View Full Version : HELP! My SETI efficiency is terrible!
I've just installed version 3.03 of the seti client, and setispy is reporting my efficiency to be around 12 cycles per flop. The very best I've seen is 8.47, and that was while the screen was blanked. I have it set up to process data all the time, and for the screensaver to go blank. The system is as follows: Athlon 1333 @ 1570, FSB @157, with the multiplier set at 10. The memory is 512 MB of Micron pc2100 DDR ram. The memory timing settings are all set to the most aggressive except for the command something-or-other, which is set to 2T, since it's unstable at 1T. When I hit cntrl-alt-del, the only programs running are explorer, systray, and seti. With these specs, I can't imagine why the efficiency is so low. I mean, I have plenty of memory, it's running at a fairly high fsb speed, and there aren't many other processes running. (The OS is win 98 version A, by the way.) I've checked the BIOS, and there's nothing obvious wrong like the L1 or L2 cache being disabled. Any ideas?
Don't run the screen saver version...it eats up alot of cycles.
Download the command line version and you'll be set. You can get them all from the team website here (www.ocsetiteam.com). Get SETI Driver, SETISpy and the CLC version and then follow the directions to set them all up. You'll be flying.
I have to second Jon here. Even though it sounds like you're doing all the right things, just give the command line version a shot and see if that changes anything.
LandShark
12-18-01, 11:07 PM
w/ ur Athlon 1.57G, i'm sure u'll dip into the mid-4hrs / WU range(!), if u switch to the CLC version.
give it a shot!! :beer:
Mictlan
12-19-01, 08:17 AM
Guys, you seems to have misread it.....
Originally posted by v8440
I've just installed version 3.03 of the seti client, and setispy is reporting my efficiency to be around 12 cycles per flop. The very best I've seen is 8.47, and that was while the screen was blanked. I have it set up to process data all the time, and for the screensaver to go blank.
Correct me if I'm wrong, v8440. You d/l the clc and installed it in your PC. Before doing that one you already installed the screensaver. Right?
The problem is that you are running two instance of CETI@home with a single CPU. Either, you unistall the screensaver or you stop the clc. You don't need the clc to run the screensaver. If you change the screensaver and later check in SETISpy the instantaneous performance, you shall see that your effic¿iency will go up.
The people who already posted above were in fact correct-I had only installed the screensaver version. After posting that, I later completely deleted the screensaver version, and installed the clc, only to get an error message saying something to the effect that this version of clc isn't compiled to run on my type of processor and operating system. The weird thing is, I installed the version designed for NT 4. It had some other operating systems listed, but I thought that if it would run on NT that it would run on 98.
LandShark
12-19-01, 10:41 AM
yes, u r right. the NT version should run on the 9x machines.
just follow Jon's direction to the team's site. from there, go to "Links", there, u'll find an article 'bout how to get CLC, SETISpy, and SETIDriver to work together....
keep crunching......;)
Now getting the CLC will most cirtanly help out times, the effiency is still horible for even the Screen Saver version. When running the SS version, I usually get just what I get with the CLC. Anywhere from 6-8CpF average. Something weird is with his system. The only thing that I can think of would be the SS displaying the grpahics, but he had it to blank screen, and that is how I had it. Anybody got suggestions for what could be pushing up his CpF?? I don't think the CLC will help out for the effiency, but who knows? Mabey the SS refuses to run on NT correctly!
JigPu
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.