[gentoo-user] swapping processor problem

[gentoo-user] swapping processor problem

[Raffaele BELARDI]

I have two gentoo boxes, X has an ASUS M2NPV-VM with AMD64 3500+ CPU, Y
has a AMD64 X2 5600+ CPU. Since I need more juice on X I thought I could
swap CPUs.

After updating X's BIOS the system with the 'new' CPU boots up to the
MythTv screen with no error but does not respond to the USB keyboard nor
to the remote control keypresses. More precisely:

- keyboard is fine at grub boot, I can select up and down or edit the
entries
- keyboard is no longer responsive at the init scripts start (when you
can press 'I' to select services)

My first thought was that X (and Y) WERE compiled with '-march=native'
GCC flag and maybe the 5600+ does not execute properly the 3500+ code.
But a quick search on wikipedia shows that 5600+ has a superset of the
3500+ so I should have problems putting the 3500+ in the Y box, not
vice-versa.

Any suggestions?

raffaele

[gentoo-user] Re: swapping processor problem

[Raffaele BELARDI]

On 11/15/2011 08:33 AM, Raffaele Belardi wrote:
> I have two gentoo boxes, X has an ASUS M2NPV-VM with AMD64 3500+ CPU, Y
> has a AMD64 X2 5600+ CPU. Since I need more juice on X I thought I could
> swap CPUs.
> 
> After updating X's BIOS the system with the 'new' CPU boots up to the
> MythTv screen with no error but does not respond to the USB keyboard nor
> 
> My first thought was that X (and Y) WERE compiled with '-march=native'
> GCC flag and maybe the 5600+ does not execute properly the 3500+ code.
> But a quick search on wikipedia shows that 5600+ has a superset of the
> 3500+ so I should have problems putting the 3500+ in the Y box, not
> vice-versa.

I did some more testing, the lock occurs also when booting from CD so I
excluded a software problem. So I tried swapping also the PSU, still
locks. I ended up swapping also the motherboards, apparently no locks
(but I need a new kernel). So it looks like the combination of ASUS
M2NPV-VM motherboard plus MD64 X2 5600+ CPU was causing the lock. I just
don't understand why.

Re: [gentoo-user] swapping processor problem

[Michael Mol]

On Tue, Nov 15, 2011 at 2:33 AM, Raffaele BELARDI
<raffaele.belardi@st.com> wrote:
> I have two gentoo boxes, X has an ASUS M2NPV-VM with AMD64 3500+ CPU, Y
> has a AMD64 X2 5600+ CPU. Since I need more juice on X I thought I could
> swap CPUs.
>
> After updating X's BIOS the system with the 'new' CPU boots up to the
> MythTv screen with no error but does not respond to the USB keyboard nor
> to the remote control keypresses. More precisely:
>
> - keyboard is fine at grub boot, I can select up and down or edit the
> entries
> - keyboard is no longer responsive at the init scripts start (when you
> can press 'I' to select services)
>
> My first thought was that X (and Y) WERE compiled with '-march=native'
> GCC flag and maybe the 5600+ does not execute properly the 3500+ code.
> But a quick search on wikipedia shows that 5600+ has a superset of the
> 3500+ so I should have problems putting the 3500+ in the Y box, not
> vice-versa.
>
> Any suggestions?

Play with your BIOS settings. Look for things like legacy USB support.
Also, double-check that all the relevant USB drivers (UHCI, EHCI,
XHCI, HID, etc) are either built-into the kernel, or are loaded as
modules. Consider rebuilding your kernel.

Just because one processor has a superset of the instructions of the
other doesn't mean there may not be other compatibilities. Some time
back, a thread on here discussed how to find out what -march=native
becomes, in terms of -march and a bunch of other parameters. I've
noticed parameters like cache line sizes and cache sizes, among a
couple others. I imagine a bungling of, e.g., cache line sizes could
break code that has heavy dependency on data locality and/or memory
models; it might have broken your USB drivers, for example.

But, really, I think BIOS settings and driver presence are the more
likely culprit.

