[4519] in linux-net channel archive
Re: seeing crashes? read this... YUP... on intel also
daemon@ATHENA.MIT.EDU (babydr)
Tue Sep 24 09:57:32 1996
Date: Mon, 23 Sep 1996 21:48:35 -0700 (PDT)
From: babydr <babydr@nwrain.net>
To: Ken Jordan <kenjordan@massmedia.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>,
submit-linux-dev-net@ratatosk.yggdrasil.com
In-Reply-To: <Pine.LNX.3.94.960921133550.32576A-100000@netserver.massmedia.com>
Ken & Alan, having recompiled with cpu defined as (486)
has made no difference the 'divide error: 0000' still persists
at the location (from /System.map)
0010ec3c t do_fast_gettimeoffset
I'd be greatly appreciative if someone could point me to where
in the code I can disable the use of 'do_fast_gettimeoffset'
the slower speed for stability tradeoff is becoming more and
more beautiful...
Tnx, JimL
On Sat, 21 Sep 1996, Ken Jordan wrote:
> Date: Sat, 21 Sep 1996 13:49:20 -0700 (PDT)
> From: Ken Jordan <kenjordan@massmedia.com>
> To: Alan Cox <alan@lxorguk.ukuu.org.uk>
> Cc: submit-linux-dev-net@ratatosk.yggdrasil.com
> Subject: Re: seeing crashes? read this... YUP... on intel also
>
>
>
> On 21 Sep 1996, Alan Cox wrote:
>
> > In article <Pine.VUL.3.94.960919194933.9092A-100000@pppbox.baby-dragons.com>,
> > babydr <babydr@nwrain.net> wrote:
> > AMD5x86-90, 64MB memory,
> > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >
> > This odd divide by 0 error is always only seen on the AMD5x86 chips. You
> > need to disable the use of the pentium fast gettimeofday for these
> > clone processors it seems.
> >
> > Alan
>
> Umm, don't confuse the AMD5x86-P90 (which is really a clock quadrupled
> 486) with the AMD-K5 (which is AMDs pentium class clone). AMD was kind of
> boneheaded when they named this chip and it creates a lot of confusion (no
> doubt a marketing decision).
>
> I've never heard of anyone having problems with the AMD5x86-P90 (unless
> you try to use this 486 class chip with the compile for pentium option or
> run it overclocked without appropriate precautions and testing). I have
> read the reports of the K5 having unexplained divide by zero errors.
>
> I've been using several AMD5x86 chips without problems (it runs faster
> than a P-90 when run at 160Mhz - one of the best bangs for the buck at
> around $120 for MB and CPU).
>
> Take it easy,
> Ken Jordan
>
>
>
