Support for new Dual G4 1 Gig machines?

Support for new Dual G4 1 Gig machines?

[Kevin B. Hendricks]

Hi,

I have not been keeping up wih the latest processor and driver support in
the kernels or on the status and stability of dual processor boxes.

But I just ordered a brand new Dual 1 GHZ G4 machine from Apple with an
ATI 7500 video card (I also have an ATI Radeon Mac PCI card).

So can anyone update date me if the version of the G4 in the latest Apple
machines is supported?

How stable is SMP support under PPC Linux?

I know I can get the ATI Radeon 32M Mac edition card to work to work with
my FlatPanel under ppc linux, but I have no idea about the ATI 7500 and
the FlatPanel display?

Any information on required patches or kernel version to make this thing
work when it comes would be greatly appreciated.  I will have the use of
my old machine to rebuiold kernels and things for about a week after the
new machine arrives (before it become's my wife's new machine) and would
like to get everything I can built in advance.

Thanks,

Kevin

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

Re: Support for new Dual G4 1 Gig machines?

[Christopher Murtagh]

On Mon, 28 Jan 2002, Jim Potter wrote:
>I've been running SMP YDL on a dual 500 for over a year;  it's quite
>stable.

 I've been meaning to pass this on to the linuxppc-dev list for some time,
but I've been *insanely* busy. Sorry for the cross post if it ticks anyone
off.

 Actually, this depends on what you are doing with the machine. My desktop
machine is a Dual-450 and it has been running very well. However, I had a
Dual-800 and a Dual-500 that I tried to put into production, and they both
crashed often. I had mentioned the 800 to BenH and he suspected a problem
with high-memory and potentially a 7450 problem. After a few 7450 fixes
were merged, I tried the most recent kernel (2.4.18pre1), and it made the
machine much more stable.

 I think that the problem is also potentially network related, as when the
Dual-800 was in my office chugging away at internal stuff (low network
access), it ran without any problems. It even handled a serious stress
test (load average above 9.8 for over 2 hours). However, as soon as I put
it into production and it started taking on moderate traffic, it barfed
several times before I re-compiled the kernel and turned of SMP.

 All of the above machines are running 2.4.18pre1 running Apache 1.3.20,
PHP 4.1.1, Postgres 7.1.3, htdig 3.1.16. Both the 500 and the 800 have 1GB
of RAM, but the 450 only has 512, so perhaps it is a high-memory with
networking problem?

 So, SMP on the PPC platform has some work needed I'm afraid. I'd be happy
to help in any testing if someone wants me to, as I'd love to actually use
that other CPU on my dual 800. :-)

 (Sorry for not answering you directly Eric, I've been swamped with work.)

Cheers,

Chris

--

Christopher Murtagh
Webmaster / Sysadmin
Web Communications Group
McGill University
Montreal, Quebec
Canada

Tel.: (514) 398-3122
Fax:  (514) 398-2017


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

Re: Support for new Dual G4 1 Gig machines?

[benh]

> .../...
>
> All of the above machines are running 2.4.18pre1 running Apache 1.3.20,
>PHP 4.1.1, Postgres 7.1.3, htdig 3.1.16. Both the 500 and the 800 have 1GB
>of RAM, but the 450 only has 512, so perhaps it is a high-memory with
>networking problem?
>
> So, SMP on the PPC platform has some work needed I'm afraid. I'd be happy
>to help in any testing if someone wants me to, as I'd love to actually use
>that other CPU on my dual 800. :-)

I recently spotted a problem with >512Mb (CONFIG_HIGHMEM or not) & SMP that
is fixed in the current bk _2_4 (2.4.18pre7). I also sent a fix to Marcelo
that should hopefully be in the next 2.4.18preX release.

You should also use sungem preferably to the old gmac driver.

Ben.


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

Re: Support for new Dual G4 1 Gig machines?

[Christopher Murtagh]

On Tue, 29 Jan 2002 benh@kernel.crashing.org wrote:
>I recently spotted a problem with >512Mb (CONFIG_HIGHMEM or not) & SMP
>that is fixed in the current bk _2_4 (2.4.18pre7). I also sent a fix to
>Marcelo that should hopefully be in the next 2.4.18preX release.

