Home » The PARIS Forums » PARIS: Main » Hyperthreading/Dual CPUs and the XP Driver
| Hyperthreading/Dual CPUs and the XP Driver [message #64646] |
Fri, 17 February 2006 14:50  |
Mike Audet
Messages: 294
Registered: December 2008
|
Senior Member |
|
|
Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file to ressolve
the c16 issues with Windows XP?
The switch tells Windows to rout all interupt requests through one CPU, instead
of spreading them out over two. It definitly solves some driver problems
in dual CPU systems.
I ran accross it trying to resolve a problem with a USB card on my Dual Athlon.
I don't have any c16s to try it with, but my testing seems to show that
it puts about 10% more load on the highest numbered CPU, which is a small
price to pay if it solves a driver issue, especially in a dual core or cpu
situation.
Just wondering....
Mike
|
|
|
|
|
|
|
|
| Re: Hyperthreading/Dual CPUs and the XP Driver [message #64668 is a reply to message #64663] |
Sat, 18 February 2006 22:48  |
Chris Ludwig
Messages: 868
Registered: May 2006
|
Senior Member |
|
|
HI,
If you r feeling real frisky you can try the Interrupt Affinity Tool in
the win2003 resource kit. You will be able to set individual pieces of
hardware to use single processors. Been fooling around with UADs on
Nvidia dual core machines it see if it a makes ant difference.
http://www.microsoft.com/downloads/details.aspx?familyid=9d4 67a69-57ff-4ae7-96ee-b18c4790cffd&displaylang=en
Chris
Mike Audet wrote:
> This flag is interesting because the system still uses both CPUs, just not
> for interupts. You still get most of the advantage of a dual CPU system,
> but the PARIS driver might work properly.
>
> Let's hope!
>
> Mike
>
>
> "Rohde Wakefield" <rohde @ iname.com> wrote:
>
>>Hi Mike.
>>
>>I ran into this flag a month ago but haven't had a chance to test it during
>
>
>>a session. I actually ran my last session (recording scratch tracks of my
>
>
>>wife) without the C16 - just forgot to hook it up and never took the time
>
> to
>
>>shutdown PARIS once we got rolling.
>>
>>When I found the switch, I also tried another switch (/onecpu). I'm about
>
>
>>80% sure one of these does the trick. That night I tried it and played
>>through a project, trying fast moves on the C-16 - didn't see the problems
>
> I
>
>>had before of the transport locking up.
>>
>>Hopefully someone using the system a bit more fulltime will have a chance
>
> to
>
>>try it out.
>>
>>-Rohde
>>
>>
>>"Mike Audet" <mike@mike.....> wrote in message news:43f64525$1@linux...
>>
>>>Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file
>
> to
>
>>>ressolve
>>>the c16 issues with Windows XP?
>>>
>>>The switch tells Windows to rout all interupt requests through one CPU,
>
>
>>>instead
>>>of spreading them out over two. It definitly solves some driver problems
>>>in dual CPU systems.
>>>
>>>I ran accross it trying to resolve a problem with a USB card on my Dual
>
>
>>>Athlon.
>>>I don't have any c16s to try it with, but my testing seems to show that
>>>it puts about 10% more load on the highest numbered CPU, which is a small
>>>price to pay if it solves a driver issue, especially in a dual core or
>
> cpu
>
>>>situation.
>>>
>>>Just wondering....
>>>
>>>Mike
>>
>>
>
--
Chris Ludwig
ADK
chrisl@adkproaudio.com <mailto:chrisl@adkproaudio.com>
www.adkproaudio.com <http://www.adkproaudio.com/>
(859) 635-5762
|
|
|
|
| Re: Hyperthreading/Dual CPUs and the XP Driver [message #64838 is a reply to message #64646] |
Thu, 23 February 2006 07:17 |
Rohde Wakefield
Messages: 11
Registered: December 2005
|
Junior Member |
|
|
OK - did some more testing yesterday. Again, this is an AMD X2 3800+ dual
core processor.
With just /intaffinity, the transport would be locked up almost immediately.
I could never use the C16 for anything.
With /intaffinity and /onecpu (/intaffinity probably unnecessary), it works
fine with no lockups.
For me, this is acceptable as I can reboot and do my video editing work with
both cores engaged. I don't need more horsepower than one core for Paris.
It seems like there was a way to force which processor was tasked to handle
the interrupts and its also possible to set the application to just one
processor. This may be another possibility that I'll look into later.
-Rohde
"Rohde Wakefield" <rohde @ iname.com> wrote in message news:...
> Hi Mike.
>
> I ran into this flag a month ago but haven't had a chance to test it
> during a session. I actually ran my last session (recording scratch tracks
> of my wife) without the C16 - just forgot to hook it up and never took the
> time to shutdown PARIS once we got rolling.
>
> When I found the switch, I also tried another switch (/onecpu). I'm about
> 80% sure one of these does the trick. That night I tried it and played
> through a project, trying fast moves on the C-16 - didn't see the problems
> I had before of the transport locking up.
