also suspend E@H after the last one uploads so the computer will take on additional work from the other projects. All 3 of my computers are AMD/window units.
Which file do you insert this code into or what code this should it replace.
that code says:
1. put on your preferences page at Einstein@Home Resource share to 1
(this will set E@H much less time to crunch than other projects)
2. during next week take a look at the (very few) results you have
3. if there still are errors, go back to step 2
4. if no errors, check message boards to see what said other crunchers
5. if bad opinions, go back to step 2
6. if good opinions, put the the Resource share back how it was before setting it to 1.
PS you can't put resource share to something less than 1, if you put 0, it will be set to 100
I have just stopped taking new wu from E@H until the problem is fixed. I will let the "QUE" work itself out. I will also suspend E@H after the last one uploads so the computer will take on additional work from the other projects. All 3 of my computers are AMD/window units.
Which file do you insert this code into or what code this should it replace.
that code says:
1. put on your preferences page at Einstein@Home Resource share to 1
(this will set E@H much less time to crunch than other projects)
2. during next week take a look at the (very few) results you have
3. if there still are errors, go back to step 2
4. if no errors, check message boards to see what said other crunchers
5. if bad opinions, go back to step 2
6. if good opinions, put the the Resource share back how it was before setting it to 1.
PS you can't put resource share to something less than 1, if you put 0, it will be set to 100
ROFL. I just finished setting up my last machine to run Rosetta until they decide what they are going to do. Personally, I doubt they have clue.
Gary
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams
RE: RE: I can only beg
)
I'm pulling my AMD's from this project, It's just not a good use of resources. I'll leave the Intels running it for now.
RE: RE: Wow.... the
)
That's what I've noticed. I'm only getting about 65% in points over time on my Athlon 64 XP 3200, but at the same time some slower 32bit Athlon machines running in the 1.1 -1.6 range have shown little change. May have to switch to Rosetta on the faster machines and let the latest dual core AMD chips as well as their slower above meantioned 32bit chips run Einstein. At least that's the way it looks from here. But tomorrow's a new day.
F. Prefect
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams
I have Einstein currently
)
I have Einstein currently running on 3 AMD computers.
My AMD X2 4800+ (2.4 GHz) is running Windows and takes about 15.5 odd hours to do a WU, getting about 35% pts/h less than before.
My AMD Opteron 285 (2.6 GHz) runs Linux and now takes around 18 hours per WU despite being a faster processor.
My AMD Opteron 275 (2.2 GHz) runs Linux for 20 hours per WU.
My question is why has Linux now become so much slower than Windows?
Is part of the 'optimising' that is to come going to level these OS's out?
None of my 3 computers are overclocked.
RE: I've finished the first
)
I know... that`s why i`m running E@H exclusively, right now. I dont have any problems with errors, but the fact, that i hit the same time to complete an s5r2 WU as AMD 850? I got twice the juice in my barton (1.88ghz).
Could you pls give me a link for those workunits of yours? How many credits did you got for them?
RE: My question is why has
)
The Linux app is much faster on AMD than the Windows app. -> Example.
The differences you found, are caused by the different size of the units. You can not compare just the time for a unit, you must take the different credits into account.
cu,
Michael
I got my first WUs with
)
I got my first WUs with version 4.17 for Windows. Is this the "only" bugfixed version or are there any Akos optimisations in?
Well, what was said
)
Well, what was said beforehand was that this version contains only bugfixes. I haven't tried it myself, it will be some time until I need new WUs despite a one-day cache- those WUs are just so large ;-)
RE: RE: RE: I have
)
that code says:
1. put on your preferences page at Einstein@Home Resource share to 1
(this will set E@H much less time to crunch than other projects)
2. during next week take a look at the (very few) results you have
3. if there still are errors, go back to step 2
4. if no errors, check message boards to see what said other crunchers
5. if bad opinions, go back to step 2
6. if good opinions, put the the Resource share back how it was before setting it to 1.
PS you can't put resource share to something less than 1, if you put 0, it will be set to 100
All apps are new (dated
)
All apps are new (dated yesterday) : http://einstein.phys.uwm.edu/apps.php
RE: RE: RE: RE: I
)
ROFL. I just finished setting up my last machine to run Rosetta until they decide what they are going to do. Personally, I doubt they have clue.
Gary
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams