Linux 2.4.22-pre2 and AthlonMP

Linux 2.4.22-pre2 and AthlonMP

[Edward Tandi]

On the Tyan Tiger board (2460) fitted with two processors, if I use
'noapic' in the lilo boot options I get the following in messages (and
eventually a crash):


Jun 28 17:54:36 machine syslogd 1.4.1: restart.
Jun 28 17:54:36 machine kernel: klogd 1.4.1, log source = /proc/kmsg
started.
Jun 28 17:54:36 machine kernel: Inspecting /boot/System.map
Jun 28 17:54:36 machine partmon: Checking if partitions have enough free
diskspace:
Jun 28 17:54:36 machine kernel:  40(40)
Jun 28 17:54:36 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:36 machine last message repeated 1127 times
Jun 28 17:54:36 machine partmon: ^[[65G[
Jun 28 17:54:36 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:36 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:36 machine partmon:
Jun 28 17:54:36 machine rc: Starting partmon:  succeeded
Jun 28 17:54:36 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:37 machine last message repeated 18 times
Jun 28 17:54:37 machine nfslock: rpc.lockd startup succeeded
Jun 28 17:54:37 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:37 machine last message repeated 5 times
Jun 28 17:54:37 machine rpc.statd[744]: Version 1.0.1 Starting
Jun 28 17:54:37 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:37 machine nfslock: rpc.statd startup succeeded
Jun 28 17:54:37 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:54:37 machine last message repeated 66 times
...
Jun 28 17:56:12 machine smbd[1925]:   Got SIGHUP
Jun 28 17:56:12 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:56:43 machine last message repeated 3057 times
Jun 28 17:57:44 machine last message repeated 6149 times
Jun 28 17:58:01 machine last message repeated 1808 times
Jun 28 17:58:01 machine ntpd[1023]: kernel time discipline status change
41
Jun 28 17:58:01 machine kernel: APIC error on CPU1: 40(40)
Jun 28 17:58:32 machine last message repeated 3058 times


The system boots, but hangs shortly thereafter. Removing the noapic
option, I get the following new messages:


Jun 28 18:27:46 machine kernel: ACPI: PCI Interrupt Link [LNKA] (IRQs 3
5 10 11, enabled at IRQ 9)
Jun 28 18:27:46 machine kernel: ACPI: PCI Interrupt Link [LNKB] (IRQs 3
*5 10 11)
Jun 28 18:27:46 machine kernel: ACPI: PCI Interrupt Link [LNKC] (IRQs 3
5 *10 11)
Jun 28 18:27:46 machine kernel: ACPI: PCI Interrupt Link [LNKD] (IRQs 3
5 10 *11)
Jun 28 18:27:46 machine kernel:     ACPI-0352: *** Error: Looking up
[Z00Q] in namespace, AE_NOT_FOUND
Jun 28 18:27:46 machine kernel: search_node c1596ac0 start_node c1596ac0
return_node 00000000
Jun 28 18:27:46 machine kernel:     ACPI-1121: *** Error: Method
execution failed [\_SB_.PCI0.ISA_.SIO_.COM1._STA] (Node c1596ac0),
AE_NOT_FOUND
Jun 28 18:27:46 machine kernel:     ACPI-0352: *** Error: Looking up
[Z00Q] in namespace, AE_NOT_FOUND
Jun 28 18:27:46 machine kernel: search_node c1596dc0 start_node c1596dc0
return_node 00000000
Jun 28 18:27:46 machine kernel:     ACPI-1121: *** Error: Method
execution failed [\_SB_.PCI0.ISA_.SIO_.COM2._STA] (Node c1596dc0),
AE_NOT_FOUND
Jun 28 18:27:46 machine kernel:     ACPI-0352: *** Error: Looking up
[Z00Q] in namespace, AE_NOT_FOUND
Jun 28 18:27:46 machine kernel: search_node c1594300 start_node c1594300
return_node 00000000
Jun 28 18:27:46 machine kernel:     ACPI-1121: *** Error: Method
execution failed [\_SB_.PCI0.ISA_.SIO_.LPT_._STA] (Node c1594300),
AE_NOT_FOUND
Jun 28 18:27:46 machine kernel: PCI: Probing PCI hardware
Jun 28 18:27:46 machine partmon: ^[[65G[
Jun 28 18:27:46 machine kernel: PCI: Using ACPI for IRQ routing
Jun 28 18:27:46 machine kernel: PCI: if you experience problems, try
using option 'pci=noacpi' or even 'acpi=off'
Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
compliance, fixing this error.


