Hi Akosf,...we have an older 400MHz Pentium-II IBM laptop running Win98 and the figures are as follows: C40=41422 sec...C41.00=37372sec. That's a 10% increase (over one hour less!)in productivity. It has validated OK as well. Your genius has done it again! Thanks for keeping the older machines current....Cheers, Rog.
I ran a controlled test using the same short Work Unit I've used for previous testing.
Yes, I know akosf did not suggest C41.00 for SSE-capable machines, but given some of the odd architecture dependent results I thought an extra data point might be of use:
The short Einstein result was r1_0265.5_2133_S4R2a_0) and I ran it on my Gallatin (Northwood-descended 2M L3 cache--the first P4 EE).
hyperthreaded, with the other BOINC job also an Einstein job:
Mostly this states again just how surprisingly badly S40.12 works on this machine--nearly as long an execution time as C41.00, which is not even using an SSE capability.
C41.00 Observation Thread
)
Sorry
after i doenload the alb_C4100.zip
my PC tell me that it is a broken file
the file size about 1020KB?? (is so big ??)
RE: Sorry after i doenload
)
Thanks. The zipped file was wrong.
Rezipped one is uploaded.
RE: C41.00 was made for old
)
My AMD-K6 400MHz don´t like the C41.
On 1407 WUs
C41 = 64,283.27 s
D40 = 56,740.65 s
I changed back to D40
Chris
Thank you for your excellent work
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
RE: RE: C41.00 was made
)
It likes, but you don't. :)
Probably your K6 is a K6-2 or K6-III because it's able to execute 3DNow! instructions.
Best,
Akos
Hi Akosf,...we have an older
)
Hi Akosf,...we have an older 400MHz Pentium-II IBM laptop running Win98 and the figures are as follows: C40=41422 sec...C41.00=37372sec. That's a 10% increase (over one hour less!)in productivity. It has validated OK as well. Your genius has done it again! Thanks for keeping the older machines current....Cheers, Rog.
RE: My AMD-K6 400MHz
)
The C41 was never ment to be faster than D40. C41 is for the processors, which doens't support SSE or 3DNow!
RE: RE: RE: C41.00 was
)
I like all Alberts from Hungary :-)
You are right, it is a K6 with 3DNow, was my mistake, sorry.
Model/Typ/FSB (MHz)/L1-Cache
8/K6-2 400 AFQ/66/100/32 kByte + 32 kByte
http://www.elektronik-kompendium.de/sites/com/0406071.htm
Sorry for the German site
Best Wishes and Happy compiling
Chris
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
I ran a controlled test using
)
I ran a controlled test using the same short Work Unit I've used for previous testing.
Yes, I know akosf did not suggest C41.00 for SSE-capable machines, but given some of the odd architecture dependent results I thought an extra data point might be of use:
The short Einstein result was r1_0265.5_2133_S4R2a_0) and I ran it on my Gallatin (Northwood-descended 2M L3 cache--the first P4 EE).
hyperthreaded, with the other BOINC job also an Einstein job:
S40.12 2345 seconds
S40.04 1933 seconds
C41.00 2486 seconds
Mostly this states again just how surprisingly badly S40.12 works on this machine--nearly as long an execution time as C41.00, which is not even using an SSE capability.
On my computer - 300MHz
)
On my computer - 300MHz Pentium is c41.00 12% faster than c40.
great work akosf - thanks
RE: On my computer - 300MHz
)
That sounds well.
I put a modified look-up table handling into C41.01, because I realized that my previous codes have a clumsy part that reduces the performance.