 Cool, thanks Ben! I'll give it a try soon.

>You should also use sungem preferably to the old gmac driver.

 Do you suspect that this is what is causing my crashes when the box is
exposed to moderately high network volume?

 Thanks again.

Cheers,

Chris


--

Christopher Murtagh
Webmaster / Sysadmin
Web Communications Group
McGill University
Montreal, Quebec
Canada

Tel.: (514) 398-3122
Fax:  (514) 398-2017


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

Re: Support for new Dual G4 1 Gig machines?

[Holger Bettag]

Apparently the GHz Macs use a new CPU variant, the MPC7455. It differs
very slightly from the earlier CPUs in that it has

 - a feature to lock parts of the L1 data cache (can probably be ignored)

 - some extra SPRGs (can probably be ignored)

 - 4 additional DBATs and 4 additional IBATs. I am not sure if those
   require initialization at early startup to not interfere with later
   operation.

The real difference is in fabrication; it is the first SOI part.

Some official info here (beware of line wrap):

http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC7455&nodeId=03M943030450467M986

  Holger

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

Re: Support for new Dual G4 1 Gig machines?

[Edward Swarthout]

Holger wrote:
> 4 additional DBATs and 4 additional IBATs. I am not sure if those
> require initialization at early startup to not interfere with later
> operation.

They are not used until enabled in HID0.

See HID0[HIGH_BAT_EN] on page 49 of
  http://e-www.motorola.com/brdata/PDFDB/docs/MPC7450TS.pdf

An debug hint for early users:

The 7455 uses bit 8 for HIGH_BAT_EN.  On the 7400/7410, bit 8 is DOZE
Enable and on the 7450 it is ignored.  So, code ported from the 7400
which happens to set bit 8 will work fine on the 7450 but cause great
grief on the 7455.  Random translations whose side effects don't get
detected until much later tend to make system debug difficult...

Ed.Swarthout@Motorola.com

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

Re: Support for new Dual G4 1 Gig machines?

[Derrik Pates]

On Tue, 29 Jan 2002 benh@kernel.crashing.org wrote:

> I recently spotted a problem with >512Mb (CONFIG_HIGHMEM or not) & SMP that
> is fixed in the current bk _2_4 (2.4.18pre7). I also sent a fix to Marcelo
> that should hopefully be in the next 2.4.18preX release.
>
> You should also use sungem preferably to the old gmac driver.

Maybe you should officially deprecate the GMAC driver in favor of SunGEM,
then? Put something on its line in the config scripts that says
***DEPRECATED***, and something in Configure.help for it that says
something like "Please DO NOT use me, use the SunGEM driver instead!" so
there's no question, so maybe someday it can be dropped?

Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN


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

Re: Support for new Dual G4 1 Gig machines?

[benh]

>
>>You should also use sungem preferably to the old gmac driver.
>
> Do you suspect that this is what is causing my crashes when the box is
>exposed to moderately high network volume?

Maybe, only tests will prove me right or wrong.

Ben.


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

Re: Support for new Dual G4 1 Gig machines?

[benh]

>The 7455 uses bit 8 for HIGH_BAT_EN.  On the 7400/7410, bit 8 is DOZE
>Enable and on the 7450 it is ignored.  So, code ported from the 7400
>which happens to set bit 8 will work fine on the 7450 but cause great
>grief on the 7455.  Random translations whose side effects don't get
>detected until much later tend to make system debug difficult...

If I didn't fuck it up too much, the latest kernels will not set
the DOZE bit on those CPUs if they have a proper cputable entry
(basically copy the 7450 one).

Ben.


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

Re: Support for new Dual G4 1 Gig machines?

