I already replied to some of this in another post, so I'm only going
to touch the unaddressed issues...

>Pro at 150MHz.  My understanding is that the P6 does not handle 16 bit
>code very well, but I would guess that NetBSD is all 32-bit code.  Is

Yes, NetBSD is entirely 32-bit code (unless you're running on an
Alpha, when it's 64-bit :-).  There is no 16-bit code that I'm aware
of anywhere in the distributed NetBSD kernels or binaries.

>don't know just how much better the PPro is with 32-bit code.  Would it
>make up for the 25% speed difference?)

It's possible a 200Mhz Pentium and a 150MHz PPro would be about the
same speed.

>I would likely go with an ASUS
>motherboard for either one, the 430HX-based T2P4 for the Pentium, or the
>440FX-based P6NP5 for the Pentium Pro (if that makes any difference). 

Asus motherboards are quite good.  I'm running a P6NP5 and a P55TP4N
(Triton-1).  Both work flawlessly.

>I've heard that the Pentium motherboard works fairly well overclocked,
>so I have considered doing as high as 250MHz on the Pentium board (if my

I seriously doubt you could get the chip to even boot at 250MHz.  20%
is generally about the max a decent chip will tolerate, from many of
the tests I've tried.  You might be able to get 233MHz out of it
(assuming it knows how to interpret the pins for 233MHz, which were
not available in any previous P5s, if I'm not mistaken -- i. e. it may
be that only the 233MHz Pentium knows how to interpret an old speed
selection as 233MHz, instead of some slower speed).

Also, avoid running the PCI bus at a multiple of anything slower than
33MHz, and/or the memory bus at a multiple of anything slower than
66MHz.  Doing so will slow down your entire system, not just the CPU.

>PCI cards would tolerate it).  Likewise if the P6-150 is stable at 166
>or 180, I'd be likely to push for the higher speed.  I realize that

166MHz would be the way to go.  See above re: 180MHz.  It's possible
your chip would actually give you slower effective performance at
180MHz, than 166, because the bus would be running 10% slower.

>especially at the overclocked speeds the P5 motherboard's bus frequency
>would be considerably higher than the P6 motherboard, so memory accesses
>and such would be faster regardless of the actual processor clock speed.

Assuming your memory could handle it.  It's possible you'll have to
insert extra wait states to make up for the extra speed, which could
actually give you the old performance, or worse.  Just so you know...

>    BTW, I assume there would be no point in considering a MMX Pentium
>for NetBSD, correct?  Are there any performance enhancements to the
>general processor, or is the presence of the MMX functions the only
>difference?

There is no MMX-specific software that I know of yet for the MMX
chips.  However, the MMX Pentiums do have a larger cache that the
non-MMX Pentiums (32K vs. 16K, if I'm not mistaken).

-----------------------------------------------------------------------------
  Michael L. VanLoon                           michaelv@MindBender.serv.net
        --<  Free your mind and your machine -- NetBSD free un*x  >--
    NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3,
        Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32...
    NetBSD ports in progress: PICA, others...
-----------------------------------------------------------------------------