Hey guys, just a quick update. My WU has about 2 hours left; total crunching time should amount to between 20.5 and 21 hours. I still don't know the exact credit value, though. Btw, I'm getting a friend from Uni to check this with one or two of his AMD boxes, so we'll get some more results. Mailing Bernd is a great idea imo.
Hi Annika!
OK, so please stay home, don't go out, and let us know in 2 hours time :-) :-) :-)
I got him to check on his Opteron X2... being quite new (goodness, that box made me drool when he bought it) this box supports anything you might want ;-) Plus, I have results under both Windows and Linux in my spreadsheet, so comparing will be easy. That was the box which got the huge 70% penalty, btw.
I got him to check on his Opteron X2... being quite new (goodness, that box made me drool when he bought it) this box supports anything you might want ;-) Plus, I have results under both Windows and Linux in my spreadsheet, so comparing will be easy. That was the box which got the huge 70% penalty, btw.
Excellent! I would expect the gap to narrow significantly, but it won't be closed completely.
Results for my WU are just in... 310 credits in 21.4 hours, so, about 14.5 credits/hour. Now I need your help interpreting this, since I don't have results with the "normal" Windows app from that box. It can do 18.3 C/H under Linux, so, Linux would still be 26% faster. But I guess the penalty would have been greater with the standard app?
Btw, first (very rough) estimates for the Opteron show that it is about 20% faster after patching.
Results for my WU are just in... 310 credits in 21.4 hours, so, about 14.5 credits/hour. Now I need your help interpreting this, since I don't have results with the "normal" Windows app from that box. It can do 18.3 C/H under Linux, so, Linux would still be 26% faster. But I guess the penalty would have been greater with the standard app?
Btw, first (very rough) estimates for the Opteron show that it is about 20% faster after patching.
Have a look at this post. To compare, just divide your c/h values through cpu clock rate.
Running at 2.2 GHz, it would be a ratio of 6.6 (C/H)/clock speed. This certainly looks better than your results with the unpatched app :-) As the WU was already a few % (not more than 10 or 12, maybe less) finished when I applied the hack it might even be a bit better than that.
Results for my WU are just in... 310 credits in 21.4 hours, so, about 14.5 credits/hour. Now I need your help interpreting this, since I don't have results with the "normal" Windows app from that box. It can do 18.3 C/H under Linux, so, Linux would still be 26% faster. But I guess the penalty would have been greater with the standard app?
Btw, first (very rough) estimates for the Opteron show that it is about 20% faster after patching.
I googled around (all the results pages are indexed :-) ) to find similar hosts, e.g. this one
And what I got was 10.5 to 11.5 CCH , which would mean a ca 30 % speedup just as expected :-). So, still no reason to crunch under Windows when Linux is available, but the diff between Win and Linux was reduced from a rediculous runtime factor of 1.6 to a more acceptable 1.3 on your machine.
This sounds good. I mean, personally, I still won't be using Windows much for crunching, but it's great we figured this out :-) and if it could be made available to everyone, e.g., all the 80% who can't or don't want to use Linux, the project could benefit quite a lot...
This sounds good. I mean, personally, I still won't be using Windows much for crunching, but it's great we figured this out :-) and if it could be made available to everyone, e.g., all the 80% who can't or don't want to use Linux, the project could benefit quite a lot...
2 team mates have already patched there AMDs, in one case 19 x A64 and 6 x A64 X2. :-))
Goodness, huge farms :-D makes my single AMD box (and even my friend's three, two of which are SSE2 capable) pale in comparison. Of course with so many boxes the effect will be much more noticeable; I'm sure it will pay of for both the team and the project. Still, what about the "not so many patched clients" policy?
RE: Hey guys, just a quick
)
Hi Annika!
OK, so please stay home, don't go out, and let us know in 2 hours time :-) :-) :-)
Your friends AMD boxes support SSE2?
CU
BRM
I got him to check on his
)
I got him to check on his Opteron X2... being quite new (goodness, that box made me drool when he bought it) this box supports anything you might want ;-) Plus, I have results under both Windows and Linux in my spreadsheet, so comparing will be easy. That was the box which got the huge 70% penalty, btw.
RE: I got him to check on
)
Excellent! I would expect the gap to narrow significantly, but it won't be closed completely.
CU
BRM
Results for my WU are just
)
Results for my WU are just in... 310 credits in 21.4 hours, so, about 14.5 credits/hour. Now I need your help interpreting this, since I don't have results with the "normal" Windows app from that box. It can do 18.3 C/H under Linux, so, Linux would still be 26% faster. But I guess the penalty would have been greater with the standard app?
Btw, first (very rough) estimates for the Opteron show that it is about 20% faster after patching.
RE: Results for my WU are
)
Have a look at this post. To compare, just divide your c/h values through cpu clock rate.
cu,
Michael
Running at 2.2 GHz, it would
)
Running at 2.2 GHz, it would be a ratio of 6.6 (C/H)/clock speed. This certainly looks better than your results with the unpatched app :-) As the WU was already a few % (not more than 10 or 12, maybe less) finished when I applied the hack it might even be a bit better than that.
RE: Results for my WU are
)
I googled around (all the results pages are indexed :-) ) to find similar hosts, e.g. this one
http://einsteinathome.org/host/930888/tasks
And what I got was 10.5 to 11.5 CCH , which would mean a ca 30 % speedup just as expected :-). So, still no reason to crunch under Windows when Linux is available, but the diff between Win and Linux was reduced from a rediculous runtime factor of 1.6 to a more acceptable 1.3 on your machine.
CU
BRM
This sounds good. I mean,
)
This sounds good. I mean, personally, I still won't be using Windows much for crunching, but it's great we figured this out :-) and if it could be made available to everyone, e.g., all the 80% who can't or don't want to use Linux, the project could benefit quite a lot...
RE: This sounds good. I
)
2 team mates have already patched there AMDs, in one case 19 x A64 and 6 x A64 X2. :-))
cu,
Michael
Goodness, huge farms :-D
)
Goodness, huge farms :-D makes my single AMD box (and even my friend's three, two of which are SSE2 capable) pale in comparison. Of course with so many boxes the effect will be much more noticeable; I'm sure it will pay of for both the team and the project. Still, what about the "not so many patched clients" policy?