WarpUp (431/442)

From:ThOmAs!
Date:30 Dec 99 at 12:21:07
Subject:Re: [warpup] Re: other WarpOS problems

On Fri, 17 Dec 1999, Sam Jordan wrote:

Hi!

I just thought I'd share my experience with my PPC 604e-233 +040/40MHz.
First of all I don't trust P5 that much initially they gave me a non
working board then they gave me a 040 board clocked at 60MHz(!) before
they managed to give me one that worked.

I have an A4000T with 7 Zorro cards, CDR, SyQuest and a few harddrives
aswell as to much external stuff so it get's pretty hot.
In the beginning I had problems partly because the versions supplied with
the cards were not working properly and as I suspect because of
overheating.
I have now bought a dual fan to put in the 5 1/4" bay aswell as a fan
monitoring which also eat up another 5 1/4" bay and controls two internal
fans and supply me with temperature readout. I do also have two parallel
PSUs which actually don't give away much extra heat.
Nevertheless after adding the extra fans I have not had a single
unexpected crash in about a year. I have machine on 7/24 and last time I
turned it off was in August 1998 because I was away for Rhodes for 14
days. Had it not been that I needed to make a reboot for upgrading from
PFS2-> PFS3 I would have still had much better uptime than our Sun
Enterprise E4500 at work ;-)

I do not use the fix found on Internet.

Just make sure not cables are blocking the fan on the PPC card, and I
found the 040 to get mighty hot aswell even though it does not even have a
heatsink(!)

I'm sorry to quote the whole post, but as this positng was pretty old
(sorry been ill!) it's easier to see what were initially suggested.

Hope this helps. BTW: These fans were not that expensive, it can easily be
bought in Norway and Sweden thru "Claes Ohlson".

> Am 17-Dez-99 schrieb Tobias Seiler:
>
> >> I bet you have a 604e/233. It seems to be the infamous LR-bug,
> >> which always trashes bit 0 or bit 2 of your LR. This failure occurs
> >> due to an overheated PowerPC. In most cases replacing the fan by
> >> a strong 12V Pentium fan might help.
> >>
> >> In the meantime you can try aminet/hard/misc/CSPPC233Fix.lha as a
> >> quick fix.
> >>
> >>
> >> --
> >> _ Frank Wille (frank@phoenix.owl.de)
> >> _ // http://home.owl.de/~frank/
> >> \X/ Phx @ #AmigaGer
>
> >Well, I had idea about this LR bug. (actually a friend of mine had it)
>
> >As this problem only occurs with 233MHz cards and as I understood it these
> >cards where all delivered with 060 processors at 60MHz our thoughts where
> >if this could be an overclocking problem to the motherboard.
>
> >For a 200MHz card the divider for the system clock would be 4 to get 50MHz.
>
> I would rather say, that 200Mhz cards have a 66Mhz bus clock with a
> multiplier of 3. There exist other boards (180Mhz, 150Mhz) which all
> have a multiplier of 3.
>
> BTW, 'showinfo' in the WarpUp directory reports the bus clock.
> For my card (180Mhz) it's 60Mhz.
>
> >I dont really think that p5 has designed the completele thing in a asyncron
> >way.
> >So when the divider remains at 4, 233/4 = >50. So what happens to the system
> >speed then ?
>
> >Maybe I'm wrong. It's just an idea.
>
> You're most likely right in the assumption, that Phase5 used boards
> for 233Mhz-CPU's which were not designed to deal with such clock
> frequencies. They simply overclocked the boards IMHO and as it seems
> they did never say anything about it, so people had to find out
> themselves, that they need a much stronger cooler.
>
> But I'm actually not sure whether the problem is the bus frequency,
> maybe it's still 66Mhz (someone with 233er can confirm or correct
> this value). I would rather guess that the increased power
> consumption/heating due to the increased core clock is the real problem.
> Overheating is the problem Nr. 1 for AMIGA hardware nowadays.
>
> bye
> --
> Sam Jordan ______
> eMail: s.jordan@haage-partner.com _______ / \
> sam_jordan@spin.ch (private) ()_______) / \
> | (__ / NCC \
> Also at HAAGE & PARTNER PowerPC | __| 1701-D |
> development section __|_(__ \ /
> ()_______) \ /
> We develop to WarpUP the AMIGA! \______/
>

Virtually Yours,

.ThOm@s!



Thomas Hansen, Department of Informatics, University of Oslo, Norway
"Minus infinity: The smallest [number that can be represented in a
particular type of variable], but not necessarily or even usually the
simple negation of plus infinity."
(Eric Raymond, The new Hacker's Dictionary, 1991)