how are my seti times......?

[nickburns]

*AMD 1ghz TB with 256mb sdram, win2k.......4.5hours/WU
*Celeron 500 with 128mb sdram, win98.......14 hours/WU (but it is heavily used)
*p3 450 with 128mb sdram, win98.......11 hours/WU

all computers are running CLi v3.0
just wanted to know how these times are. When the 1 ghz was running win98SE it was doing WU's in about 3.5 hours? can it be that the O/S makes a difference?

 

[Robor]

I've found WinNT/Win2K faster in SETI, especially if you get some of the tougher WU's.

I think your first 2 times look okay if your Celeron500 is used a lot. The P3-450 should be able to do a WU faster than 11 hours. Are you using a screen saver of any sort? How long have you been watching them to get the average times? SETI WU's do vary a lot, especially with the 3.0 client. My main box (P3-700@1000/256MB CAS 2-2-2) has completed WU's in just under 3 hours and taken as long as 7 hours on others. Playing No One Lives Forever doesn't help either!

Good luck!

Rob

 

[ElFenix]

that p3 450 time is about an hour slower than my p2 400 time. that may be the operating system, but i also have half the ram so there should be some hdd thrashing on my part.

 

[Jalapeno]

Hi guys,

any chance one of you could e-mail me client version 3.0???

Much appreciated!

 

[danzigrules]

you have mail

 

[Assimilator1]

I also think the PIII 450 time is under ,it should be nearer 8.5 hrs

Jalapeno

Hehe ,you need it twice? ,you're being greedy
BTW the CLi v3.0 ,you just run it in the directery you want ,it's not an install thingy!

 

[Poof]

Ditto what Assimilator1 said - I briefly had a PIII 450 running NT 4.0 and it did 3.x WUs in around 8-8.5 hours...

 

[Jalapeno]

OK, how the HECK am I going to get it to the directory I want to run it from?

As soon as my mouse pointer even comes close to the file, it opens.

Somebody please give a "command line" challenged guy a walk-through...

Thanks

 

[Assimilator1]

Create a new folder(directory) where you want it(eg SETI CLi in Program files).Move (right click) the .exe to the new folder.
Create a shortcut(right click) & place it where you want ,eg Desktop.
In the shortcut properties ,shortcut tab add this to the target

"C:\Program Files\SETI CLI\setiathome-3.0.i386-winnt-cmdline.exe" -verbose

Note the 'program files\SETI CLI' is just where I happen to place it.
The verbose command makes SETI give a progress readout

 

[Jalapeno]

Thanks man,

I'll try that as soon as my current WU is done.

 

[nickburns]

well the p450 is used alot of the time but ususally just surfing the net. but now that it hasnt been used that oftern the times have come down to about 9 hours..and since posting this thread,for the last 3 WUs the celeron time has come down to about 9.5 hours???? thats just weird. I have booted my 1 ghz back into wind 98 and it is back to 3.5 hour WUs and sometimes even quicker than that, so i guess it likes win98 better on my box

 

[Jalapeno]

Assimilator,

running in command line now...

 

[Assimilator1]

Jalapeno

Nice ,something else you might want to consider is a WU storing program in case the SETI servers go down.SETIQ is a bit of a DOS command/batch file nightmare ,so I would recommend using SETI Driver instead ,I believe it also has the ability to hide the SETI DOS box.Unfortunatly I don't use it so I am unable to help you, but I'm sure someone will real soon

Nickburns
With v3.0 of SETI WU times vary by a much greater degree.Check out this & this FAQ page.

Here's an excert

Why is there so much variability in workunit completion time with version 3.x?
With version 3.0, we introduced 2 new algorithms to search for pulsed signals. One is a generalized pulse finder and the other is a fast triplet finder. See for description of the algorithms. The generalized pulse finder, while highly optimized, is still very compute intensive. In order to balance sensitivity to pulsed signals (a good thing) with workunit throughput (also a good thing), the client will apply the pulse finder more often in some workunits than others.

The finer the time resolution (ie, the greater the number of bins in the time dimension), the more effective the pulse finding will be. The pulse finder always works with a chunk of data covering 1 telescope beam (0.1 degrees on the sky). For any given frequency resolution, the time resolution of 1 beam is determined entirely by the slew rate of the telescope at the time that the data were acquired. On one hand, the pulse finder does not bother with very coarse time resolutions and on the other, it avoids superfine time resolutions, as the execution time becomes prohibitively large. There is always a tradeoff and we have tried to achieve the optimal balance. Slew rate also affects whether or not the client executes the gaussian finder.

The following chart shows what gets executed at various slew rates. The corresponding angle range is also given. You can see the angle range for your workunit in the workunit header. The horizontal axis is frequency resolution in both FFT length and Hz. For any given slew rate / frequency resolution pair, you can see whether gaussian finding (G), pulse finding (P) or triplet finding (T) is executed.

Sounds like you had some slow WU's

 

[Jalapeno]

Not bad,

I have gone from 8hrs down to 5 1/2 per WU.

 

[BadThad]

I recommend seti driver. It does hide the dos box. It just puts a little dish icon in your system tray. Also, nab seti watch and have driver create the setilog.csv file for it. Then you can keep a close watch on all wu info.