-- 
:wq

Re: [gentoo-user] swapping processor problem

[Raffaele BELARDI]

On 11/16/2011 08:11 AM, Michael Mol wrote:
> Play with your BIOS settings. Look for things like legacy USB support.
> Also, double-check that all the relevant USB drivers (UHCI, EHCI,
> XHCI, HID, etc) are either built-into the kernel, or are loaded as
> modules. Consider rebuilding your kernel.
> 
> Just because one processor has a superset of the instructions of the
> other doesn't mean there may not be other compatibilities. Some time
> back, a thread on here discussed how to find out what -march=native
> becomes, in terms of -march and a bunch of other parameters. I've
> noticed parameters like cache line sizes and cache sizes, among a
> couple others. I imagine a bungling of, e.g., cache line sizes could
> break code that has heavy dependency on data locality and/or memory
> models; it might have broken your USB drivers, for example.
> 
> But, really, I think BIOS settings and driver presence are the more
> likely culprit.
> 

I don't think it was a software issue because the same lock happened
also booting from a live-CD. I ended up swapping the motherboards, now
live-CD boots fine but I need a new kernel.

The march=native thread looks interesting, do you remember the subject
line so I can search it in the archives?

Re: [gentoo-user] swapping processor problem

[Dale]

Michael Mol wrote:
> Play with your BIOS settings. Look for things like legacy USB support. 
> Also, double-check that all the relevant USB drivers (UHCI, EHCI, 
> XHCI, HID, etc) are either built-into the kernel, or are loaded as 
> modules. Consider rebuilding your kernel. Just because one processor 
> has a superset of the instructions of the other doesn't mean there may 
> not be other compatibilities. Some time back, a thread on here 
> discussed how to find out what -march=native becomes, in terms of 
> -march and a bunch of other parameters. I've noticed parameters like 
> cache line sizes and cache sizes, among a couple others. I imagine a 
> bungling of, e.g., cache line sizes could break code that has heavy 
> dependency on data locality and/or memory models; it might have broken 
> your USB drivers, for example. But, really, I think BIOS settings and 
> driver presence are the more likely culprit. 

Here it is:

gcc -Q --help=target  -march=native

I agree tho that checking those BIOS setting is a good start.  If that fails, boot a CD or something, chroot in, do a emerge -e system.  Maybe make some corrections to the kernel then try booting.  Oh, I'd rebuild the input drivers to, mouse and keyboard.  Check the USE flags too.  I'm not sure what all options they have.


Dale

:-)  :-)

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] swapping processor problem

[Raffaele BELARDI]

On 11/16/2011 08:34 AM, Dale wrote:
> 
> I agree tho that checking those BIOS setting is a good start.  If that fails, boot a CD or something, chroot in, do a emerge -e system.  Maybe make some corrections to the kernel then try booting.  Oh, I'd rebuild the input drivers to, mouse and keyboard.  Check the USE flags too.  I'm not sure what all options they have.
> 

Boot from CD also was failing with a lock so I think it was a mobo/CPU
incompatibility issue. Nevertheless I'd like to understand the points
you and Michael are making.

The Award BIOS has almost no CPU-related options.
There is no way I can control cache size from BIOS.
Since the memory controller is integrated in the Athlon64 I'd think DRAM
settings page could be affected by a CPU swap but I have it all set to
'auto' so I suppose that the BIOS should fix them after the swap.
USB is implemented on southbridge, not on CPU.

I found no parameter in the kernel config that could be affected by the
CPU swap except for SMP setting, but a misconfiguration there should not
cause a hang, just a core not being used.

Which drivers should be affected by a CPU swap? All the ones I can think
of are motherboard-related, not CPU-related.

Thanks for the hint on gcc, I'll check the differences on the two
systems once I have them back up and working.

Re: [gentoo-user] swapping processor problem

[Dale]

Raffaele BELARDI wrote:
> On 11/16/2011 08:34 AM, Dale wrote:
>> I agree tho that checking those BIOS setting is a good start.  If that fails, boot a CD or something, chroot in, do a emerge -e system.  Maybe make some corrections to the kernel then try booting.  Oh, I'd rebuild the input drivers to, mouse and keyboard.  Check the USE flags too.  I'm not sure what all options they have.
>>
> Boot from CD also was failing with a lock so I think it was a mobo/CPU
> incompatibility issue. Nevertheless I'd like to understand the points
> you and Michael are making.
>
> The Award BIOS has almost no CPU-related options.
> There is no way I can control cache size from BIOS.
> Since the memory controller is integrated in the Athlon64 I'd think DRAM
> settings page could be affected by a CPU swap but I have it all set to
> 'auto' so I suppose that the BIOS should fix them after the swap.
> USB is implemented on southbridge, not on CPU.
>
> I found no parameter in the kernel config that could be affected by the
> CPU swap except for SMP setting, but a misconfiguration there should not
> cause a hang, just a core not being used.
>
> Which drivers should be affected by a CPU swap? All the ones I can think
> of are motherboard-related, not CPU-related.
>
> Thanks for the hint on gcc, I'll check the differences on the two
> systems once I have them back up and working.
>

You could be right that it is a mobo CPU issue.  When you are grasping 
at straws, just grab all you can.  Since you got it to work then I 
wouldn't think it was software, a failing hardware such as a bad CPU or 
a bad kernel.