[Tom Rini]

On Tue, Jan 29, 2002 at 10:25:45AM -0700, Derrik Pates wrote:
>
> On Tue, 29 Jan 2002 benh@kernel.crashing.org wrote:
>
> > I recently spotted a problem with >512Mb (CONFIG_HIGHMEM or not) & SMP that
> > is fixed in the current bk _2_4 (2.4.18pre7). I also sent a fix to Marcelo
> > that should hopefully be in the next 2.4.18preX release.
> >
> > You should also use sungem preferably to the old gmac driver.
>
> Maybe you should officially deprecate the GMAC driver in favor of SunGEM,
> then? Put something on its line in the config scripts that says
> ***DEPRECATED***, and something in Configure.help for it that says
> something like "Please DO NOT use me, use the SunGEM driver instead!" so
> there's no question, so maybe someday it can be dropped?

Well, I think the plan is to simply rm the file in 2.5.x :)  Dropping it
from 2.4.x is probably a bad idea, since it's not broken for old things.
The defconfigs should probably be updated tho. :)

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

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

Re: Support for new Dual G4 1 Gig machines?

[Eric Leblond]

le lun 28-01-2002 à 21:34, Christopher Murtagh a écrit :
> On Mon, 28 Jan 2002, Jim Potter wrote:
> >I've been running SMP YDL on a dual 500 for over a year;  it's quite
>  Actually, this depends on what you are doing with the machine. My desktop
> machine is a Dual-450 and it has been running very well. However, I had a
> Dual-800 and a Dual-500 that I tried to put into production, and they both
> crashed often. I had mentioned the 800 to BenH and he suspected a problem


The thing is that I was unable to boot. It hangs at the start at :
	smp_core99_kick_cpu done

Maybe was it a misconfiguration in the OpenFirmware :
when I list the value of the vars in OF, i've got two vars :
	betty_cpu0 passed
	betty_cpu1 ......
Oops, I can't log today on the computer, I don't remember the second
value :-( which was different from the first.
I test it on 2 G4 with exactly the same hardware configuration.

By the way, MacOS 9.2 was highly unstable : it often hangs on netscape.

Cheers,
--
Eric Leblond
Mail: regit@regit.org
Page Linux : http://www-anp.lip6.fr/~leblond/linux/
*---------------------------------------------------------------------------*
 Computers are like air conditioners. They don't work when Windows are
open.
*---------------------------------------------------------------------------*

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

Re: Support for new Dual G4 1 Gig machines?

[Alain RICHARD]

Just for information, why do you consider sungem better than gmac ? What
kind of functionnalities is it supposed to better handle than gmac ?

Also about the video cards, there is two options :

- ATI Radeon 7500 (32 Mo)
- NVIDIA GeForce4 MX (64 Mo of RAM)

Is there any clue about the support of each one ?

For information, I have used NVidia cards under intel linux boxes and it
is very boring having to get the driver from Nvidia and to patch the
kernel and the XFree with their driver. Also I don't think they produce a
driver for ppc linux.

Le mardi 29 janvier 2002, à 08:08 , benh@kernel.crashing.org a écrit :

>
>>
>>> You should also use sungem preferably to the old gmac driver.
>>
>> Do you suspect that this is what is causing my crashes when the box is
>> exposed to moderately high network volume?
>
> Maybe, only tests will prove me right or wrong.
>
> Ben.
>
>
>
>
-------------------------------------------------------
Alain RICHARD <mailto:alain.richard@equation.fr>
EQUATION SA <http://www.equation.fr/>
Tel : +33 477 79 48 00	 Fax : +33 477 79 48 01
Applications client/serveur, ingénierie réseau et Linux


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

Re: Support for new Dual G4 1 Gig machines?

[benh]

>Just for information, why do you consider sungem better than gmac ? What
>kind of functionnalities is it supposed to better handle than gmac ?

no copy, better perfs (after the latest tuning we did), better management
of link (including fallback to forced speeds if autoneg fails, ...), support
for ethtool ioctls, better power management (especially the chip isn't shut
down immediately on ifconfig down, but only after about 10 seconds which means
a dhcp client will work better).

>Also about the video cards, there is two options :
>
>- ATI Radeon 7500 (32 Mo)
>- NVIDIA GeForce4 MX (64 Mo of RAM)
>
>Is there any clue about the support of each one ?

ATI will probably work properly with 2D accel, end eventually DRI soon,
while nVidia will be problematic since they don't provide us with specs
nor drivers.

Ben.


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