Troubleshooting heavy DPC (Deferred Procedure Calls) use?

[BillW50]

Yesterday on this netbook, Process Explorer showed that DPC was eating
50% or more CPU time. The machine was really slow and sluggish. I
cheated and restored from an earlier backup and all is well again.
Although I did also saved a backup of it acting up all of the time. So I
would like to try to troubleshoot it if for nothing else for educational
purposes. But I don't know how to troubleshoot DPC problems. I believe a
Windows update, touchpad driver update, or something else may have
caused the original problem. Any ideas?

--
Bill
Asus EEE PC 702G8 ~ 2GB RAM ~ 16GB-SDHC
Windows XP SP2

[Jose]

On Oct 1, 11:16*am, "BillW50" > wrote:
> Yesterday on this netbook, Process Explorer showed that DPC was eating
> 50% or more CPU time. The machine was really slow and sluggish. I
> cheated and restored from an earlier backup and all is well again.
> Although I did also saved a backup of it acting up all of the time. So I
> would like to try to troubleshoot it if for nothing else for educational
> purposes. But I don't know how to troubleshoot DPC problems. I believe a
> Windows update, touchpad driver update, or something else may have
> caused the original problem. Any ideas?
>
> --
> Bill
> Asus EEE PC 702G8 ~ 2GB RAM ~ 16GB-SDHC
> Windows XP SP2

Hard to troubleshoot when it is behaving.

Google result show that people are sometimes instantly happy with
manipulating their ACPI battery settings in Device Manager or merely
removing their battery temporarily - with some happy faces when the
miracle occurs. Sounds very ACPI/battery related, but I have never
seen this problem.

Other folks use Vista only tools to isolate it to a few specific driver
(s) and I don't know what XP equivalents might be.

[Paul]

BillW50 wrote:
> Yesterday on this netbook, Process Explorer showed that DPC was eating
> 50% or more CPU time. The machine was really slow and sluggish. I
> cheated and restored from an earlier backup and all is well again.
> Although I did also saved a backup of it acting up all of the time. So I
> would like to try to troubleshoot it if for nothing else for educational
> purposes. But I don't know how to troubleshoot DPC problems. I believe a
> Windows update, touchpad driver update, or something else may have
> caused the original problem. Any ideas?
>

A large number of DPCs, could come from a large number of hardware
interrupts. But that might not be the only mechanism.

http://en.wikipedia.org/wiki/Deferred_Procedure_Call

Another measure of "badness", is DPC Latency. That is a
measurement of the service time for a DPC. On some computers,
even when there are only a handful of DPCs being serviced, some
of them see a 10 millisecond delay, instead of the normal
hundreds of microseconds delay, in getting serviced.
Something that can cause a high latency, is this.

http://en.wikipedia.org/wiki/System_management_interrupt

The OS doesn't know it is happening, but the symptoms can be
observed. Sometimes a BIOS update fixes it. This tool can be
used, to detect a latency problem (caused by an invisible,
unscheduled activity, stealing the processor).

http://www.thesycon.de/deu/latency_check.shtml

In your case, I would have fired up the Performance plugin and
looked at interrupt counts. Your problem isn't latency, as
much as it is the sheer number of them.

Paul

[BillW50]

In
,
Jose typed on Thu, 1 Oct 2009 11:53:59 -0700 (PDT):
> On Oct 1, 11:16 am, "BillW50" > wrote:
>> Yesterday on this netbook, Process Explorer showed that DPC was
>> eating 50% or more CPU time. The machine was really slow and
>> sluggish. I cheated and restored from an earlier backup and all is
>> well again. Although I did also saved a backup of it acting up all
>> of the time. So I would like to try to troubleshoot it if for
>> nothing else for educational purposes. But I don't know how to
>> troubleshoot DPC problems. I believe a Windows update, touchpad
>> driver update, or something else may have caused the original
>> problem. Any ideas?
>
> Hard to troubleshoot when it is behaving.
>
> Google result show that people are sometimes instantly happy with
> manipulating their ACPI battery settings in Device Manager or merely
> removing their battery temporarily - with some happy faces when the
> miracle occurs. Sounds very ACPI/battery related, but I have never
> seen this problem.
>
> Other folks use Vista only tools to isolate it to a few specific
> driver (s) and I don't know what XP equivalents might be.

In ,
Paul typed on Thu, 01 Oct 2009 17:35:41 -0400:
>
> A large number of DPCs, could come from a large number of hardware
> interrupts. But that might not be the only mechanism.
>
> http://en.wikipedia.org/wiki/Deferred_Procedure_Call
>
> Another measure of "badness", is DPC Latency. That is a
> measurement of the service time for a DPC. On some computers,
> even when there are only a handful of DPCs being serviced, some
> of them see a 10 millisecond delay, instead of the normal
> hundreds of microseconds delay, in getting serviced.
> Something that can cause a high latency, is this.
>
> http://en.wikipedia.org/wiki/System_management_interrupt
>
> The OS doesn't know it is happening, but the symptoms can be
> observed. Sometimes a BIOS update fixes it. This tool can be
> used, to detect a latency problem (caused by an invisible,
> unscheduled activity, stealing the processor).
>
> http://www.thesycon.de/deu/latency_check.shtml
>
> In your case, I would have fired up the Performance plugin and
> looked at interrupt counts. Your problem isn't latency, as
> much as it is the sheer number of them.
>
> Paul

Thanks Jose and Paul for your very informative information! I believe I
nailed down what the problem is on this one machine. When you get 18
inches or closer to another machine with WiFi, DPCs use goes up
drastically.

--
Bill
Asus EEE PC 702G8 ~ 2GB RAM ~ 16GB-SDHC
Windows XP SP2