Douglas A. Tutty wrote: ... > Here's the software that I need to run on the box (beyond what is in 4.2 > base): > > vim
easy > mc easy > mutt easy > tex considering it dates back to the low two-digit CPU speed days, I suspect "easy", though it might be true that you would appreciate more. > python depends on your app. My VERY limited experience with python is it is a pig. BUT patience is a virtue, you may have sufficient virtue. > some kind of printfilter to serve my Epson LQ-2080 impact printer. probably not a problem. > > Here's the hardware-type I'll envisioning: > > Multiple CPU so that multiple apps can run better on limited individual > CPUs, running under 200 MHz THIS may be a problem. On i386, <200MHz means PPro or Pentium, and I've had near zero luck on SMP PentiumPro machines (not totally zero luck...I have a 4x200MHz 2G RAM which Just Worked...and pumped out something like 700W of heat...) > Probably PCI bus. (please!) > Paralell port for the printer (or I would just use a USB adapter) > USB for future needs > serial port for console > multi-port serial for terminal(s) and my external 3Com Courier modem. > 10 or 10/100 Ethernet > Multiple hard drives: IIRC, the older boxes had 9 GB SCSI drives. I > don't know if one can plunk new eg. 250 GB SCSI drives in them. oooo... 9G. those were BIG. :) But yes, respecting the interface, you can put REALLY big drives in really old machines. But go IDE. An old machine with a SCSI drive does nothing for you. You don't want to buy new 140G or 300G SCSI disks. > SCSI HBA for a tape drive > > > Any suggestions for good old boxes like this that will run modern > OpenBSD and be reasonably reliable? you take what you can get... but yes, get a spare. :) A few thoughts... 1) 1M RAM/1G Disk if you don't want swap on fsck. And, trust me, you don't want swap on fsck if you are in a hurry to get back up and running. This will be a problem. <200MHz machines usually use 72 pin SIMMs, finding bigger than 32M is difficult, and more than four slots is rare. However, if you aren't running X and X apps, you will probably find 64M to 128M A lot. Unless you put a 500G drive on the thing with a 400G /home and trip over the power cord... On the other hand, just because you have a 500G drive doesn't mean you have to partition it to all be available. Or even 10% of it to be available. 2) you may find a PII you can underclock. 3) old machines with heavy metal cases might shield better than new machines with plastic cases. 4) MAKE SURE all shielding is in place. I suspect a lot of people are like me and don't re-install all the silly metal bits that various regulatory agencies desire. Make sure all screws are tightened down properly. 5) Remove CDROMs, floppies, and put in metal plates in their place. Attach them firmly. 6) Is she going to be better with two 200MHz chips than one, say, 300MHz chip? Interesting experiment. (hint: don't refer to her as "lab rat" :) It sounds like she has reactions to both frequency and intensity... Also, I'm not entirely sure about this, but a pair of 200MHz processors may have an RF signal closer to that of a 400MHz processor than you wish. 7) Not sure if it is true anymore, but home machines used to be more strict about RF emissions than "business" machines or servers. 8) You (not necessarily your wife) will be better off with an old, heavy-duty workstation than an old server. You don't want an old server. Trust me (lots of wacko parts). 9) Most non-i386 machines were sold for data centers, probably have higher RF emissions than i386, home-oriented machines. 10) The CPU isn't your only issue: keyboards, disks, printers, almost everything now has its own little processor on it, and is thus a potential source of irritation. 11) Old compaqs sometimes had P-I processors but SDRAM. Some other P-I machines had chipsets with serious problems with more than 64M RAM (which was sad, as many 486 systems could handle many times that much long before...) 12) Most CPU heat sinks are not grounded. Might be interesting to ground it to the case. 13) Old system with PCI slot and add-in modern PCI IDE disk adapter can provide very decent disk performance. How much RF does the adapter emit? 14) Don't obsess on processors and ignore the rest of the machine which is (probably) radiating large amounts of irritation, too. I'd suspect even all NICs aren't equal in this regard. I've got some old P90 machines which would be fascinating tests. All metal case, only large hole, assuming no CDROM drive, is the floppy bay. I did a lot of my early OpenBSD faq work on those machines. If your goals are realistic, a 100MHz machine can do a lot of work for you, but don't trip over the power cord if you put a 500G disk on it. Even that's not a fatal problem -- I got a backup machine at work, does rsync backups to a 500G disk, only 128M of RAM. But, if someone trips over the power cord, no big deal if it takes an hour or more to come back up. For many years, I ran a P166 is my main OpenBSD NFS server. The only time the processor was ever taxed is when I was ripping CDs to MP3 files. I recently retired my 200MHz P-I mail server (it was really a P166, overclocked to 200MHz, for giggles. The extra 33MHz didn't change my life one bit, I can assure you.) I replaced it with a PII-450MHz, as I figure one of these days, I should put more aggressive spam and virus filtering on the thing. Best bit of advice: doctors are not E-Mag specialists, and I can't think of many E-mags I'd want to have giving me medical advice, so don't trust anyone's diagnosis or advice blindly. Keep your brain engaged and do a lot of experimenting. I can speculate as to what is going on as a lot of people here have been doing (and most are obviously neither doctors NOR e-mags), but ultimately, you need to do lots of tests yourself. Good luck... Nick. (I'm not an e-mag, but I did take an e-mag class. One. Long ago. Didn't do very well in it. Oh...e-mag...that's what we used to call the Electromagnetic Engineering students. Curious lot, brilliant people..but wouldn't take a bit of medical or health advice from 'em).
