System "stutter" and slowdowns, related to high "System" process CPU usage

Associate
Joined
11 Dec 2006
Posts
1,091
Location
Lincoln
Hi all,

I've been having some very frustrating system issues for the past month or so and I am pretty much out of idea regarding how to narrow it down to a cause. Any ideas would be gratefully received.

For starters specs:
Intel 2600K @ 4.2
8Gb Corsair Dominator (1866)
ASUS P8P67 Pro
Crucial M4 64Gb SSD (OS)
OCZ Vertex 3 (Games)
WD Black 2Tb (Data)
780Ti (Gigabyte GHz edition)
ASUS Xonar DX

The issue is what I can only describe as severe stuttering, which varies from freezes lasting a second or 2 repeating every 10-20 seconds, up to freezes that last for 30 seconds with the system only unfreezing for a couple of second before locking up again, so bad that even getting to the restart menu option takes several minutes. It seems to mostly present when I'm running DCS World (DX9 based flight sim) but it does not go away when exiting the game, no matter how long the system is left running (the process isn’t left running). I can also play other games and use the system for hours without issue occurring.

When it happens stutters are accompanied by an increase in CPU usage on the “system” process (up to 10-15% CPU), when the CPU usage drops to normal levels the stutter ends.

This started presenting a month or so ago and was quite sporadic, I couldn't put a finger on the possible cause. Having recently upgraded to Win 8.1 I put it down to a Win 8 issue and coped in the hope that a windows updater would solve it, obviously it didn't. I've since updated all hardware drivers to the latest available, tried alternate drivers for my sound card, and also rolled back to all applicable graphics drivers. None of these helped so suspecting a drive failure I switched out the SSDs by adding the 64Gb Crucial as the OS drive in place of the 128 Vertex and taking the previous 256Gb OCZ Agility game drive out of the equation, this had no effect.

I then did the same driver reinstalls a further time, and shortly afterwards did a reinstall of Win 8.1, again no effect on the issue.
Last night I switched back to Win 7 with a full clean install, again installing the most recent drivers available for all hardware, and this evening the problem occurred again.

So my current thinking is that is must be a hardware issue, but I just can’t figure out what. SMART checks out ok on all driver, the only issue being a dodgy temp sensor in the OCZ Vertex reporting 128C. CPU and GPU temps are all fine, MEMTEST reports no problems, and running Prime 95 for several hours goes fine.

All my Googling suggests the “system” process high CPU usage is linked to either a driver or physical hardware issue, but I am out of ideas when it comes to narrowing down to exactly what. So any ideas? Hopefully I'm just missing something "obvious" but I'm totally stumped. Help!
 
Yeah tried that. Last night I removed each drive again one by one (restoring windows from image each time) and running the usual tests, no change.

I've just spent an hour this morning taking the whole system apart and reassembling, just in case there is a loose cable or mis-seated card somewhere that was causing it. We'll see what happens. Definitely not a easy one to fault find.
 
This sounds similar to an issue I was having, except System was using 80-100% of the CPU.

The issue turned out to be a dodgy SATA cable. I tried everything from trying different SATA ports, trying each HDD (I have two) in different ports / configurations / on their own / etc. The last thing I tried was swapping the SATA cable for a new one, and that fixed the issue.

Might give you something else to test if you didn't try new / different SATA cables - you didn't specify whether you tried this or not so thought I'd chip in.
 
Last edited:
Just an update for future reference etc.

Now back on Win 8.1 again and checking event viewer showed nvlddmkm event 14 & event 13 errors each time this fault occurred.

So it's now narrowed down to Nvidia video drivers. The fact that playing games other than DCS does not lead to this issue suggests to me that it isn't an actualy hardware fault, but just a software/driver issue.

Some more Goolge research points towards an issue with timeout detection and recovery, so I've used the recommended registry edit to change the timeout delay (to off). Hopefully that will fix the issue, only time and more testing will tell.
 
Back
Top Bottom