which benchmark should I tweak with for SETI?

[cakin]

I want my machines to be optimal for SETI.
If I use Sandra, should I optimize using the CPU, CPU Multimedia or Memory Benchmark settings?
Or is there another benchmarking program which would be better to use?

I can see several arguements:
CPU, its raw computing that matters,
Multimedia, since its CPU and having to write to the disk,
or Memory,

Also which is most important Integer of Float?

 

[Poof]

With SETI, FSB and memory bandwidth seem to make all the difference. Plus being able to run your memory at CAS 2.

 

[Sukhoi]

Yep, Poof covered it.

Are you on Team AnandTech? I don't recognize your name.

 

[JWMiddleton]

Hi cakin,

What Poof said is correct. I have a couple of CUSL2 boards. One is running a PIII-700e @945 MHz. Getting Micron CAS2 RAM (www.crucial.com) and running it at CAS2 with most optimized settings knocked over 30 minutes off my per WU time.

I also have a couple boards based on VIA chipsets. The same memory set to 4-Way Interleave, CAS2, and TURBO, along with a BIOS upgrade knocked over 1 hours off my time with the previous version of SETI.

You will find that the PIII-667 will do the WUs as fast, or faster, than the Cel2-667/900. I have a Cel2-700 that will only run at 904 MHz and it really SUX at SETI.

I use Sandra's memory benchmark to tell if my memory tweaking is helping. Then SETI itself becomes your best benchmark. You just need to understand that the Angle-Range(AR) of the WU will effect the time. So, compare those of a similar AR. For info on AR search this forum. SetiSpy and SetiQ are good source to determine times and AR.

Good luck! And, if you haven't joined our TeAm then consider this an invitation!