Tracing hardware interrupts and DPC with xperf

Message Topic Search Topic Options Post Reply Create New Topic Printable Version

   Hey guys
Recently, my quad core AMD started having phantom CPU usage (doesn't show up in any discernible process in task manager).
After a little digging, I found that process explorer DOES show the missing cpu usage (~30%)
but its source is hardware interrupts and DPC.

Now, I'm trying to trace what is generating the interrupts, but I'm not having any luck.

I have run xperf (after much trouble trying to find out how to get it installed),
with the simple command line of:

xperf -on DiagEasy+PROFILE -stackwalk Profile

when I open the merged.etl file, I just get a screen of graphs, and can't seem to drill down to any details with regard to DPC/interrupts.  I don't know whether I'm using the viewer wrong, or if I'm actually not capturing the specific info I need.

Any help much appreciated!  thanks

Author	Message Topic Search Topic Options Post Reply Create New Topic Printable Version
bluenote Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 November 2009 Online Status: Offline Posts: 3	Quote Reply Topic: Tracing hardware interrupts and DPC with xperf Posted: 03 November 2009 at 11:26pm
	Hey guys Recently, my quad core AMD started having phantom CPU usage (doesn't show up in any discernible process in task manager). After a little digging, I found that process explorer DOES show the missing cpu usage (~30%) but its source is hardware interrupts and DPC. Now, I'm trying to trace what is generating the interrupts, but I'm not having any luck. I have run xperf (after much trouble trying to find out how to get it installed), with the simple command line of: xperf -on DiagEasy+PROFILE -stackwalk Profile when I open the merged.etl file, I just get a screen of graphs, and can't seem to drill down to any details with regard to DPC/interrupts. I don't know whether I'm using the viewer wrong, or if I'm actually not capturing the specific info I need. Any help much appreciated! thanks

bluenote Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 November 2009 Online Status: Offline Posts: 3	Quote Reply Posted: 03 November 2009 at 11:37pm
	Ok, I managed to figure out that the summary table has the individual data labeled. For interrupt CPU usage graph, the data is like this: 1. acpi.sys (address) 316,551 2. Unknown (3 entries, of which 2 are miniscule) - 164,006 3. hal.dll 55,430 Leading me to believe that my problem is associated with acpi and/or with the unknown entry. How could I translate the memory address into a program/driver/file ? thanks EDIT: And how do I go about troubleshooting something as intangible as ACPI? Edited by bluenote - 03 November 2009 at 11:39pm

molotov Members Profile Send Private Message Find Members Posts Visit Members Homepage Add to Buddy List Moderator Group Joined: 04 October 2006 Online Status: Offline Posts: 17287	Quote Reply Posted: 04 November 2009 at 4:14am
	Hi bluenote, What OS are you using? The unknown entries are curious... What arch are you using - x86, or x64? Do you have Daemon Tools installed?
	Daily affirmation: net helpmsg 4006

bluenote Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 November 2009 Online Status: Offline Posts: 3	Quote Reply Posted: 06 November 2009 at 9:24pm
	Hey molotov I'm using vista x64, and I do have daemon tools installed (but not running, to my knowledge), however daemon tools pre-dated the problem by at least a couple months. What I've done for now which has eliminated the high CPU usage, is disabled ACPI aware-OS support in the bios,although this isn't ideal from my perspective, I don't feel like there is much else that is likely. (although, I will uninstall daemon tools just to confirm). thanks