Massive stutters, spikes, lag, crashes, maybe related to frametimes...

Nova Swing

Nova Swing

Nova Swing

Associate
Joined
29 Mar 2012
Posts
91
Hi all, posting in general hardware because I've been unable to work out exactly what component is causing my issue but hoping somebody can help...

Since Thursday last week, any game I play stutters and lags like mad, to the point where it will usually gives up the ghost and crash. The symptoms have been different, FPS dips down to 1 and then perhaps ok for a few seconds before a dip down again. Not always a slow as 1 but sometimes slower - like 3 or 4 frozen frames over 10 secs or so.

Symptoms changed last night after a while and now I'm getting micro-stutters on moving the mouse, which makes games just as unplayable. Funny thing is, using AWDS to move my player around doesn't seem to cause the stutter but maybe that's slower movement so not stressing whatever the problem is.

I tried changing mouse USB port but is fine. 99.99% sure it's not the mouse, it's a few months old M65 and besides, I'm getting all the other symptoms. I also tried a keyboard with a mouse pad and same stuttering.

Hardware monitor shows lots of fluctuations in CPU temp and voltage on idle and sometimes in game. However, nothing is overheating at all in game or on desktop. All around 50 - 60 degrees max.

Playing R6: Seige last night, I had the more jerky FPS issue to start, then for some strange reason it went away for about an hour and was perfect. Then after a restart (it hung when alt-tabbing) the micro stuttering on mouse movement started. I had MSI Afterburner overlay on and all figures looked fine when the stuttering is happening (all the time at this point)... except for frametimes. On the AB log frametimes are jumping to 1000 (i believe that's as far as it logs, could be 5000 for all I know).

CPU utilisation is highest on core one - around 80% with other cores on a rough average of 40-60%.

The issues are not game sepecifc, it all happens in BF1 and Just Cause 3 too.

My spec:

Win 10, Intel I7 2600K, Gigabyte 1080 G1, 8 gb ram - all stock speeds, OCz 650 w PSU. Dell SG2716dg monitor running at 1440p with gsync @ 144hz.

PC is about 5 years old I think, was originally an OCUK Battlefield Commander model.

All games have been buttery smooth at very high settings until this. Now all games are unplayable.

I'm wondering if power supply is on it's way out, but at the moment I really don't know what's causing it.

Or maybe CPU has finally given up the ghost. Poor thing has had a hammering.

No hardware changes to make it happen from Thursday. On Friday I thought I'd let the Creator's Update happen to see if it would clear it. It didn't so have since rolled back. I checked PhysX to be on the GPU also.

Perhaps it could be some g-sync, v-sync, refresh rate thing. But nothing changed to make this happen, apart from turning deep sleep off on the monitor. Can't see why that would be an issue.

Really sorry for such a long post but just trying to get everything down so anybody who might be able to help has has much info as possible.

So, if anybody still reading at this point has any thoughts or suggestions of what the problem might be, or how to narrow it down, it would be much appreciated.

One final thing to add, desktop, web browsing is still pretty ok, but something feels a bit off. So I think the issue is always there but shows itself much more when something is under heavy load.

Thanks!
 
Nova Swing said:
Hardware monitor shows lots of fluctuations in CPU temp and voltage on idle and sometimes in game. However, nothing is overheating at all in game or on desktop. All around 50 - 60 degrees max.
Click to expand...
I'm sure you've checked, but it's the one thing you haven't mentioned, what are your GFX temps when the stuttering occurs?
 
Can you run any benchmarks?
Try something cpu specific like x264 and then something gpu related like 3d mark.

If it's not temperature related it might be Windows or drivers.

If you have a spare hdd I would try a quick reinstall.

From the symptoms I would suspect memory first, but it could be anything :)
 
Plec said:
I'm sure you've checked, but it's the one thing you haven't mentioned, what are your GFX temps when the stuttering occurs?
Click to expand...

GPU temp is fine, around 65 - 70 from recall last night. No temps in HW monitor or MSI AB are cause for concern at all. Although I am sure that the CPU voltage and temps didn't used to fluctuate at idle like they are now. Not massive fluctuations, say between 35-40 degrees and just a few points on voltage - I think but could be wrong it was going between say, 1.18 to 1.23. that is avery rough description but can look tonight if it's important.

Will try the benchmarks and see what happens.

