no SMP without ACPI? (amd64)

[Chad Leigh -- Shire.Net LLC]

On Sep 5, 2007, at 3:07 AM, Ted Mittelstaedt wrote:

>
>
>> -----Original Message-----
>> From: owner-freebsd-questions at freebsd.org
>> [mailto:owner-freebsd-questions at freebsd.org]On Behalf Of Kris  
>> Kennaway
>> Sent: Wednesday, September 05, 2007 1:57 AM
>> To: Chad Leigh -- Shire.Net LLC
>> Cc: User Questions
>> Subject: Re: no SMP without ACPI? (amd64)
>>
>
>> Note that this may well be hardware related: without acpi you are  
>> only
>> using one CPU, etc, so if one of the others is bad it will only fail
>> when you have ACPI enabled -- even if ACPI itself is not to blame.
>>
>
> Easily testable by running that Other Operating system on the thing
> which I would have expected Chad to have done.

Sorry to be MIA for the last day.

No, I did not test it with that OS which shall not be named.   
However, I did attempt to install Sol10 on it.  Sol10 also relies on  
ACPI btw.  Anyway, the same error happened a good way through the  
installation -- it rebooted itself during the high IO of installing  
the system onto the system array.  Sol10 uses a 32bit kernel for the  
installation and I think, but am not sure, that it only activates 1  
CPU core.

I do not think, however, that it is a bad CPU.  I can boot the system  
with ACPI and activate all 4 CPU cores and I can run CPU intensive  
things, so far, as long as they do not generate lots of IO, and  
processes run fine on all 4 CPUs and will run for hours.  However, if  
I do something that generates IOs, like a build world on the FreeBSD  
source, about 75% of the time it does not make it all the way through  
with the build.  It takes me about 45min on this system to do a "make  
buildworld" and it usually will happen 10-30 minutes into the build.   
25% of the time it will make it through 1 build world, and once or  
twice has made it through 2, but always fails in the same way on a  
subsequent retry of the build. (Ie, I just keep doing builds over and  
over until it fails, which is 75% of the time in the first build, 24%  
of the time in the second, and 1% of the time in the third, roughly  
speaking).

I also do not think it is the memory.  For one, that would show up  
even with ACPI disabled, and 2, I ran multiple complete passes of the  
standalone memtest86 program without error.

I do not think it is thermally related as the temperature never got  
very high and I had additional fans temporarily blowing or I also  
stuck it directly over my A/C floor register with the A/C on blowing  
cold air directly at the intake of the system.

I monitored the Areca system while it was building and its  
temperatures never got anywhere near the limits Areca sets for their  
boards.  (The Areca has its own ethernet port and I was directly  
attached to that from my powerbook during several tests when it failed).

This machine is now in production with only the 1 CPU core.  I cannot  
do extensive testing but I can get pieces of info or do a reboot with  
ACPI if there are data items for me to grab if people want to take a  
look at what it is doing etc.

Thanks!
Chad

---
Chad Leigh -- Shire.Net LLC
Your Web App and Email hosting provider
chad at shire.net