is there any table of which client optimization fits best for given processor type ?
I plan to start running E@H on Celeron 1.70Ghz (has 478pins and support SSE2) with 266Mhz ram speed.
But as far as i know celerons have very small cash L1 and L2 if i recall right 32KB and i learned by try and error that certain DC projects require big cash (run much faster on Pentium than on same frequency Celerons). So which client version should i d/l for my CPU (i run Windows XP Pro)
right now i run albert_4.37_windows_intelx86.exe
Copyright © 2024 Einstein@Home. All rights reserved.
which client to run ?
)
Give S39L a try it was built for a small cash.
Hope for the best :)
Anders n
RE: RE: is there any
)
I agree S-39L is very likely best of current offerings for the Einstein science application on your host. That is not, by the way, the client.
S39L is faster than S39 even
)
S39L is faster than S39 even on machines with larger caches. Unless you've got an older (Pre-XP) athlon or a really old pentium (P1/pro) and don't have SSE S39L is the fastest science app. On nonSSE machhines C40's king of the hill.
I don't know what SSE is.
)
I don't know what SSE is. Which version of Akos magic do you recommend for my 1.1 Ghtz Duron with WIN98SE? Thanks, Tom
Look on the web for a program
)
Look on the web for a program called WCPUID. You can use that on your machine to determine all the specifics of your processor and what code it supports. A very handy utility. :)
"Chance is irrelevant. We will succeed."
- Seven of Nine
RE: I don't know what SSE
)
Download CPUZ from here, It will tell you all you need to know about your cpu.
Andy
RE: I don't know what SSE
)
1.1GHz Duron has Morgan core, it supports SSE.
The fastest client for your processor is S39L, but I got the first valid result for the 3DNow! code (D40). :-)
SSE (Streaming SIMD Extensions) is a part of the instruction set of the processors. These instructions can handle 4 numbers simultaneously. SIMD (Single Instruction Multiple Data)
...I got the first valid
)
...I got the first valid result for the 3DNow! code (D40). :-)
When do you think we'll be able to see this new D40 app? How much faster do you think it is compared to the S39L?
Keep up the great work!
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: When do you think we'll
)
I would like to get at least 20-30 valid results without faults. I will upload D40 after this test.
The comparison of 3DNow! against SSE is interesting, because SSE handles four numbers together in front of two, so 3DNow! needs about two times more instruction for the same task.
I hope that 3DNow! will be faster (D40 S40), i like that. :-)
RE: I would like to get at
)
10 Valid results so far ^^
*thumps up*
Might be the first DC Project that makes use of 3dnow!