Powerbook hard freeze summary (sort of)

Powerbook hard freeze summary (sort of)

[Bernhard Reiter]

[-- Attachment #1: Type: text/plain, Size: 1063 bytes --]

Okay, we have a couple of people running stable Lombards with
more than 64 MB.

On the other hand, we have people having hard freezes with
Lombard and Pismos even with 64 MB.

So my analysis is as follows: 

High bus speed (400 Mhz) seems to be necessary to trigger this hard
freeze. 
More memory makes it more likely for hard freezes to appear.

If there is a problem with the memory it is more likely that it is with the 
original 64 MB.


Is there somebody out there having a >=400 Mhz Lombard
and Pismo who is _not encountering hard freezes_ under high pressure?
(Preferably with more that 64MB. Because I never got the hard freeze
when I only used 64MB with the mem=64MB switch.)

It might as well be a design problem of the Powerbooks. Can any of the 
hardware kernel hackers comments on the differences in architecture
for powerbooks and the other G3s? Maybe a linux driver is written sloppy
in this area not taking in account how the hardware actually works?

	Bernhard

ps.: Just I reminder: I am not subscribed ot the list.

[-- Attachment #2: Type: application/pgp-signature, Size: 236 bytes --]

Re: Powerbook hard freeze summary (sort of)

[Benjamin Herrenschmidt]

On Wed, May 24, 2000, Bernhard Reiter <bernhard@intevation.de> wrote:

>
>It might as well be a design problem of the Powerbooks. Can any of the
>hardware kernel hackers comments on the differences in architecture
>for powerbooks and the other G3s? Maybe a linux driver is written sloppy
>in this area not taking in account how the hardware actually works?

The Lombard PowerBook uses the same architecture as the first generation
iMacs (and this same memory controller is also used on the beige and B&W
G3s at least). If there is a problem, it's either a memory controller
problem (bogus revision ? incorrect setup ?), or a specific issue with
the design of the memory connector/logic.

The Pismo uses the same chipset as the new iMacs, G4s (sawtooth) and iBook.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Powerbook hard freeze summary (sort of)

[Bernhard Reiter]

[-- Attachment #1: Type: text/plain, Size: 1375 bytes --]

Thanks for the answer.
How can we proceed to further debug this problem?

I have to admit that  am not very experienced with low level hardware 
issues and apple hardware in general.
It seems that there is more than one person having these kind of problems.

	Bernhard

On Wed, May 24, 2000 at 04:49:58PM +0200, Benjamin Herrenschmidt wrote:
> On Wed, May 24, 2000, Bernhard Reiter <bernhard@intevation.de> wrote:
> 
> >
> >It might as well be a design problem of the Powerbooks. Can any of the 
> >hardware kernel hackers comments on the differences in architecture
> >for powerbooks and the other G3s? Maybe a linux driver is written sloppy
> >in this area not taking in account how the hardware actually works?
> 
> The Lombard PowerBook uses the same architecture as the first generation
> iMacs (and this same memory controller is also used on the beige and B&W
> G3s at least). If there is a problem, it's either a memory controller
> problem (bogus revision ? incorrect setup ?), or a specific issue with
> the design of the memory connector/logic.
> 
> The Pismo uses the same chipset as the new iMacs, G4s (sawtooth) and iBook.
> 

-- 
Professional Service around Free Software                (intevation.net)  
The FreeGIS Project				            (freegis.org)
Association for a Free Informational Infrastructure            (ffii.org)

[-- Attachment #2: Type: application/pgp-signature, Size: 236 bytes --]

Re: Powerbook hard freeze summary (sort of)

[Albrecht Dress]

> Is there somebody out there having a >=400 Mhz Lombard
> and Pismo who is _not encountering hard freezes_ under high pressure?
> (Preferably with more that 64MB. Because I never got the hard freeze
> when I only used 64MB with the mem=64MB switch.)

Hi Bernhard,

we have a Lombard, 333 MHz, 320 MB Ram, and two Pismo's, 500 MHz, 384 MB
Ram each, and all run without any problems with LinuxPPC 2000, but with
the kernel upgraded to 2.2.15pre20.  I ran the HINT benchmark on all
machines (which should be some kind of high pressure), and a colleague ran
data reduction software on one of the PISMO's (high preasure, too).
Again, no problems/freezes yet (which, as I hope, will last.. ;-)

Yours, Albrecht.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Powerbook hard freeze summary (sort of)

[Bernhard Reiter]

[-- Attachment #1: Type: text/plain, Size: 1236 bytes --]

Albrecht;
Thanks for the update. The pressure is not easy to create. I can 
create it with compiling a big thing a couple of times. I assume that
you also did that.

It must be hardware then on my part. :-/

	Bernhard


On Thu, May 25, 2000 at 12:22:24AM +0200, Albrecht Dress wrote:
> > Is there somebody out there having a >=400 Mhz Lombard
> > and Pismo who is _not encountering hard freezes_ under high pressure?
> > (Preferably with more that 64MB. Because I never got the hard freeze
> > when I only used 64MB with the mem=64MB switch.)

