Hi!
Just downloaded again, after a long time, the setiathome client. I
wanted to look how smooth my tyan dual works. So i started the client
and after a few seconds it gets and `floating point exception'. No
problem till now, cause it seems to be seti bug. Ok.
Right after that my window manager segfaults. Ok, switch to console,
restart it and go. No! Can't start any programs anymore, no login. All
tasks die one after the other, up to the complete lock of the machine.
Even alt-sysrq doesn't work.
So, this is kernel 2.4.17 and i'll try other kernels right after this
email.
Anyone knows what's going on?
Christian Thalinger wrote:
>
> Hi!
>
> Just downloaded again, after a long time, the setiathome client. I
> wanted to look how smooth my tyan dual works. So i started the client
> and after a few seconds it gets and `floating point exception'. No
> problem till now, cause it seems to be seti bug. Ok.
>
> Right after that my window manager segfaults. Ok, switch to console,
> restart it and go. No! Can't start any programs anymore, no login. All
> tasks die one after the other, up to the complete lock of the machine.
> Even alt-sysrq doesn't work.
>
> So, this is kernel 2.4.17 and i'll try other kernels right after this
> email.
>
> Anyone knows what's going on?
What CPU do you have? Do you have the FPU emulator compiled in? Are
there any oops messages?
--
Brian Gerst
On Wed, 2002-01-16 at 00:28, Brian Gerst wrote:
> What CPU do you have? Do you have the FPU emulator compiled in? Are
> there any oops messages?
>
> --
> Brian Gerst
>
I mentioned in my first mail the dual tyan, so athlon xp, no fpu
emulator ;-) and no oops messages.
On 16 Jan 2002, Christian Thalinger wrote:
> I mentioned in my first mail the dual tyan, so athlon xp, no fpu
> emulator ;-) and no oops messages.
Dual Athlon XP problem. Thanks for playing.
--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
On Wed, 16 Jan 2002 12:58:35 +0100 (CET)
Dave Jones <[email protected]> wrote:
> On 16 Jan 2002, Christian Thalinger wrote:
>
> > I mentioned in my first mail the dual tyan, so athlon xp, no fpu
> > emulator ;-) and no oops messages.
>
> Dual Athlon XP problem. Thanks for playing.
Interesting. That's the first actual report I've seen of problems caused by
using XPs instead of MPs. I'd been wondering if I could get away with XPs for
my next SMP box; now I know better ;)
Christian Thalinger wrote:
>
> On Wed, 2002-01-16 at 00:28, Brian Gerst wrote:
> > What CPU do you have? Do you have the FPU emulator compiled in? Are
> > there any oops messages?
> >
> > --
> > Brian Gerst
> >
>
> I mentioned in my first mail the dual tyan, so athlon xp, no fpu
> emulator ;-) and no oops messages.
Last I checked, Athlon XP's weren't certified for SMP, only MP's.
That's likely what the problem is. And for the record, Tyan also makes
Intel boards too.
Processor manufacturing 101: All processors of a given family come off
the same production line. Due to variations in the process, some
processors have defects that only show up at higher clock speeds, SMP
mode, etc. At the end of the line the processor is tested. If it fails
at higher clock speeds it is marked at a lower speed. If it fails SMP
it is marked as an XP. Market demand can also cause a chip to be rated
lower than it really is, so you can sometimes get away with
overclocking, etc. but it's just random luck if it really works.
--
Brian Gerst
On Wed, 16 Jan 2002, Brian Gerst wrote:
> Last I checked, Athlon XP's weren't certified for SMP, only MP's.
> That's likely what the problem is. And for the record, Tyan also
> makes Intel boards too.
>
> Processor manufacturing 101: All processors of a given family come
> off the same production line. Due to variations in the process, some
> processors have defects that only show up at higher clock speeds, SMP
> mode, etc. At the end of the line the processor is tested. If it
> fails at higher clock speeds it is marked at a lower speed. If it
> fails SMP it is marked as an XP. Market demand can also cause a chip
> to be rated lower than it really is, so you can sometimes get away
> with overclocking, etc. but it's just random luck if it really works.
Could you be more specific on this "random luck" bit? Let's say we have
a production line making processors that should *all* run SMP at, say,
1800 MHz. What fraction of them will actually run SMP and 1800? What
fraction of them will actually run at 1800 UP? What fraction of them
will run at 1700 SMP, 1600 SMP, etc.? And what fraction of them run at
1800 SMP at the end of the line but croak when they get stuck in
<ducking> Aunt Tillie's motherboard?
I'm not looking for anyone's proprietary yield statistics here -- just a
rough idea of what kind of distributions we're dealing with here. For my
application, the 1.3 GHz Athlon I've got now is overkill. I wanted a
dual when I got the system last March, but there weren't any
motherboards available that I could find.
--
M. Edward Borasky
[email protected]
The COUGAR Project
http://www.borasky-research.com/Cougar.htm
Q. How do you tell when a pineapple is ready to eat?
A. It picks up its knife and fork.
On Wed, 2002-01-16 at 14:14, Bruce Harada wrote:
> On Wed, 16 Jan 2002 12:58:35 +0100 (CET)
> Dave Jones <[email protected]> wrote:
>
> > On 16 Jan 2002, Christian Thalinger wrote:
> >
> > > I mentioned in my first mail the dual tyan, so athlon xp, no fpu
> > > emulator ;-) and no oops messages.
> >
> > Dual Athlon XP problem. Thanks for playing.
>
> Interesting. That's the first actual report I've seen of problems caused by
> using XPs instead of MPs. I'd been wondering if I could get away with XPs for
> my next SMP box; now I know better ;)
Don't be too scared. Everything works except this seti thingy.
In article <[email protected]>,
Christian Thalinger <[email protected]> wrote:
| On Wed, 2002-01-16 at 14:14, Bruce Harada wrote:
| > On Wed, 16 Jan 2002 12:58:35 +0100 (CET)
| > Dave Jones <[email protected]> wrote:
| >
| > > On 16 Jan 2002, Christian Thalinger wrote:
| > >
| > > > I mentioned in my first mail the dual tyan, so athlon xp, no fpu
| > > > emulator ;-) and no oops messages.
| > >
| > > Dual Athlon XP problem. Thanks for playing.
| >
| > Interesting. That's the first actual report I've seen of problems caused by
| > using XPs instead of MPs. I'd been wondering if I could get away with XPs for
| > my next SMP box; now I know better ;)
|
| Don't be too scared. Everything works except this seti thingy.
Does this run correctly for UP? And is this the right version of
setiathome for this CPU. Not using SSE7, 4Dthen, or some other
proprietary FP method? And until it is proven to work with the MP part,
should it ever actually be shipped instead of advertized, I wouldn't be
totally sure about XP or kernel being at fault, or even RAM problems
under load, etc.
--
bill davidsen <[email protected]>
CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.