No spare HDD but my game libraries are split between OS SSD and an HDD. Perhaps I should test a game on the HDD to see if that makes a difference.

Thanks
 
Have you checked if Event Monitor shows anything for those stuttering moments?
You could also use LatencyMon to check for possible hints of driver problems.
http://www.resplendence.com/latencymon

Unless you've overvolted it CPU is very unlikely to broke.
And likely results would be major errors/total crashes instead of slow downs.

But that PSU is definitely in suspect list.
I don't remember if OCZ has ever had PSUs with high quality capacitors.
In general they've spend money on bling bling instead of quality.
With precise model name/number it would be easy to check how cheap parts they used in it.
 
EsaT said:
Have you checked if Event Monitor shows anything for those stuttering moments?
You could also use LatencyMon to check for possible hints of driver problems.
http://www.resplendence.com/latencymon

Unless you've overvolted it CPU is very unlikely to broke.
And likely results would be major errors/total crashes instead of slow downs.

But that PSU is definitely in suspect list.
I don't remember if OCZ has ever had PSUs with high quality capacitors.
In general they've spend money on bling bling instead of quality.
With precise model name/number it would be easy to check how cheap parts they used in it.
Click to expand...

It's an OCZ ZS Series 650W if that rings any bells.

I'm wondering if I should just get a new PSU as if it is faulty I think it can taker other components with it.

Re CPU, the only overclock it's ever had is the OCUK one, up tyo 4.2. Never played with voltage myself so don't think it could have been borked that way.

Haven't had chance to try the benchmarks or latecy tests tonight but will have a go tomorow.
 
Nova Swing said:
Win 10, Intel I7 2600K, Gigabyte 1080 G1, 8 gb ram - all stock speeds, OCz 650 w PSU. Dell SG2716dg monitor running at 1440p with gsync @ 144hz.
Click to expand...
Nova Swing said:
Re CPU, the only overclock it's ever had is the OCUK one, up tyo 4.2. Never played with voltage myself so don't think it could have been borked that way.
Click to expand...
Just to clarify - is the CPU at manufacturer's stock or at the 'stock clock' that OcUK originally set?

If it still has the OcUK clock this may have become unstable (not unusual) - and would need to be removed to rule it out as the problem.
 
That (I think OCZ's last PSU serie before their bankruptcy) is their only PSU line which had quality caps so from that point it shouldn't die fast.
Though otherwise it was real basic old design platform from not so top tier maker Sirfa and who knows if those caps were properly sized to last long.

So without either finding clear cause from elsewhere or trying other PSU we can't say anything to one way or another about its guilt.



Plec said:
If it still has the OcUK clock this may have become unstable (not unusual) - and would need to be removed to rule it out as the problem.
Click to expand...
That overclocking likely involves some amount of overvolting from default so depending on motherboard VRM might be starting to wear from stress.
People have had framerate/stutter problems with power hog Bulldozer CPUs overheating VRMs causing throttling.


EVGA's non-reference 1070/1080 cards had once VRM overheating issues because custom cooler not cooling VRM enough (some VRMs ending up going up in flames) but Gigabyte doesn't seem to have had those.
And anyway GPU's VRM should surely handle desktop/web browing use if it tolerates gaming at all.
 
Have you got the Windows 10 Creators Update? If so, disable Game Mode. Also check your Power profile is set to Performance.
 
Sounds like it's worth saving your bios settings and removing the overclock to check that's not the issue as it will only take a few minutes.
 
Not currently using the OCUK overclock. Had a few problems a year or so ago, when it started to apply the 4.2 clock permanently rather than boosting to it, so I just disabled it. Everything currently at stock and has been for a long time.

It's not a creators update issue, it started before I updated. But have since rolled back for troubleshooting purposes so as not to introduce a new variable.

Weird last night - all the problems in R6 Seige but straight after BF1 ran perfectly. I only had a little time with PC though so didn't get chance to switch back to R6 to see how that was running. Hopefully will get more time tonight for this... what a pain, but thanks for help and suggestions so far.
 
I had the same issue after some motherboard drivers failed. In the end it's a very fast job to back up files and clean install windows 10... worked fine afterwards with the same BIOS settings.

Instead of hours/days of messing with all the hardware highly recommend it. Back up files, note all the programs you need, download all those installers on a drive ready, wipe the OS drive and install clean.
 
