[Asterisk-Users] Re: hyperthreading and zaptel

Steven asterisk at tescogroup.com
Thu May 4 04:53:44 MST 2006


cat /proc/interrupts
           CPU0       CPU1       CPU2       CPU3
  0:        422          0          0   46196367    IO-APIC-edge  timer
  8:          0          0          0        155    IO-APIC-edge  rtc
  9:          0          0          0          0   IO-APIC-level  acpi
 14:     414685          0          0        241    IO-APIC-edge  ide0
169:          0          0          0          0   IO-APIC-level  uhci_hcd
177:          0          0          0          0   IO-APIC-level  uhci_hcd
185:          0          0          0          0   IO-APIC-level  uhci_hcd
193:          0          0          0       2105   IO-APIC-level  ehci_hcd
201:      61965          0          0       4612   IO-APIC-level  megaraid
209:          0          0          0   46015177   IO-APIC-level  wct4xxp
217:     399933          0          0        333   IO-APIC-level  eth0
NMI:   46196423   46196379   46196377   46196376
LOC:   46196579   46196300   46196579   46196294
ERR:          0
MIS:          0

The Interrupt addresses were the same with and without hyperthreading, just the number of CPUs was two.

Mark suggested that the binaries might have been HT optimized.
I did a quick search of the code and didn't find anything, but I am not exactly sure what the keyword might be for that.
I did recompile and install zaptel with hyperhtreading off (with no success) and that is the build I am using with it back on now.





-- 
-- 
Steven

http://www.glimasoutheast.org



"James Harper" <james.harper at bendigoit.com.au> wrote in message news:AEC6C66638C05B468B556EA548C1A77DCEC240 at trantor...
Turning on hyperthreading may have changed the way interrupts are
routed. Were you using the same kernel (eg SMP kernel even with
hyperthreading disabled)? The BIOS may have configured things
differently too if you disabled it there.

I'm not sure, but you may be able to keep hyperthreading on in the BIOS
and boot into a UP kernel and have the same net effect as having ht
disabled.

You mention you have looked at /proc/interrupts, are there any
differences between the interrupt numbers assigned in the ht enabled and
ht disabled cases?

When the kernel boots, it dumps some info about IRQ routing, compare
those.

Maybe post /proc/interrupts and the relevant bits of the kernel boot
logs here if you aren't sure, someone might be able to spot something
out of the ordinary.

James

> -----Original Message-----
> From: asterisk-users-bounces at lists.digium.com [mailto:asterisk-users-
> bounces at lists.digium.com] On Behalf Of Steven
> Sent: Thursday, 4 May 2006 09:18
> To: asterisk-users at lists.digium.com
> Subject: [Asterisk-Users] hyperthreading and zaptel
>
> My Dell 2800 Dual 64bit  Proc. machine came in with hyperthreading
> enabled. (they call it virtual processor??)
>
> I have been intending for a month to disable it.
>
> Tonight, I rebooted, turned it off, and let the system come up.
>
> zaptel loaded and asterisk loaded, but both of my t1s were red. (it is
a
> TE411P)
> /proc/interrupts looked OK, zttest gave OK numbers.
> I doublechecked all of the files in case I changed something else
> accidentally.
> I tried various combinations of unloading, loading the modules and
ztcfg,
> etc.
>
> Finally, I decided to turn hyperthreading back on, and everything is
back
> to normal.
>
> Unless there is somewhere in CentOS 4.3 that has the processor count
> hardcoded from the install, I am baffled by this.
>
>
>
>
>
>
> --
> --
> Steven
>
> http://www.glimasoutheast.org
>
>
>
>
>
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> Asterisk-Users mailing list
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users

_______________________________________________
--Bandwidth and Colocation provided by Easynews.com --

Asterisk-Users mailing list
To UNSUBSCRIBE or update options visit:
   http://lists.digium.com/mailman/listinfo/asterisk-users






More information about the asterisk-users mailing list