On November 19, 2005 1:19 pm, Randall R Schulz wrote:

Mike,

On Saturday 19 November 2005 09:27, Mike wrote:

...

I have a dual boot machine. When I checks the stats on the seti page I noticed that they were different when running XP, or SuSE. The machine is mostly idle. Is anyone else running setiathome? Any thoughts on boosting perfromance?

Running under XP ... % of time BOINC client is running 81.8484 %

Running under SuSE 9.3 with X ... % of time BOINC client is running 58.6482 %

How similarly are the two BOINC clients configured? In particular, is one set to compute constantly and the other tied to screen-saver

They have the share the same configuration file. At first I thought that maybe it was computing the average including the time that the Other OS was running, but the number haven't changed much since I stopped running XP on the machine. I thought I had 9.3 on the machine but it is an upgrade 9.0 would a 2.6 kernel make that much difference? Just tried an upgrade but it yast can't read the packages from the SuSE dvd. -- Collector of vintage computers http://www.ncf.ca/~ba600 Machines to trade http://www.ncf.ca/~ba600/trade.html Open Source Weekend http://www.osw.ca

Hans du Plooy wrote:

...

On Saturday 19 November 2005 09:27, Mike wrote:

...

I have a dual boot machine. When I checks the stats on the seti page I noticed that they were different when running XP, or SuSE. The machine is mostly idle. Is anyone else running setiathome? Any thoughts on boosting perfromance?

I just use the old setiathome binary (haven't actually had time to bother with the BOINC client, and seeing as setiathome still works, I'm not worried). Let it run in the background with -nice 19 and it will eat up an idle cycle without making your PC unresponsive. I don't know about the CPU per hour, but my machine (2ghz AthlonXP) takes on avarage 4 hours to complete a packet. Sometimes a little less, sometimes a little more.

Hans

Hans, Do not foret that, accordingly to berkeley.edu, the old seti@home client will cease to work on this December, 15. -- Rui Santos http://www.ruisantos.com/

On Mon, 28 Nov 2005 16:30:49 +0200 H du Plooy wrote:

I checked out the BOINC client, I'm not that impressed. You now require a username and password. I collaborated with a friend using his old (university, doesn't exist anymore) e-mail address, so now we can't continue.

As far as the (web) *server* is concerned, you can still log on with the account ID instead of with the e-mail address. I'm less certain about the (pc) *client*, but I believe that __all__ you have to do is to REPEAT whatever it was that you entered on the "change password" page at the web server. mikus

Mike, On Saturday 19 November 2005 19:55, Mike wrote:

On November 19, 2005 12:27 pm, Mike wrote:

...

I have a dual boot machine. When I checks the stats on the seti page I noticed that they were different when running XP, or SuSE. The machine is mostly idle. Is anyone else running setiathome? Any thoughts on boosting perfromance?

Measured floating point speed 733.63 million ops/sec Measured integer speed 1895.05 million ops/sec % of time BOINC client is running 58.6482 %

Just finished an upgraded to 9.3 and not the change I was expecting

Measured floating point speed 676.59 million ops/sec Measured integer speed 2025 million ops/sec % of time BOINC client is running 59.1031 %

What change did you expect and why did you expect any change at all? On how many runs were each of these reports based? How tightly did you control the mix of other processes running concurrently? Judging from those numbers, I'd suspect that they are within the range of normal variation and in fact reflect no real difference. Keep in mind, too, that cache utilization is a critical performance factor. Any other processes executing concurrently compete for cache use and degrade the performance of the BOINC client. The bottom line is that I don't think you can really infer any change based on those observations. Randall Schulz

Theo v. Werkhoven wrote:

Sat, 19 Nov 2005, by kenziem@sympatico.ca:

...

I have a dual boot machine. When I checks the stats on the seti page I noticed that they were different when running XP, or SuSE. The machine is mostly idle. Is anyone else running setiathome? Any thoughts on boosting perfromance?

Running under XP GenuineIntel Intel(R) Celeron(R) CPU 2.93GHz Microsoft Windows XP SP2, (05.01.2600.00)

Measured floating point speed 1390.1 million ops/sec Measured integer speed 2412.94 million ops/sec

Running under SuSE 9.3 with X GenuineIntel Intel(R) Celeron(R) CPU 2.93GHz Linux 2.4.21-297-default

Measured floating point speed 733.63 million ops/sec Measured integer speed 1895.05 million ops/sec

...

From what I read on the einstein@home forum there are significant differences between the Intel compiler that was used for the Windows client and gcc for Linux, with the Intel one being (far) superior in speed optimization for i686 code (I think gcc delivers smaller object-file though). Of course the Intel compiler /only/ makes i386 code, not code for a dozen other processor like gcc. And, being the "inventor", they

I've seen the same thing here. Even in wine the Windows client is faster than the Linux one. probably know some tricks the gcc developers don't, or don't think is important enough to implement.

Theo

Does anyone know if there is a gui BOINC client manager for KDE similar to KSetiSpy or Ksetiwatch? Jim

On Monday 21 November 2005 18:47, Linuxjim wrote:

Does anyone know if there is a  gui BOINC client manager for KDE similar to KSetiSpy or Ksetiwatch?

KBoincSpy: http://kboincspy.sourceforge.net/ -- Marcus Lund