Sounds exactly like what I had before I bought a better AIO cooler. Your CPU is thermally throttling. Is your core cpu temperatures reaching 91 Degrees C at all?
 
Just reporting in, things magically seemed to be back to normal last night and so far tonight it seems ok. It doesn't feel quite right though and I am expecting it to happen again.

I looked deeper into the Afterburner logs and though I don't really understand it myself. When playing a game, the massive frametime spikes seem to coincide with a Power % drop from a steady approx 90 to approx 20.

GPU usage drops from high 90s to mid teens.

FB usage drops from around 50 to low teens.

Core clock Mhz (GPU) drops from 1936 to 1202 for every event. No variations always those two figures (at least on this particular log).

Fluctuations in memory usage (MB) from a level of around steady 2800 down to the 1200s. This goes up and down within the event but maybe the measurement is just more granular than the others.

Voltage limit drops from 1 to 0.

No load limit spikes from 0 to 1.

I believe these are all GPU measurements. Anything I haven't mentioned stays pretty steady and doesn't seem to change when the frametimes spike.

For CPU measurements, the following happens when frametimes spike:

All eight CPU temps drop from a steady 60 low 50s.

All eight CPU usage % drop from between 80-high 90s to anywhere between 20 and 40.

Just wondering if any of that would shed any light onto what mighty be happening.

Guess I could try to load the file or a screen shot if it helps but don't really know how to do that.

Again, thanks for all responses so far. Although it seems ok now, I'm still feeling like I might buy a new PSU as I don't really want to risk hurting my 2 months old 1080!
 
Nova Swing said:
Just reporting in, things magically seemed to be back to normal last night and so far tonight it seems ok. It doesn't feel quite right though and I am expecting it to happen again.

I looked deeper into the Afterburner logs and though I don't really understand it myself. When playing a game, the massive frametime spikes seem to coincide with a Power % drop from a steady approx 90 to approx 20.

GPU usage drops from high 90s to mid teens.

FB usage drops from around 50 to low teens.

Core clock Mhz (GPU) drops from 1936 to 1202 for every event. No variations always those two figures (at least on this particular log).

Fluctuations in memory usage (MB) from a level of around steady 2800 down to the 1200s. This goes up and down within the event but maybe the measurement is just more granular than the others.

Voltage limit drops from 1 to 0.

No load limit spikes from 0 to 1.

I believe these are all GPU measurements. Anything I haven't mentioned stays pretty steady and doesn't seem to change when the frametimes spike.

For CPU measurements, the following happens when frametimes spike:

All eight CPU temps drop from a steady 60 low 50s.

All eight CPU usage % drop from between 80-high 90s to anywhere between 20 and 40.

Just wondering if any of that would shed any light onto what mighty be happening.

Guess I could try to load the file or a screen shot if it helps but don't really know how to do that.

Again, thanks for all responses so far. Although it seems ok now, I'm still feeling like I might buy a new PSU as I don't really want to risk hurting my 2 months old 1080!
Click to expand...

Sounds like exactly what happens when the cpu overheats - it downclocks to the base 1200Mhz frequency. Except you are saying your core temps are all at 50.

Maybe faulty temperature sensors? Maybe the actual temperature is higher than it is saying it is? Have you tried touching the heatsink on the cpu and/or if you have watercooling checking the temperature of the coolant?

Also - windows power settings - are they set to balanced? What about other settings in the power options?

Also, I would advise against using performance and instead use Balanced. Balanced will allow the cpu to downclock when idle and will save a lot of temperature and wear and tear at running at full pelt for no reason. You could check / try it to see what happens though.

Have you tried a system stability test by running OCCT at full test for a couple of minutes?
 
MarcusX said:
Also, I would advise against using performance and instead use Balanced. Balanced will allow the cpu to downclock when idle and will save a lot of temperature and wear and tear at running at full pelt for no reason. You could check / try it to see what happens though.
Click to expand...
Agreed, it's not an ideal setting but the balanced setting can sometimes cause performance issue like this (replicating throttling - which admittedly it could still be if hes sensors are borked) - and setting it to performance can sometimes resolve the issue.

Not the ideal but needs to be tested and ruled out - and needs must etc..
 
You must log in or register to reply here.
Back
Top Bottom