> we have a Lombard, 333 MHz, 320 MB Ram, and two Pismo's, 500 MHz, 384 MB
> Ram each, and all run without any problems with LinuxPPC 2000, but with
> the kernel upgraded to 2.2.15pre20.  I ran the HINT benchmark on all
> machines (which should be some kind of high pressure), and a colleague ran
> data reduction software on one of the PISMO's (high preasure, too).  
> Again, no problems/freezes yet (which, as I hope, will last.. ;-)
> 
> Yours, Albrecht.

-- 
Professional Service around Free Software                (intevation.net)  
The FreeGIS Project				            (freegis.org)
Association for a Free Informational Infrastructure            (ffii.org)

[-- Attachment #2: Type: application/pgp-signature, Size: 236 bytes --]

Re: Powerbook hard freeze summary (sort of)

[Chris Leishman]

On Wed, May 24, 2000 at 04:18:37PM +0200, Bernhard Reiter wrote:
> Okay, we have a couple of people running stable Lombards with
> more than 64 MB.
>
> On the other hand, we have people having hard freezes with
> Lombard and Pismos even with 64 MB.
>
> So my analysis is as follows:
>
> High bus speed (400 Mhz) seems to be necessary to trigger this hard
> freeze.
> More memory makes it more likely for hard freezes to appear.
>
> If there is a problem with the memory it is more likely that it is with the
> original 64 MB.
>
>
> Is there somebody out there having a >=400 Mhz Lombard
> and Pismo who is _not encountering hard freezes_ under high pressure?
> (Preferably with more that 64MB. Because I never got the hard freeze
> when I only used 64MB with the mem=64MB switch.)
>
> It might as well be a design problem of the Powerbooks. Can any of the
> hardware kernel hackers comments on the differences in architecture
> for powerbooks and the other G3s? Maybe a linux driver is written sloppy
> in this area not taking in account how the hardware actually works?
>
> 	Bernhard
>
> ps.: Just I reminder: I am not subscribed ot the list.


Just to let you know that I'm also running a Pismo with 192 megs ram,
and it seems very stable.  I'm compiled kernels and run tripwire over the
filesystem (lots of IO doing that) and no problems.  I'm running the
linux-pmac-stable kernel.

Chris

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Powerbook hard freeze summary (sort of)

[Seanano]

  As I said before, I'm running a 400mhz lombard with 192 Megs of RAM
without any problems.  I've run multiple compiles at the same time for
upwards of an hour without any problem.  Let's sit back and examine this
for a minute.  So far I've seen maybe 4 reports of this problem.  Yet I
see a lot of peole on this list and the linuxppc-user list have lombards
and pismos matching the description for machines that may have this
problem that never report it.  It would seem that the problem then does
not lie in a "sloppy driver" or something universally wrong with the
pismo/lombard but rather something specific to those machines. at least,
IMHO...

Sean



On Thu, 8 Jun 2000, Chris Leishman wrote:

>
> On Wed, May 24, 2000 at 04:18:37PM +0200, Bernhard Reiter wrote:
> > Okay, we have a couple of people running stable Lombards with
> > more than 64 MB.
> >
> > On the other hand, we have people having hard freezes with
> > Lombard and Pismos even with 64 MB.
> >
> > So my analysis is as follows:
> >
> > High bus speed (400 Mhz) seems to be necessary to trigger this hard
> > freeze.
> > More memory makes it more likely for hard freezes to appear.
> >
> > If there is a problem with the memory it is more likely that it is with the
> > original 64 MB.
> >
> >
> > Is there somebody out there having a >=400 Mhz Lombard
> > and Pismo who is _not encountering hard freezes_ under high pressure?
> > (Preferably with more that 64MB. Because I never got the hard freeze
> > when I only used 64MB with the mem=64MB switch.)
> >
> > It might as well be a design problem of the Powerbooks. Can any of the
> > hardware kernel hackers comments on the differences in architecture
> > for powerbooks and the other G3s? Maybe a linux driver is written sloppy
> > in this area not taking in account how the hardware actually works?
> >
> > 	Bernhard
> >
> > ps.: Just I reminder: I am not subscribed ot the list.
>
>
> Just to let you know that I'm also running a Pismo with 192 megs ram,
> and it seems very stable.  I'm compiled kernels and run tripwire over the
> filesystem (lots of IO doing that) and no problems.  I'm running the
> linux-pmac-stable kernel.
>
> Chris
>
>


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Powerbook hard freeze summary (sort of)

[nailz]

Just thought I'd throw my experiences in the pot.  I have a 500Mhz pismo,
that came with 64 megs.  I slapped in a 128meg SO-DIMM bringing it up to
192.  Linuxppc would crash very reliably (it was rock solid with
64).  Macos would also crash, although not nearly as much.  I assumed it
was a bad DIMM (and it may very well have been) and sent it back.  But now
I'm kind of cagey about buying another DIMM for it.

--
Nolan


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/