100% Cpu-ness

Im running an AMD System:
OS: Windows Vista Business X64 SP2
Proc: Amd Athlon x2 6400
MotherBoard: Asus M3N78-VM
Ram: G-Skills PC2 6400 6GB
Network: D-Link DWA 552
Video: Nvidia 9500GT 512MB

The problem is that at random times, mostly when I am not doing anything processor intensive (like iTunes, or Word Processing) the processor will spike from 7-13% to 100%..without me even doing anything..If i restart, or just even go into standby and then return, it settles down and goes to idle again, until it occurs again.

If I am gaming, or video/photo editing it runs fine..

I have full formatted and reinstalled..i have reinstalled a different OS..32bits and 64bits..updated all drivers, software, BIOS..I have tested the RAM and it passes..I even tried it with different ram..still same problem..I removed all add-on cards(wireless, FireWire, video) and it still occured..I even swapped the Athlon X2 6400 for an old Athlon X2 4000 and it still did it..

Could a MOBO be casuing this random spike...please help..I don't wanna go and get a new mobo and then have it be something simple/dumb..

PS: The problem occured before SP1, while SP1, and while SP2..i have spent about 6 months now trying to figure out the problem and am about to give up...other then that problem, the system runs flawlessly..its just that random spike..I can't leave the computer unattended..its like a child :/

Thanks

A few questions:

- Are you saying the processor is pegged at 100% and you cannot get it to come down without rebooting or going to standby?

- Does Task Manager show all processor cores to be affected (they're all individually at 100%) or just one?

- Have you ever noted any pattern with respect to which process is tagged as being responsible?

- Does the same issue occur while the machine is running in safe mode?

- Can you reproduce it in any way? (Is there a known trigger?)

- Are you saying the processor is pegged at 100% and you cannot get it to come down without rebooting or going to standby?
I cannot get it to come down unless I suspend it or shut it down some how..it will be 86-100% idle..and when i move the mouse its fine..but don't even think about clicking the start menu..itll take like a good 5 minutes..

- Does Task Manager show all processor cores to be affected (they're all individually at 100%) or just one?
The task manager shows both cores affected..it does not however attach the use to any process..but the system idle does go down to 5%-0%..but there is no process showing high usage..and the CPU core temps do remain idle..
<100F idle (37C) (when CPU stressed Max is ~145F(62C))
when i use process explorer it shows svchost as the suspect..but if i end svchost..another random process will jump..and so on..

- Have you ever noted any pattern with respect to which process is tagged as being responsible?
The pattern is..that there is no pattern :/..Close one, another one gets picked..most of the time..it is svchost or nothing at all..but i do no believe that is it

- Does the same issue occur while the machine is running in safe mode?
I have not tried it in safe mode..I will have to try and it and report back to u..

- Can you reproduce it in any way? (Is there a known trigger?)
The only way i know for sure to reproduce it, is to leave to computer alone running something simple for about 10 minutes..whether its leaving itunes open, or firefox..sometimes its just when im doing nothing..I have scanned with Malware Bytes Mcafee and spybot and avg..nothing was found..i do not suspect virus or spyware

OK, thanks for the clarifications.

It's still possible that the cause is software, in which case the safe mode test might provide an interesting additional data point, but I suspect you've got a hardware device periodically going nuts and issuing interrupts at a very high rate.

If you start PERFMON and add the following counters, you may see substantial increases during the problem period:

PROCESSOR\% Processor Time (so you can see the onset of the spike)
PROCESSOR\% Interrupt Time
PROCESSOR\DPC Rate
PROCESSOR\DPCs Queued/sec
PROCESSOR\Interrupts/sec

The thing about hardware interrupts is that they're handled in what's called an arbitrary thread context. Whichever thread/process happens to be using the processor at the time will be hijacked to deal with the hardware interrupt. Hence, an "interrupt storm" can make it look like various processes are responsible for the high processor usage - you kill one and another one takes over, just as you're describing.

If you watch the counters I've listed above during a normal (non-problem) period, you'll get a feel for a baseline value that's OK for your machine and your software environment. If those counters then skyrocket during the processor 100% plateau, that almost proves a hardware link.

Ok. I shall run her tonight and see if she gets a spike tonight in safe mode and then post back in the am. As well as run the perfmon and see if it does anything during the day.

Thanks so much for the help.

problem did not occur in safe mode from what I could see..but then again I couldn't do much in safe mode anyway..I tried to repeat the problem with my usual, play iTunes and leave it overnight, and when I woke up this morning..the cpu was still idle :/..I will get the perfmon data once the situation occurs again..

suprafanatic26 said:

problem did not occur in safe mode from what I could see..but then again I couldn't do much in safe mode anyway..I tried to repeat the problem with my usual, play iTunes and leave it overnight, and when I woke up this morning..the cpu was still idle :/..I will get the perfmon data once the situation occurs again..

Click to expand...

OK. It would be good if it turns out that the machine consistently behaves itself in safe mode because of the implication that an app or driver is part of the issue.

If you're comfortable with Perfmon, it may be useful to log those counters to a file instead of just adding them to the graphing UI, although the latter is a good start too. By logging to a file and examining the counter graphs afterwards, you give yourself (and others) the chance to review the info more than once, plus you reduce the load at the time of the incident - it's marginally easier for perfmon to log to a file than to drag lines across the screen - and that can improve the accuracy of the data in edge conditions.

soooooooooooo..i have a dual monitor setup..i had a 19 wide and a 15..the inverter board for the 15 started being stupid..and the back light would cut in and out..a new inverter board would have been $80..so instead i just got a new monitor..the 15 had speakers under it..i had a headphone split on the audio out jack from the mobo..one feed to the monitor..one feed to an amplifier..when i unplugged the 15 from the audio jack to put the new monitor in (w/o speakers) the cpu jumped to 100%..so i went to standby and came back and all was good....when i went to go and plug the amplifier back into the jack..the cpu jumped up again!!..so i went to standby and came back...

all-in-all, since i put the new monitor, and only have one feed for the audio coming out..i have not had the problem..computer has been on for about 2days straight now..before i couldn't even leave it on 1 day without the cpu spike....could it have been a bad contact with the y-split that was driving the mobo crazy???

(if you notice, in vista if u unplug the audio device, it detects it, and puts a red "X" on the sound icon in the system tray)

suprafanatic26 said:

all-in-all, since i put the new monitor, and only have one feed for the audio coming out..i have not had the problem..computer has been on for about 2days straight now..before i couldn't even leave it on 1 day without the cpu spike....could it have been a bad contact with the y-split that was driving the mobo crazy???

Click to expand...

Your situation reminds me of this in some ways:

http://www.vistax64.com/general-discussion/219311-monitor-causes-strangely-high-cpu-usage.html

In short, yes, what you're describing is not unprecedented. Hardware problems can easily manifest themselves as high processor usage.

so it happened again >:-|

took a few days this time which was odd ..but anyway

perfmon reported DURING the spike

% Processor Time: 192mean
% Interrupt Time: 2 mean
DPC Rate: 9mean
DPCs Queued/sec: 350 mean
Interrupts/sec: 4,234 mean

weird thing is..perfmon won't give me reports for the processor when its not in a spike
it won't even give me that section...

inspike:

noSpike:

perfmon is reporting IN SPIKE that it is stuck between 89-100% cpu(when I am not doing anything), but under the processes..the highest is side bar at 22%..then rundll at 13%..and the rest are <5%..with over 90% of them being below .5%..

i am pretty sure it is hardware