I presume this last set of messages is due to the recent ACPI changes.
If I try booting with the suggested 'pci=noacpi', the machine hangs
during boot before it gets to the SCSI driver. Setting 'acpi=off' gets
rid of the messages and the box appears to run OK.

Ed-T.

Re: Linux 2.4.22-pre2 and AthlonMP

[Alan Cox]

On Sad, 2003-06-28 at 21:50, Edward Tandi wrote:
> Jun 28 18:27:46 machine kernel: PCI: Using ACPI for IRQ routing
> Jun 28 18:27:46 machine kernel: PCI: if you experience problems, try
> using option 'pci=noacpi' or even 'acpi=off'
> Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> compliance, fixing this error.

Start by upgrading to their current BIOS

Re: Linux 2.4.22-pre2 and AthlonMP

[Edward Tandi]

On Sat, 2003-06-28 at 22:51, Alan Cox wrote:
> On Sad, 2003-06-28 at 21:50, Edward Tandi wrote:
> > Jun 28 18:27:46 machine kernel: PCI: Using ACPI for IRQ routing
> > Jun 28 18:27:46 machine kernel: PCI: if you experience problems, try
> > using option 'pci=noacpi' or even 'acpi=off'
> > Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> > compliance, fixing this error.
> 
> Start by upgrading to their current BIOS

Believe or not, it _is_ the latest bios for that board
(Tyan S2460 BIOS v1.05, 2nd Jan 2003).

Ed-T.

Re: Linux 2.4.22-pre2 and AthlonMP

[Alan Cox]

On Sad, 2003-06-28 at 23:50, Edward Tandi wrote:
> > > using option 'pci=noacpi' or even 'acpi=off'
> > > Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> > > compliance, fixing this error.
> > 
> > Start by upgrading to their current BIOS
> 
> Believe or not, it _is_ the latest bios for that board
> (Tyan S2460 BIOS v1.05, 2nd Jan 2003).

Then I guess you have a problem. We try and fix up BIOS problems but there
is a limit to what we can do, and if it has problems like the one that is
logged I'd be worried what else it might do - eg I suspect Nvidia 4x AGP cards
aren't too solid on it.

The APIC errors also suggest something isn't happy at all at the hardware
layer. Are you using MP processors ?

Re: Linux 2.4.22-pre2 and AthlonMP

[Edward Tandi]

On Sun, 2003-06-29 at 00:17, Alan Cox wrote:
> On Sad, 2003-06-28 at 23:50, Edward Tandi wrote:
> > > > using option 'pci=noacpi' or even 'acpi=off'
> > > > Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> > > > compliance, fixing this error.
> > > 
> > > Start by upgrading to their current BIOS
> > 
> > Believe or not, it _is_ the latest bios for that board
> > (Tyan S2460 BIOS v1.05, 2nd Jan 2003).
> 
> Then I guess you have a problem. We try and fix up BIOS problems but there
> is a limit to what we can do, and if it has problems like the one that is
> logged I'd be worried what else it might do - eg I suspect Nvidia 4x AGP cards
> aren't too solid on it.

It does have an AGP NVidia card in it. I'm using the standard XFree
drivers with it at the moment but I have played UT on it for hours
before (using NVidia drivers) without problems. It might be an AGP x2
card though. The computer is now mostly a back-end server and I haven't
really pushed it on the graphics side recently.

Could the problem be caused by some BIOS setting? I could spend some
time looking at them.

> The APIC errors also suggest something isn't happy at all at the hardware
> layer. Are you using MP processors ?

Yes, MP processors. This is not a new machine, It has been running quite
nicely for nearly two years. There have been some kernel releases in the
past that have shown some instability, but I can usually find a fairly
recent version and tweak the kernel-build settings so that it becomes
stable.

The version running prior to this one was 2.4.21-rc3. This version
allowed me to specify noapic.

Ed-T.

> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Linux 2.4.22-pre2 and AthlonMP?

[Edward Tandi]

The what processor thread...

On Sun, 2003-06-29 at 01:04, Edward Tandi wrote:
> On Sun, 2003-06-29 at 00:17, Alan Cox wrote:
> > On Sad, 2003-06-28 at 23:50, Edward Tandi wrote:
> > > > > using option 'pci=noacpi' or even 'acpi=off'
> > > > > Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> > > > > compliance, fixing this error.
> > > > 
> > > > Start by upgrading to their current BIOS
> > > 
> > > Believe or not, it _is_ the latest bios for that board
> > > (Tyan S2460 BIOS v1.05, 2nd Jan 2003).
> > 
> > Then I guess you have a problem. We try and fix up BIOS problems but there
> > is a limit to what we can do, and if it has problems like the one that is
> > logged I'd be worried what else it might do - eg I suspect Nvidia 4x AGP cards
> > aren't too solid on it.
> > 
> > The APIC errors also suggest something isn't happy at all at the hardware
> > layer. Are you using MP processors ?
> 
> I have to admit, I have noticed something a little odd coming out of
> /proc/cpuinfo:
> 
> processor       : 0
> vendor_id       : AuthenticAMD
> cpu family      : 6
> model           : 6
> model name      : AMD Athlon(tm) MP
> stepping        : 1
> cpu MHz         : 1194.690
> cache size      : 256 KB
> fdiv_bug        : no
> hlt_bug         : no
> f00f_bug        : no
> coma_bug        : no
> fpu             : yes
> fpu_exception   : yes
> cpuid level     : 1
> wp              : yes
> flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> cmov pat
> pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> bogomips        : 2385.51
>                                                                                 
> processor       : 1
> vendor_id       : AuthenticAMD
> cpu family      : 6
> model           : 6
> model name      : AMD Athlon(tm) Processor
> stepping        : 2
> cpu MHz         : 1194.690
> cache size      : 256 KB
> fdiv_bug        : no
> hlt_bug         : no
> f00f_bug        : no
> coma_bug        : no
> fpu             : yes
> fpu_exception   : yes
> cpuid level     : 1
> wp              : yes
> flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> cmov pat
> pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> bogomips        : 2385.51
> 
> What confuses me here is how on earth the second processor reports
> itself without the "MP" bit and with a stepping of 2. They were
> identical processors when I put them in and I haven't touched them
> since. Is there any way this could be reported wrongly?

Further info on this, x86info gives the following results:

x86info v1.7.  Dave Jones 2001
Feedback to <davej@suse.de>.
 
Found 2 CPUs
CPU #1
Family: 6 Model: 6 Stepping: 1 [Athlon 4 (Palomino core) Rev A2]
Processor name string: AMD Athlon(tm) MP
 
PowerNOW! Technology information
Available features:
        Temperature sensing diode present.
 
CPU #2
Family: 6 Model: 6 Stepping: 2 [Athlon MP]
Processor name string: AMD Athlon(tm) Processor
 
PowerNOW! Technology information
Available features:
        Temperature sensing diode present.
 

It looks like the processors may have been from two different batches!
How bizarre. Should this make any difference?

Ed-T.

Re: Linux 2.4.22-pre2 and AthlonMP?

[Willy TARREAU]

Hi Edward,

I too got rather strange results with ASUS boards :

- on the first one (first rev A7M266D, without on-board USB),
  I stuck 2 Athlon-XP 1800 (yes, I risked XPs because it's only
  a dev machine and at that time, XPs were less than half the
  price of MPs). They were seen as XPs by the BIOS (rev 1004)
  and by Linux. I had a few problems with APIC at that time
  and Alan suggested me to try a more recent BIOS, which I did.
  Now, with rev 1007, the BIOS tells me I have two AthlonMP,
  which are both MP capable. Linux now sees a model name changed
  to AthlonMP :

  willy@pcw:willy$ cat /proc/cpuinfo
  processor       : 0
  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 6
  model name      : AMD Athlon(TM) MP 1800+
  stepping        : 2
  cpu MHz         : 1546.160
  cache size      : 256 KB
  fdiv_bug        : no
  hlt_bug         : no
  f00f_bug        : no
  coma_bug        : no
  fpu             : yes
  fpu_exception   : yes
  cpuid level     : 1
  wp              : yes
  flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
  bogomips        : 3039.23

  processor       : 1
  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 6
  model name      : AMD Athlon(TM) MP 1800+
  stepping        : 2
  cpu MHz         : 1546.160
  cache size      : 256 KB
  fdiv_bug        : no
  hlt_bug         : no
  f00f_bug        : no
  coma_bug        : no
  fpu             : yes
  fpu_exception   : yes
  cpuid level     : 1
  wp              : yes
  flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
  bogomips        : 3088.38

==> the BIOS seems to change this itself at boot time. Now I
    have absolutely no way to tell if I really have MPs or XPs,
    except opening the box and removing the HS/Fan, or downgrading
    the BIOS ! My mistake is not to have checked if
    family/model/stepping were the same before the BIOS upgrade,
    but honnestly I didn't expect the BIOS to affect the way the
    processor reports itself to the cpuid instruction ! But x86info
    also tells me I have MPs, and I think I would have noticed it
    if it had already been the case before...


  willy@pcw:willy$ x86info
  x86info v1.11.  Dave Jones 2001, 2002
  Feedback to <davej@suse.de>.

  Found 2 CPUs
  CPU #1
  Family: 6 Model: 6 Stepping: 2
  CPU Model : Athlon MP (palomino)
  Processor name string: AMD Athlon(TM) MP 1800+

  PowerNOW! Technology information
  Available features:
        Temperature sensing diode present.


- the second board, a new generation ASUS A7M266D (with working
  on-board USB) :
  I bought this card to use at work as a heavy computation/compilation
  server. This time I didn't want to play with stability, so I bought
  two real AthlonMP 2200 sold in an AMD box labeled AthlonMP, etc...
  I believe the BIOS is rev 1007 too. And guess what ? at boot, the bios
  stops and tells me that CPU1 is NOT MP capable !!! It doesn't come from
  the CPU since I swapped them and only CPU1 is reported not to be MP
  capable ! Which means that, according to that crappy BIOS, my XPs are
  more MP capable than true MPs ! And the must :

  root@aluminium:root# cat /proc/cpuinfo
  processor       : 0
  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 8
  model name      : AMD Duron(TM) MP Processor
  stepping        : 0
  cpu MHz         : 1800.040
  cache size      : 256 KB
  fdiv_bug        : no
  hlt_bug         : no
  f00f_bug        : no
  coma_bug        : no
  fpu             : yes
  fpu_exception   : yes
  cpuid level     : 1
  wp              : yes
  flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
  bogomips        : 3547.13

  processor       : 1
  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 8
  model name      : AMD Duron(TM) MP Processor
  stepping        : 0
  cpu MHz         : 1800.040
  cache size      : 256 KB
  fdiv_bug        : no
  hlt_bug         : no
  f00f_bug        : no
  coma_bug        : no
  fpu             : yes
  fpu_exception   : yes
  cpuid level     : 1
  wp              : yes
  flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
  bogomips        : 3596.28


Notice how they now are named "Duron MP" !
Fortunately, x86info is not fooled by this stupid BIOS :

  x86info v1.11.  Dave Jones 2001, 2002
  Feedback to <davej@suse.de>.
  
  Found 2 CPUs
  CPU #1
  Family: 6 Model: 8 Stepping: 0
  CPU Model : Athlon MP (Thoroughbred)[A0]
  Processor name string: AMD Duron(TM) MP Processor


===> To conclude, I would say don't worry about the model name, since
     the BIOS seems to have the ability to change it to whatever it thinks
     is appropriate (I would laugh to see a 'GenuineIntel' here :-)). But
     your different steppings are more worrying. You may want to swap the
     two CPUs to see if the steppings follow the CPU or the socket, but you
     need some thermal grease if you dismount the heatsink/fan.

BTW, concerning stability, the two machines (XP and MP) have been rock solid,
sometimes spending a full day compiling kernels in parallel. I even made my
XPs more silent by slowing down the fans, and they can go as high as 92 degrees
after a few hours of intensive compilation, but I never have a stability problem
with them. OTOH, a collegue of mine bought two MP2000 which crashed in more than
one hour of intensive computation, and once the reseller finally replaced the
CPUs which were already AMD box, recommended fan..., the problems were gone !
So I begin to wonder if XPs would not be more reliable than MPs since they are
sold far more often...

Cheers,
Willy

[I leave your mail here for reference]

On Sun, Jun 29, 2003 at 01:58:54AM +0100, Edward Tandi wrote:
> The what processor thread...
> 
> On Sun, 2003-06-29 at 01:04, Edward Tandi wrote:
> > On Sun, 2003-06-29 at 00:17, Alan Cox wrote:
> > > On Sad, 2003-06-28 at 23:50, Edward Tandi wrote:
> > > > > > using option 'pci=noacpi' or even 'acpi=off'
> > > > > > Jun 28 18:27:46 machine kernel: BIOS failed to enable PCI standards
> > > > > > compliance, fixing this error.
> > > > > 
> > > > > Start by upgrading to their current BIOS
> > > > 
> > > > Believe or not, it _is_ the latest bios for that board
> > > > (Tyan S2460 BIOS v1.05, 2nd Jan 2003).
> > > 
> > > Then I guess you have a problem. We try and fix up BIOS problems but there
> > > is a limit to what we can do, and if it has problems like the one that is
> > > logged I'd be worried what else it might do - eg I suspect Nvidia 4x AGP cards
> > > aren't too solid on it.
> > > 
> > > The APIC errors also suggest something isn't happy at all at the hardware
> > > layer. Are you using MP processors ?
> > 
> > I have to admit, I have noticed something a little odd coming out of
> > /proc/cpuinfo:
> > 
> > processor       : 0
> > vendor_id       : AuthenticAMD
> > cpu family      : 6
> > model           : 6
> > model name      : AMD Athlon(tm) MP
> > stepping        : 1
> > cpu MHz         : 1194.690
> > cache size      : 256 KB
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 1
> > wp              : yes
> > flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> > cmov pat
> > pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> > bogomips        : 2385.51
> >                                                                                 
> > processor       : 1
> > vendor_id       : AuthenticAMD
> > cpu family      : 6
> > model           : 6
> > model name      : AMD Athlon(tm) Processor
> > stepping        : 2
> > cpu MHz         : 1194.690
> > cache size      : 256 KB
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 1
> > wp              : yes
> > flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> > cmov pat
> > pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> > bogomips        : 2385.51
> > 
> > What confuses me here is how on earth the second processor reports
> > itself without the "MP" bit and with a stepping of 2. They were
> > identical processors when I put them in and I haven't touched them
> > since. Is there any way this could be reported wrongly?
> 
> Further info on this, x86info gives the following results:
> 
> x86info v1.7.  Dave Jones 2001
> Feedback to <davej@suse.de>.
>  
> Found 2 CPUs
> CPU #1
> Family: 6 Model: 6 Stepping: 1 [Athlon 4 (Palomino core) Rev A2]
> Processor name string: AMD Athlon(tm) MP
>  
> PowerNOW! Technology information
> Available features:
>         Temperature sensing diode present.
>  
> CPU #2
> Family: 6 Model: 6 Stepping: 2 [Athlon MP]
> Processor name string: AMD Athlon(tm) Processor
>  
> PowerNOW! Technology information
> Available features:
>         Temperature sensing diode present.
>  
> 
> It looks like the processors may have been from two different batches!
> How bizarre. Should this make any difference?
> 
> Ed-T.
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Linux 2.4.22-pre2 and AthlonMP?

[Alan Cox]

On Sul, 2003-06-29 at 08:45, Willy TARREAU wrote:
> ===> To conclude, I would say don't worry about the model name, since
>      the BIOS seems to have the ability to change it to whatever it thinks
>      is appropriate (I would laugh to see a 'GenuineIntel' here :-)). But

You can twiddle the model data if you have the right (NDA) docs, and yes
some old BIOSen get it wrong.

Re: Linux 2.4.22-pre2 and AthlonMP

[Alan Cox]

On Sul, 2003-06-29 at 00:52, Edward Tandi wrote:
> It does have an AGP NVidia card in it. I'm using the standard XFree
> drivers with it at the moment but I have played UT on it for hours
> before (using NVidia drivers) without problems. It might be an AGP x2
> card though. The computer is now mostly a back-end server and I haven't
> really pushed it on the graphics side recently
> 
> Could the problem be caused by some BIOS setting? I could spend some
> time looking at them.

The BIOS has magic tuning tables for AMD76x chipsets for various video
cards. Its one of the reasons that new BIOSes sometimes make AGP 4x
work, or more reliable.

> The version running prior to this one was 2.4.21-rc3. This version
> allowed me to specify noapic.

Out of interest, compile out ACPI support and see what it does

Re: Linux 2.4.22-pre2 and AthlonMP

[Edward Tandi]

On Sun, 2003-06-29 at 11:42, Alan Cox wrote:
> On Sul, 2003-06-29 at 00:52, Edward Tandi wrote:
> > It does have an AGP NVidia card in it. I'm using the standard XFree
> > drivers with it at the moment but I have played UT on it for hours
> > before (using NVidia drivers) without problems. It might be an AGP x2
> > card though. The computer is now mostly a back-end server and I haven't
> > really pushed it on the graphics side recently
> > 
> > Could the problem be caused by some BIOS setting? I could spend some
> > time looking at them.
> 
> The BIOS has magic tuning tables for AMD76x chipsets for various video
> cards. Its one of the reasons that new BIOSes sometimes make AGP 4x
> work, or more reliable.
> 
> > The version running prior to this one was 2.4.21-rc3. This version
> > allowed me to specify noapic.
> 
> Out of interest, compile out ACPI support and see what it does

OK, compiled without ACPI. The system boots and runs fine with or
without noapic. No nasty warnings.

The AMD76x power management has been off in all tests to date.

Ed-T.

Re: Linux 2.4.22-pre2 and AthlonMP?

[Edward Tandi]

On Sun, 2003-06-29 at 01:58, Edward Tandi wrote:
> The what processor thread...

For purely academic interest, and to satisfy my own curiosity, I have
pulled the two processors out of the box to examine their markings.
Here they are:


Processor 1
-----------
| AHX1200AMS3C == Athlon MP 1200MHz (0.18um), CPGA package, 1.75V, 95'C
max temp, 256K L2 cache, 266MHz max bus speed.

| AGKGA 0137MPMW
Athlon XP Palomino core, manufactured 37th week of the year 2001.


Processor 2
-----------
| AHX1200AMS3C == Athlon MP 1200MHz (0.18um), CPGA package, 1.75V, 95'C
max temp, 256K L2 cache, 266MHz max bus speed.

| AGHCA 0129CPAW
Unknown core, manufactured 29th week of the year 2001.


So although they look identical, have the same model number and were
produced only 8 weeks apart, they really do have a different stepping
(and core) identifier. I couldn't find what the AGHCA corresponded to.

Ed-T.


> > I have to admit, I have noticed something a little odd coming out of
> > /proc/cpuinfo:
> > 
> > processor       : 0
> > vendor_id       : AuthenticAMD
> > cpu family      : 6
> > model           : 6
> > model name      : AMD Athlon(tm) MP
> > stepping        : 1
> > cpu MHz         : 1194.690
> > cache size      : 256 KB
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 1
> > wp              : yes
> > flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> > cmov pat
> > pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> > bogomips        : 2385.51
> >                                                                                 
> > processor       : 1
> > vendor_id       : AuthenticAMD
> > cpu family      : 6
> > model           : 6
> > model name      : AMD Athlon(tm) Processor
> > stepping        : 2
> > cpu MHz         : 1194.690
> > cache size      : 256 KB
> > fdiv_bug        : no
> > hlt_bug         : no
> > f00f_bug        : no
> > coma_bug        : no
> > fpu             : yes
> > fpu_exception   : yes
> > cpuid level     : 1
> > wp              : yes
> > flags           : fpu vme de tsc msr pae mce cx8 apic sep mtrr pge mca
> > cmov pat
> > pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
> > bogomips        : 2385.51
> > 
> > What confuses me here is how on earth the second processor reports
> > itself without the "MP" bit and with a stepping of 2. They were
> > identical processors when I put them in and I haven't touched them
> > since. Is there any way this could be reported wrongly?
> 
> Further info on this, x86info gives the following results:
> 
> x86info v1.7.  Dave Jones 2001
> Feedback to <davej@suse.de>.
>  
> Found 2 CPUs
> CPU #1
> Family: 6 Model: 6 Stepping: 1 [Athlon 4 (Palomino core) Rev A2]
> Processor name string: AMD Athlon(tm) MP
>  
> PowerNOW! Technology information
> Available features:
>         Temperature sensing diode present.
>  
> CPU #2
> Family: 6 Model: 6 Stepping: 2 [Athlon MP]
> Processor name string: AMD Athlon(tm) Processor
>  
> PowerNOW! Technology information
> Available features:
>         Temperature sensing diode present.