[KLUG Members] Re: Well, that does it. -- Dual-Athlon is calling ...

[Adam Bultman]

Well, I put in a new mobo and proc (Gigabyte, XP 1800+), with the same
everything else, and it loaded beautifully.  Fast, too.

If you haven't, I still urge you all to try beehive linux.
-- 
Adam Bultman
adam@glaven.org
[ http://www.glaven.org ]


On 6 Jun 2002, Bryan J. Smith wrote:

> [ Moving to HARDWARE ]
>
> On Thu, 2002-06-06 at 00:02, Adam Bultman wrote:
> > Oh, I'm way ahead of you.  I've tried infinite matchings of:
> > IDE CDROM drive.
> > ATA hard drive
>
> Good.
>
> > PCI Video card
>
> Good.
>
> > 256 MB stick of RAM
> > 64 MB stick of ram
>
> Oh yeah, I meant to mention this.  Only _registered_ (4-bit IC widths)
> 256MB DIMM modules are supported on the i440BX chipset.  Most 256MB
> DIMMs sold today are NOT of that type -- the i815/845 take a different
> 256MB DIMM type (8 or 16-bit ICs, non-registered).
>
> The 64MB DIMM would remove that and you only need 1 since the chipset is
> i440BX.
>
> > 1 proc
> > dual proc
>
> Good.
>
> > different procs in different slots, one at a time, two at a time, one w/
> > the filler card, one w/o fille rcard, etc.
> > ...
> > Different SCSI settings on the drives.
> > I've done about a million things, and spent the last 5 days trying to get
> > the correct settings going.  I don't know if you want to call it
> > 'sticktoitivness' or 'stupidity'.
>
> No, call it "thoroughness."  You could have missed something, but it's
> pretty much something you want to forget about for awhile.  It may be
> that you really exhausted all avenues, or that something else will come
> to you in a few days.  Either way, you've got the right idea in moving
> past it.
>
> > The only thing I haven't tried is different processors. Howver, I know
> > those two procs are good (I ran them on another box) so they mustb e OK.
> > At this point, I'm going to hang it on my wall so I am reminded never to
> > trust Ebay, and 'r8r' especially.
>
> Well, the mainboard's VRM _might_ not be the latest as the processors,
> but I _doubt_ it.  A 400MHz Pentium II is VRM8.2 if I remember
> correctly, and _all_ i440BX chipset mainboards _at_least_ do VRM8.2, if
> not newer.
>
> > IF I come across a single proc mobo that'll hold one/two of those procs,
> > or I have a friend with a dead proc, I might use/give it away. Until then,
> > on the wall it stays.
>
> I have a ViA 693A/598B (north/south) Slot-1 PC133 mainboard I can send
> you.  The thing sucks with a PPGA-to-Slot-1 converter, but a Pentium II
> 266MHz ran fine in it.  It does 66/100 on the CPU, and 66/100/133 on the
> memory.  I'm in Orlando though, but I'm willing to ship for a small fee
> -- it's just sitting in my room.
>
> > Anyway, I'm ready to shoot it, but I'm holding out, and I'm gunna go and
> > buy me a fast motherboard and processor combo, with some DDR dimms, so I
> > have a bloody fast linux machine running beehive linux.  Period.  And I
> > will laugh. On into the night.
>
> Tyan Tiger MP (S2460) dual-Athlon dude -- only about $160 nowdays for
> the mainboard.  PC2100 (266MHz DDR) memory, registered ECC recommended.
> You'll need to get either older, non-XP Athlons, Durons or Athlon MPs as
> the Athlon XP is "locked out" on the package from dual-processor.  I'm
> running dual-Athlon 1.4GHz (last speed before XP) -- no issues.  Unlike
> the Tyan Thunder K7, which has a special PS, you just need a regular PS
> -- but I recommend the Enermax VP465VE (435W) as my Antec PP412X (400W)
> didn't cut it (although the new Antec TruePower 430 should).
>
> 64-bit PCI, CPUs can operate independently thanx to the Alpha EV6 bus so
> one can be accessing memory while the other is doing I/O.  On servers,
> equals Intel's dual-memory channels in most operations, kicks Intel P4
> butt on workstations -- even 1.26GHz Intel P3 is better than all but the
> latest 2GHz+ P4s.
>
> BTW, I recommend the Tyan Tiger MP (S2460) _over_ the Tyan Tiger MPX
> (S2466) _unless_all_ of your 64-bit PCI cards _are_66MHz_.  Otherwise, a
> 33MHz in one of the two 64-bit PCI cards on the MPX will not only result
> in slowing the 64-bit PCI cards to 33MHz anyway (like the MP, only the
> MP has four 64-bit PCI cards instead of just two) but also slows your
> north-to-southbridge link which then has four more PCI cards (can create
> a bottleneck with the ATA -- long story).  It's also about $40 cheaper
> (for the "cost" of no on-board 10/100 NIC).  See this post for more on
> the MP (762+766) v. MPX (762+768):
> http://www.matrixlist.com/pipermail/pc_support/2001-December/000450.html
>
> > However, I still thank you all for the help with this very frustrating
> > motherboard/proc issue.  If you wanna take a whack at it, be my guest, but
> > I don't take any responsibility if you shoot yourself as a result.
>
> No problem.  Sorry if some of my commentary/posts were "redundant" or
> just "rehashes" of what you tried.  You did more "due diligence" than I
> probably would have before I got frustrated.  I surrender to you man.
>
> -- Bryan
>
> P.S.  Dual-Opteron (x86-64 Dual-ClawHammer and Quad-to-8-way
> SledgeHammer) is not due until 2003H1 -- so figure April/June of 2003.
> This is inspite of the fact that the Athlon "64" (x86-64 uniprocessor
> ClawHammer -- still "unnamed," but I suspect "64" will be it) will be
> out 2002Q4.  So dual-Athlon is not a bad move for the next 9-12 months.
>
>