>
> Hopefully someone using the system a bit more fulltime will have a chance
> to try it out.
>
> -Rohde
>
>
> "Mike Audet" <mike@mike.....> wrote in message news:43f64525$1@linux...
>>
>> Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file to
>> ressolve
>> the c16 issues with Windows XP?
>>
>> The switch tells Windows to rout all interupt requests through one CPU,
>> instead
>> of spreading them out over two. It definitly solves some driver problems
>> in dual CPU systems.
>>
>> I ran accross it trying to resolve a problem with a USB card on my Dual
>> Athlon.
>> I don't have any c16s to try it with, but my testing seems to show that
>> it puts about 10% more load on the highest numbered CPU, which is a small
>> price to pay if it solves a driver issue, especially in a dual core or
>> cpu
>> situation.
>>
>> Just wondering....
>>
>> Mike
>
>
|
|
|
|
| Re: Hyperthreading/Dual CPUs and the XP Driver [message #64877 is a reply to message #64838] |
Thu, 23 February 2006 19:31 |
Mike Audet
Messages: 294
Registered: December 2008
|
Senior Member |
|
|
Thanks for trying and also for letting us know how it went, Rohde. I'm glad
it's working for you!
All the best,
Mike
"Rohde Wakefield" <rohde @ iname.com> wrote:
>OK - did some more testing yesterday. Again, this is an AMD X2 3800+ dual
>core processor.
>
>With just /intaffinity, the transport would be locked up almost immediately.
>I could never use the C16 for anything.
>
>With /intaffinity and /onecpu (/intaffinity probably unnecessary), it works
>fine with no lockups.
>
>For me, this is acceptable as I can reboot and do my video editing work
with
>both cores engaged. I don't need more horsepower than one core for Paris.
>
>It seems like there was a way to force which processor was tasked to handle
>the interrupts and its also possible to set the application to just one
>processor. This may be another possibility that I'll look into later.
>
>-Rohde
>
>"Rohde Wakefield" <rohde @ iname.com> wrote in message news:...
>> Hi Mike.
>>
>> I ran into this flag a month ago but haven't had a chance to test it
>> during a session. I actually ran my last session (recording scratch tracks
>> of my wife) without the C16 - just forgot to hook it up and never took
the
>> time to shutdown PARIS once we got rolling.
>>
>> When I found the switch, I also tried another switch (/onecpu). I'm about
>> 80% sure one of these does the trick. That night I tried it and played
>> through a project, trying fast moves on the C-16 - didn't see the problems
>> I had before of the transport locking up.
>>
>> Hopefully someone using the system a bit more fulltime will have a chance
>> to try it out.
>>
>> -Rohde
>>
>>
>> "Mike Audet" <mike@mike.....> wrote in message news:43f64525$1@linux...
>>>
>>> Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file
to
>>> ressolve
>>> the c16 issues with Windows XP?
>>>
>>> The switch tells Windows to rout all interupt requests through one CPU,
>>> instead
>>> of spreading them out over two. It definitly solves some driver problems
>>> in dual CPU systems.
>>>
>>> I ran accross it trying to resolve a problem with a USB card on my Dual
>>> Athlon.
>>> I don't have any c16s to try it with, but my testing seems to show that
>>> it puts about 10% more load on the highest numbered CPU, which is a small
>>> price to pay if it solves a driver issue, especially in a dual core or
>>> cpu
>>> situation.
>>>
>>> Just wondering....
>>>
>>> Mike
>>
>>
>
>
|
|
|
|
| Re: Hyperthreading/Dual CPUs and the XP Driver [message #64914 is a reply to message #64877] |
Fri, 24 February 2006 14:36 |
Rohde Wakefield
Messages: 11
Registered: December 2005
|
Junior Member |
|
|
Just did a little more testing.
With just /intaffinity set, the system goes considerably longer if the
affinity for the Paris process is set to the last processor. However, I
would eventually get a locked transport.
Seems /onecpu is the way to go (/intaffinity tossed in for good measure).
-Rohde
"Mike Audet" <mike@mikeF-SPAMaudet.com> wrote in message
news:43fe6fe4$1@linux...
>
> Thanks for trying and also for letting us know how it went, Rohde. I'm
> glad
> it's working for you!
>
> All the best,
>
> Mike
>
>
> "Rohde Wakefield" <rohde @ iname.com> wrote:
>>OK - did some more testing yesterday. Again, this is an AMD X2 3800+ dual
>
>>core processor.
>>
>>With just /intaffinity, the transport would be locked up almost
>>immediately.
>
>>I could never use the C16 for anything.
>>
>>With /intaffinity and /onecpu (/intaffinity probably unnecessary), it
>>works
>
>>fine with no lockups.
>>
>>For me, this is acceptable as I can reboot and do my video editing work
> with
>>both cores engaged. I don't need more horsepower than one core for Paris.
