Another VLAR question...

[OhioDude]

All my systems run CLI and only one is a Win98 machine. All the rest are Win2k, NT4, and one SCO UNIX. The Win98 box is a PIII 500 and so far it has been sent two wu's by SetiQueue and both have been VLAR wu's (.017 and .013). Each required over 18 hours to process. So it doesn't look like SetiQueue differentiates between OS when routing VLAR's, just GUI or CLI.

I have the Win98 box (PIII 500) and a PIII 700 Win2k Server box. Both use an AOpen AX6BC motherboard with a 100 MHz FSB and generic CAS3 RAM. In the last 24 hours, each has processed a wu with an AR of .013. The Win98 box required 18:06 to process its wu. The Win2k box required 9:12 to process its wu. :Q Naturally, the 700 should outperform the 500, but not by nearly 100% when you consider only physical configurations. It looks like OS has a ton to do with processing times when it comes to VLAR's.

With all this in mind would it be a good recommendation to run the GUI on the Win98 box, or at least one system in the herd?

BTW - SetiQueue is one slick system. If you're a stats-aholic, SetiQueue has got just the fix you need.

 

[artemedes]

Well I know from the two systems that I have running GUIs, that the normal processing by the clis is considerable faster and the few number of WUs that are VLAR, I am almost positive that I get more work done by using the CLIs. All of my machines are win98se. As example, I have 2 identical gateway 1 ghz athlons at work. The cli will averages close to 7 hr, wheras the one running the gui almost never gets below 8 hrs.
The situation gets worse on slower computers. I have two other identical machines at work, both pII400. The CLI will average 11-13 hours depending on use, and the gui never gets below 15 hrs, and averages more than 16 hrs.


So, all in all despite the problem, I still get more work done using the CLIs.


Re: Stats - Did you know that if you have your setiq set up to send stats to the maker of the program, that we are also ranked among setiq users. So if you were to point your setiQ at orangekids setiQ, we would move up the Setiq rankings faster. Check them out here.

To point your Q to the team's, just go to settings -> Queue settings, and under connection type make sure it says "WinInet proxy", and the address is orangekid.teamanadtech.com:5001



Help your fellow team members kick arse in one more stat! Please!

 

[Poof]

OhioDude - does your *nix box run SETI and if so, does it have access to your Setiqueue? If it does, you could have it set so that the VLARs route to that rather than to any of the windoze boxes.

You would need to Edit the settings for your 2 windoze clients (under the "Clients" setting) to enable VLAR routing. If your queue gets results from your SCO box, then it should route VLARs away from anything that is the CLI, to the SCO.

 

[OhioDude]

artemedes

I see your point. When looking at the overall picture, it appears to be far better to run CLI on all systems as opposed to worrying about having a GUI to process the VLAR wu's.

It would certainly seem like a better VLAR routing solution for SetiQ to be able to route by OS. I think I'll be upgrading that Win98 box to Win2k...

SetiQ:
I changed my queue server to orangekid.teamanandtech.com, port 5001. I had to use Winsock Proxy as my connection type. When I tried WinInet Proxy, SetiQ would send only send one wu and then hang and wouldn't retrieve any new wu's.

 

[OhioDude]

poof
I see from looking at my stats that the SCO system doesn't seem to be affected by the AR as much as the other OSes. Unfortunately, my SCO box is a pig, even though it is a PIII 600. Something about the SCO/Intel combo. It's averaging about 13 hours per wu. I've got a couple of Celeron 400's running Win2k that nearly beat it in average wu time and a PII 400 with Win2k that does...

Perhaps some kernel tuning would help, I don't know.