1

I have an Intel S5520SC server board setup in a barebones test bench.

  • Board has 1x E5540 (an approved processor that at one time worked in this board)
  • Also tested on another E5540, issue persists
  • 2x 2GB RDIMMS (tested, supported and installed in the correct positions)
  • Windows 7 Pro x64 newly installed
  • A known good simple graphics card
  • Completely default BIOS settings
  • All firmware up to date
  • A more than capable PSU, both 24 and 8 pin connectors firmly connected
  • Ran the OS drive through a checkdisk and it passes

I've zeroed this board as far as I can, I've narrowed the issue down as far as I can. Yet for some reason I'm experiencing a periodic CPU spike (about once a second).

I believe something on the board is causing the interrupts (given that this issue started after the board was in storage), I just don't know how to proceed in my diagnostics.

So given a periodic CPU spike registering on resource manager as "system interrupts" how should I proceed in testing / what can I do?

2
  • which Operation system do you use? Feb 1 '14 at 6:26
  • OP states Windows 7 Pro x64
    – Dave M
    Feb 2 '14 at 19:24
0

OK this was the strangest thing ever. Thanks to this thread I was able to figure out the issue.

Seriously I've been working with computers for a LONG time and this, this was one of the strangest things to happen ever.

So here's the deal, if this happens to you, plug both of the Gigabit nics (network interfaces) into something, anything really and the interrupts just stop.

I'm going to have a look at the driver and see if it's in some kind of dual-gigabit mode, but for now this works.

0

The vmstat utility will show you the number of interrupts per second, not just the CPU usage. It is worth looking at it to make sure you actually have a problem with the number of interrupts and not just a driver taking more processing time than expected. If the data suggests a driver problem, unloading the modules one by one to see when the load stops seems like the fastest way to identify the culprit. Also take a look at /proc/interrupts - it is giving you a more detailed breakdown on the kind of interrupts happening.

Another thing to consider are power states - a CPU might be set into power-saving states where it would have significantly reduced processing abilities. Percentages would naturally go up as the cake of total processing time is shrinking. Use powertop to see what amount of time your CPU is running in different available C-states and at what frequencies. Disable PM (e.g. by setting maximum performance in BIOS setup) and see if this helps.

Last but not least, it may be a hardware problem with an extension card or your NIC - consider unplugging everything you can to see if this might be the case.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.