>>
>>It seems like there was a way to force which processor was tasked to
>>handle
>
>>the interrupts and its also possible to set the application to just one
>
>>processor. This may be another possibility that I'll look into later.
>>
>>-Rohde
>>
>>"Rohde Wakefield" <rohde @ iname.com> wrote in message news:...
>>> Hi Mike.
>>>
>>> I ran into this flag a month ago but haven't had a chance to test it
>>> during a session. I actually ran my last session (recording scratch
>>> tracks
>
>>> of my wife) without the C16 - just forgot to hook it up and never took
> the
>>> time to shutdown PARIS once we got rolling.
>>>
>>> When I found the switch, I also tried another switch (/onecpu). I'm
>>> about
>
>>> 80% sure one of these does the trick. That night I tried it and played
>
>>> through a project, trying fast moves on the C-16 - didn't see the
>>> problems
>
>>> I had before of the transport locking up.
>>>
>>> Hopefully someone using the system a bit more fulltime will have a
>>> chance
>
>>> to try it out.
>>>
>>> -Rohde
>>>
>>>
>>> "Mike Audet" <mike@mike.....> wrote in message news:43f64525$1@linux...
>>>>
>>>> Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file
> to
>>>> ressolve
>>>> the c16 issues with Windows XP?
>>>>
>>>> The switch tells Windows to rout all interupt requests through one CPU,
>
>>>> instead
>>>> of spreading them out over two. It definitly solves some driver
>>>> problems
>>>> in dual CPU systems.
>>>>
>>>> I ran accross it trying to resolve a problem with a USB card on my Dual
>
>>>> Athlon.
>>>> I don't have any c16s to try it with, but my testing seems to show that
>>>> it puts about 10% more load on the highest numbered CPU, which is a
>>>> small
>>>> price to pay if it solves a driver issue, especially in a dual core or
>
>>>> cpu
>>>> situation.
>>>>
>>>> Just wondering....
>>>>
>>>> Mike
>>>
>>>
>>
>>
>
|
|
|
|
| Re: Hyperthreading/Dual CPUs and the XP Driver [message #65981 is a reply to message #64838] |
Thu, 30 March 2006 22:16 |
Aaron Allen
Messages: 1988
Registered: May 2008
|
Senior Member |
|
|
Rohde, did you get to dive into the IRQ splitting yet per chance? I think
this could show some nice promise!
AA
"Rohde Wakefield" <rohde @ iname.com> wrote:
>OK - did some more testing yesterday. Again, this is an AMD X2 3800+ dual
>core processor.
>
>With just /intaffinity, the transport would be locked up almost immediately.
>I could never use the C16 for anything.
>
>With /intaffinity and /onecpu (/intaffinity probably unnecessary), it works
>fine with no lockups.
>
>For me, this is acceptable as I can reboot and do my video editing work
with
>both cores engaged. I don't need more horsepower than one core for Paris.
>
>It seems like there was a way to force which processor was tasked to handle
>the interrupts and its also possible to set the application to just one
>processor. This may be another possibility that I'll look into later.
>
>-Rohde
>
>"Rohde Wakefield" <rohde @ iname.com> wrote in message news:...
>> Hi Mike.
>>
>> I ran into this flag a month ago but haven't had a chance to test it
>> during a session. I actually ran my last session (recording scratch tracks
>> of my wife) without the C16 - just forgot to hook it up and never took
the
>> time to shutdown PARIS once we got rolling.
>>
>> When I found the switch, I also tried another switch (/onecpu). I'm about
>> 80% sure one of these does the trick. That night I tried it and played
>> through a project, trying fast moves on the C-16 - didn't see the problems
>> I had before of the transport locking up.
>>
>> Hopefully someone using the system a bit more fulltime will have a chance
>> to try it out.
>>
>> -Rohde
>>
>>
>> "Mike Audet" <mike@mike.....> wrote in message news:43f64525$1@linux...
>>>
>>> Has anyone tried using the "/INTAFFINITY" switch in the boot.ini file
to
>>> ressolve
>>> the c16 issues with Windows XP?
>>>
>>> The switch tells Windows to rout all interupt requests through one CPU,
>>> instead
>>> of spreading them out over two. It definitly solves some driver problems
>>> in dual CPU systems.
>>>
>>> I ran accross it trying to resolve a problem with a USB card on my Dual
>>> Athlon.
>>> I don't have any c16s to try it with, but my testing seems to show that
>>> it puts about 10% more load on the highest numbered CPU, which is a small
>>> price to pay if it solves a driver issue, especially in a dual core or
>>> cpu
>>> situation.
>>>
>>> Just wondering....
>>>
>>> Mike
>>
>>
>
>
|
|
|
|
Goto Forum:
Current Time: Sun Nov 24 14:58:25 PST 2024
Total time taken to generate the page: 0.01949 seconds
|
The PARIS Forums
Search
Help
Members
Register
Login
Home
