punkwalrus (punkwalrus) wrote,

  • Mood:
  • Music:

Tech - Seti => Seti@BOINC

It's winter, and that means I can haul out my Seti array. In order to keep costs down, I prefer not to have the array running during the hotter months, because I don't want to pay for the electricity AND cool the guest room. Keeping the guest room air conditioning off when no one is using it saves about $30-40 a month during the summer (it has it's own AC/Heating system independent from the house). I discovered last year that my Seti Array generates enough heat to keep the guest room at about 50 degrees Fahrenheit on most cold days, which is enough to keep the pipes from freezing over, while greatly reducing the energy needs for a normal heater. I think the total draw for the array is the same as a 200 watt bulb, because none of the machines draw more than 25-40 watts at any time. Why? I don't use their graphics cards much, the hard drives aren't used except to store minute Seti data every hour, and the CD-ROMs? Almost never even spin; some machines don't even have them. I even keep the monitor off when not in use. Keeping the guest room heater off and the Seti running saves me money.

This year, I transferred all my machines to BOINC. Seti Classic is being sunsetted, and all of my credits since May 14 have been for naught. BOINC has a lot more flexible stats, a way to chart each computer's progress, and it's open source. But the machines' dormancy in the summer seems to have had an odd toll. Some of the machines are crashing a lot more. I am not sure if it's the BOINC software or something else. On some machines, BOINC just stops working. On others, the whole machine seizes up. As a result, I have some odd computer stats, like P3/800 machines with 128mb RAM doing a lot better than a 1700mhz box with 256mb RAM.

All my machines are Windows or Linux. It upsets me that my lead performer is a P4 2.2ghz Win2000 box. My highest end model, a Red Hat Linux box with a P4 (Hyperthreading) 3.0ghz chip and 1gb of RAM, comes in second and third a lot (currently it is 3rd behind a 1.8ghz/512mb RAM WinXP Home box... gaaaah!). This offends my Geek Hot Rod sensibilities. I also have boxes that are essential clones of one another, and the stats they generate vary wildly. Christine has been giving me some of her old machines from her work, and all of them are the same make and model of Dell. I suspect some of the issues may have to due with how they were used until I got them, which means that computers "wear out" more than I realize. I also suspect having an X server running is a real deep CPU drain, because I have a 2.4ghz box that slows down considerably (stats-wise) when I am on it, doing something in KDE (like I am doing right now).

Maybe FreeBSD is worth a try. My OpenBSD systems are not worth it, the highest one is a Pentium 133, and OpenBSD is optimized for security (as it should be), not speed.

I also thought I had cleared all my Seti@Home Classic machines. Either they were turned off for good, or they had been upgraded to BOINC. Yet, my Seti@Home Classic stats are still creeping upwards, which means I have a box somewhere, and it has to be at work, where Seti@Home Classic is still chugging away... it's probably on some old Wardialer or Wardialer server, although I thought I had all of those shut down (they certainly aren't responding to pings).

Sometimes people ask me why I bother. "The chances of you finding intelligent life," they say, "are remote against the odds." Yeah, but at least I can help by weeding out the misses while someone can focus on the hits. I have always wanted to do this since I was a teen.
  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded