Help: Stuttering with a lot of games on GTX 1060. HUGE UPDATE: NVIDIA'S FAULT CONFIRMED.

[JFST]

JUMP TO THE POST WITH THE UPDATE: http://www.tomshardware.co.uk/answers/id-3443807/stuttering-lot-games-gtx-1060/page-2.html#19968622


Hi all,
Hope you can help me:

(Important note: i could get a refund on my card so i got it, but i am still really interested on knowing why is this happening so i can buy a new one.)

-Since i bought my EVGA GTX 1060 i have had random stuttering, lag, frames drops on a lot of games:

Spoiler

CS:GO, Killing Floor 2 (changing some things i have got a smoother framerate but still there), Solarix (first level with huge drops, second seems smooth), Black Mesa (mainly when loading enemies).

Also i had huge stuttering on games like The Division (Free weekends), Ghost Recon Wildlands Beta, Call of Duty Infinite Warfare, and Black ops 3 (Free weekends on both). On all these does not matter if it had highest posible settings or the lowest.

-I had no problems with:

Spoiler

DOOM demo, Steep Beta, Fallout 4 had some slowdowns when i was on a city (Steam free weekend) but were stable,...
I could play ARK and it only sutters when a new day starts.

Things i have tried:

Spoiler

-Uninstalled Geforce Experience: got a performance boost but still stutters.
-Several drivers "clean-installed" with DDU (only driver and Physx, but now i see Audio and 3D is installed too -Windows Update? anyway does not seem to affect-).
-On Nvidia Control Panel Set Maximun number of prerendered frames to 1: got rid of some stuttering, but not all.
-GPU and CPU temps are fine.
-Diabling all startup programs and all MSconfig services (no Microsoft ones, of course)
-MOST IMPORTANT: i installed my old ATI HD6870 and the stuttering on CSGO and Solarix is gone without a doubt and on Killing Floor 2 i think is better too but now i have some frame drops (no stuttering as i had with 1060, i am almost sure it is "stable" frame drops -bad optimization i guess-). Also i could play The Division (everything on low) in a pretty smooth way (usually below 30fps, but smooth, while the 1060 has stuttering even in low).

Info i have found:

Spoiler

-A lot of people are having almost the same problems with almost the same games (can list some more if you want). All of them with different specs (some with way better specs).
-Some people "fixed" it changing HDD for SDD, others adding more RAM; on others these didn't work.
-Some say that anything lower than 6XX series (unsure about 7XX) work without a problem.
-Someone said he had these problems, did a RMA and got rid of them without doing anything else (waiting the report of someone else that is going to do the same).
-On the threads i have read i have not found a lot of users with ATIs with these stuttering problems. Maybe because there are less ATIs than NVIDIAS? But still very few ATis.

Now let me explain:
Well, as i said maybe a first batch of cards-chips are faulty and some compensate it with an SSD and RAM, and others can't. Could it be?

Also i contacted NVIDIA for help and all they could say is that as my motherboard has PCIe 2.0, and Pascal "requires" (does it?) 3.0 that is all my problems from.
Asked someone and told me he has stuttering with 3.0 on the same games than me, fixed with an SSD.
Also searched for benchmarks and it should only be a slighty difference between 2.0 and 3.0 on a few games (but that should have lower performance, no stuttering, right?).

I contacted Ubisoft too, for The Division stuttering and they told me that it was the CPU, but again, i don't think this could cause stuttering, just a worse performance than others.
Also they told me that i should uninstall some programs (DAEMON Tools Lite, MSI Afterburner, Unigine Valley, Nero 10, MegaDownloader), i barely use them (never runing games) and i have disabled it on startup (task manager), so i think they should have no effect, am i wrong? (as i don't use them i think i can uninstall them and install them if i need them -one of them has never been used-).



So,
What do i do?
Do you know any test i could do to see if the GPU is faulty or if any other thing is causing this?
A lot of 1XXX are having problems without a cause?
Should I ask EVGA too?
Should i RMA (or buy the same model)? could newer cards be faulty too?
Do AMD-ATIs have these problems too? If they do why are there so little posts about them? if they don't have stuttering problems (unless faulty) what could be a quick recomendation for an ATI card? (will probably create a thread to find out anyway 😉 ).








Hope you can help me.

Thank you very much in advance.




