Re: SMP or UP???

Re: SMP or UP???

[Mikael Pettersson]

On Mon, 12 Jan 2004 11:20:39 +0100, DervishD <raul@pleyades.net> wrote:
>    I have an used Gigabyte GA-7ZXE mobo, with chipset VIA KT133A,
>wearing a Duron 1000 processor, since December. AFAIK this mobo is
>uniprocessor (it has only a socket, that should be a good
>evidence...), but when booting I get these messages (I just show
>those relevant to the issue):
>
>kernel: Linux version 2.4.21 (root@DervishD) (gcc version 3.2.2) #1 Wed Jul 2 17:25:21 CEST 2003
>kernel: found SMP MP-table at 000fb210
>
>    What, SMP table?
...
>    This is not an issue, because the system seems to work OK, but
>for me is very strange and I'm not sure wether this may cause
>problems or not...

You have an anti-problem. The chipset includes an I/O-APIC
(good) and your mobo manufacturer was decent enough to include
the appropriate BIOS MP tables to describe it to the OS.

Other manufacturers skip the MP table, forcing you to enable
ACPI and pray it actually works. 

This is one area where Gigabyte seems to be consistently
better (more user-friendly) than ASUS.

/Mikael

Re: SMP or UP???

[DervishD]

    Hi Mikael :)

 * Mikael Pettersson <mikpe@csd.uu.se> dixit:
> >kernel: found SMP MP-table at 000fb210
> >    What, SMP table?
> You have an anti-problem. The chipset includes an I/O-APIC
> (good) and your mobo manufacturer was decent enough to include
> the appropriate BIOS MP tables to describe it to the OS.

    Oh, nice. I thought that the mobo was a simple reisuing of a SMP
mobo from Gigabyte with one socket removed O:)

> Other manufacturers skip the MP table, forcing you to enable
> ACPI and pray it actually works. 

    Excuse my ignorance but: why a UP system needs the MP table? Why
the I/O-APIC needs anything related with multiprocessor in an UP
system?. I lost my way on hardware back in the 486, I think...

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736
http://www.pleyades.net & http://raul.pleyades.net/

Re: SMP or UP???

[Helge Hafting]

DervishD wrote:
>     Hi Mikael :)
> 
>  * Mikael Pettersson <mikpe@csd.uu.se> dixit:
> 
>>>kernel: found SMP MP-table at 000fb210
>>>   What, SMP table?
>>
>>You have an anti-problem. The chipset includes an I/O-APIC
>>(good) and your mobo manufacturer was decent enough to include
>>the appropriate BIOS MP tables to describe it to the OS.
> 
> 
>     Oh, nice. I thought that the mobo was a simple reisuing of a SMP
> mobo from Gigabyte with one socket removed O:)
> 
> 
>>Other manufacturers skip the MP table, forcing you to enable
>>ACPI and pray it actually works. 
> 
> 
>     Excuse my ignorance but: why a UP system needs the MP table? Why
> the I/O-APIC needs anything related with multiprocessor in an UP
> system?. I lost my way on hardware back in the 486, I think...
> 
The MP table tells the kernel details about that I/O-APIC.
(Advanced Programmable Interrupt Controller).

This isn't really about SMP, but every SMP board has
one or more APICs.  They have to.  It is optional
on a uniprocessor board, but it is nice to have as it
gives lower interrupt latency.

Helge Hafting

Re: SMP or UP???

[DervishD]

    Hi Helge :)

 * Helge Hafting <helgehaf@aitel.hist.no> dixit:
> >    Excuse my ignorance but: why a UP system needs the MP table? Why
> >the I/O-APIC needs anything related with multiprocessor in an UP
> >system?. I lost my way on hardware back in the 486, I think...
> The MP table tells the kernel details about that I/O-APIC.
> (Advanced Programmable Interrupt Controller).
> This isn't really about SMP, but every SMP board has
> one or more APICs.  They have to.  It is optional
> on a uniprocessor board, but it is nice to have as it
> gives lower interrupt latency.

    Thanks a lot for the information :)

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736
http://www.pleyades.net & http://raul.pleyades.net/

SMP or UP???

[DervishD]

    Hi all :))

    I have an used Gigabyte GA-7ZXE mobo, with chipset VIA KT133A,
wearing a Duron 1000 processor, since December. AFAIK this mobo is
uniprocessor (it has only a socket, that should be a good
evidence...), but when booting I get these messages (I just show
those relevant to the issue):

kernel: Linux version 2.4.21 (root@DervishD) (gcc version 3.2.2) #1 Wed Jul 2 17:25:21 CEST 2003
kernel: found SMP MP-table at 000fb210

    What, SMP table?

kernel: hm, page 000fb000 reserved twice.
kernel: hm, page 000fc000 reserved twice.
kernel: hm, page 000f5000 reserved twice.
kernel: hm, page 000f6000 reserved twice.
kernel: Intel MultiProcessor Specification v1.1
kernel:     Virtual Wire compatibility mode.

    Is this chipset then SMP capable? 

kernel: OEM ID: VIA      Product ID: KT133        APIC at: 0xFEE00000
kernel: Processor #0 Pentium(tm) Pro APIC version 17

    What Pentium? This is an AMD mobo :??

kernel: I/O APIC #2 Version 17 at 0xFEC00000.
kernel: Enabling APIC mode: Flat.?Using 1 I/O APICs
kernel: Processors: 1

    When booting some distros, like Knoppix, I have an 'Error: only
one processor found' log message.

kernel: Initializing CPU#0
kernel: Detected 1000.077 MHz processor.
kernel: Calibrating delay loop... 1992.29 BogoMIPS
kernel: CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
kernel: CPU: L2 Cache: 64K (64 bytes/line)
kernel: CPU: AMD Duron(tm) stepping 00

    Well, finally my Duron got detected...

    This is not an issue, because the system seems to work OK, but
for me is very strange and I'm not sure wether this may cause
problems or not...

    Thanks in advance :)

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736
http://www.pleyades.net & http://raul.pleyades.net/

Re: SMP or UP???

[Matthew D. Pitts]

Raul,

I have a Gigabyte GA-7VAX and get the same messages. It probably is normal,
since the AMD CPU is functionally the same as an Intel one.

Matthew