Generally if it works on a CD, then it is software.  If it also fails 
from a CD, then it should be hardware.  Sounds like yours is a hardware 
issue but just a compatibility problem, not failure.

Maybe check if there is a BIOS update available?  Might be worth a shot.

Dale

:-)  :-)

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] swapping processor problem

[Raffaele BELARDI]

On 11/16/2011 09:23 AM, Dale wrote:
> You could be right that it is a mobo CPU issue.  When you are grasping 
> at straws, just grab all you can.

Yep, that's why I wrote about it here, I see very good suggestions on
this list.

> Maybe check if there is a BIOS update available?  Might be worth a shot.

I already upgraded the BIOS to have the dual-core CPU recognized,
otherwise the kernel would not even start. There is yet another update
on the ASUS site but it's mainly for AM3 processor compatibility and
labelled as 'beta' so I'm a bit nervous to try it ;-)

Re: [gentoo-user] swapping processor problem

[Dale]

Raffaele BELARDI wrote:
> On 11/16/2011 09:23 AM, Dale wrote:
>
>> Maybe check if there is a BIOS update available?  Might be worth a shot.
> I already upgraded the BIOS to have the dual-core CPU recognized,
> otherwise the kernel would not even start. There is yet another update
> on the ASUS site but it's mainly for AM3 processor compatibility and
> labelled as 'beta' so I'm a bit nervous to try it ;-)
>

I have a beta for mine too.  It'll will be there when it gets marked 
stable tho.  I'm in no hurry.  Sometimes waiting is a good idea.

Dale

:-)  :-)

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] swapping processor problem

[James Broadhead]

On 16 November 2011 08:55, Dale <rdalek1967@gmail.com> wrote:
> Raffaele BELARDI wrote:
>>
>> On 11/16/2011 09:23 AM, Dale wrote:
>>
>>> Maybe check if there is a BIOS update available?  Might be worth a shot.
>>
>> I already upgraded the BIOS to have the dual-core CPU recognized,
>> otherwise the kernel would not even start. There is yet another update
>> on the ASUS site but it's mainly for AM3 processor compatibility and
>> labelled as 'beta' so I'm a bit nervous to try it ;-)
>>
>
> I have a beta for mine too.  It'll will be there when it gets marked stable
> tho.  I'm in no hurry.  Sometimes waiting is a good idea.

The latest version of my Gigabyte BIOS has been 'in beta' for about 4
years now -- _and_ it's the version where they add support for
Quad-Core 2s!

Sometimes it means that they added some features, but can't be
bothered doing a full test suite.

Re: [gentoo-user] swapping processor problem

[Dale]

James Broadhead wrote:
> On 16 November 2011 08:55, Dale<rdalek1967@gmail.com>  wrote:
>> Raffaele BELARDI wrote:
>>> On 11/16/2011 09:23 AM, Dale wrote:
>>>
>>>> Maybe check if there is a BIOS update available?  Might be worth a shot.
>>> I already upgraded the BIOS to have the dual-core CPU recognized,
>>> otherwise the kernel would not even start. There is yet another update
>>> on the ASUS site but it's mainly for AM3 processor compatibility and
>>> labelled as 'beta' so I'm a bit nervous to try it ;-)
>>>
>> I have a beta for mine too.  It'll will be there when it gets marked stable
>> tho.  I'm in no hurry.  Sometimes waiting is a good idea.
> The latest version of my Gigabyte BIOS has been 'in beta' for about 4
> years now -- _and_ it's the version where they add support for
> Quad-Core 2s!
>
> Sometimes it means that they added some features, but can't be
> bothered doing a full test suite.
>
>

Mine has been there for quite a while too.  According to the timestamps 
on mine, it was downloaded back in August.  Hmmmm, glad mine is working 
as is.  lol

Dale

:-)  :-)

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!