My specs:
- Gigabyte GA-X58A-UD3R (DDR3-LGA 1366)
- Intel i7 930
- RAM G.Skill 3x2GB DDR3 1600 (F3 12800 CL 7-8-7-24 1.5v)
- Cooler Master Hyper 212 Plus
- Lancool PC-K60 Dragonlord
- Seasonic X-650 Modular SS-650KM (80 Plus Hybrid-silent)
- Western Digital Caviar Black 1TB FAEX
- EVGA GTX 1060 SC 6GB (now: ATI HD6870)
- Windows 10


P.S. I accept advices on this OP if you think it can be done more clear to read 😉 .

 

[helpstar]

Check again to be sure that it´s not your system in general.

Hardware is often not that logical and obvious, there are so many things that can result in such symptoms.

Did you overclock your system?

resetted your BIOS by jumper?

6GB of RAM could be to less for some games, 12GB with your triple channel motherboard should get you bit more comfort

you can set the RAM to 1066MHz (native speed of your CPU) with good timings

please run prime95 and watch the temperature as well as the frequency of the CPU, shouldn´t go over 67°C and below 2.80GHz

nvidia driver was a bit outdated, but don´t think it´s important

 

[JFST]

Check again to be sure that it´s not your system in general.

Hardware is often not that logical and obvious, there are so many things that can result in such symptoms.

Did you overclock your system?

resetted your BIOS by jumper?

6GB of RAM could be to less for some games, 12GB with your triple channel motherboard should get you bit more comfort

you can set the RAM to 1066MHz (native speed of your CPU) with good timings

please run prime95 and watch the temperature as well as the frequency of the CPU, shouldn´t go over 67°C and below 2.80GHz

nvidia driver was a bit outdated, but don´t think it´s important

Ok, but give me some time as it take a lot of time to run these tests for HDD (arround 3 hours) and RAM (3 hours too to do it 3 times) (hopefully this weekend).

-Nope, no overclock (unless you count overclock to the XMP).
-No, BIOS is with the last update FH, and i thought it is better for a new card to have the latest (well, from 2012).
-I know 6GB is not a lot but games like CSGO or Solarix should have no problem with that.
-Ups, i said 1033, and i wanted to put 1066, sorry.
-Which Prime95 version? (I read that anything over 26.6 can damage computer) And has any effect on GPUs? HD6870 has not the best cooling.
-Well, these test are arround 8 months old so drivers were up to date then 😛 .

Still i find it weird that the problem can be on my hardware if others have almost the same problems, and mainly if with my HD6870 i don't have them, i mean, if it is a hardware problem not related to GPU, shouldn't i have it with the ATI too? and also have it on all games?
But well, some tests won't hurt (i guess xD ).

 

[TerryLaze]

Thank you all for the answers 😉 .

TerryLaze, don't get angry on me as i don't know a lot about subjet ( 😛 😉 ) but, why would just a change of GPU make the CPU work harder on same games-settings?

How many FPS where you getting with the old ATI card and how many do you get now?
If a GPU is only capable of let's say 30FPS then the CPU has only 30FPS worth of work to do,if your GPU is capable of 100FPS your CPU has to be able to do 100FPS worth of work.

 

[helpstar]

because your CPU doesn´t support AVX, it´s not relevant which version you are using to stress test.
If it would use AVX with a newer CPU, it simply would get hotter than in normal stress testing matters, which can lead to a thermal shutdown. It won´t damage the CPU because of internal safety features of the CPU.

 

[JFST]

...
How many FPS where you getting with the old ATI card and how many do you get now?
If a GPU is only capable of let's say 30FPS then the CPU has only 30FPS worth of work to do,if your GPU is capable of 100FPS your CPU has to be able to do 100FPS worth of work.

I can understand that 😉 . On the other hand on games like Solarix i had stable 62fps (weird number i know but not the problem as i have it on both), and i can't remember if i had a lot more than now on CSGO. But i now understand what you said, thanks 😉 .

because your CPU doesn´t support AVX, it´s not relevant which version you are using to stress test.
If it would use AVX with a newer CPU, it simply would get hotter than in normal stress testing matters, which can lead to a thermal shutdown. It won´t damage the CPU because of internal safety features of the CPU.

Ok, used 29.2, but i "hate" stress tests, they stress me xD 😛 .
Here are the results of all tests (yes i lied a little when i said i will do it on weekend xD 😛 ):

-HDD

Spoiler

-RAM:

Spoiler

-Prime95 29.2 (ambient temperature at almost 40ºC -my mouse knows it's true xD-):
Result:

Spoiler

[Jun 20 1803] Worker starting
[Jun 20 1803] Beginning a continuous self-test on your computer.
[Jun 20 1803] Please read stress.txt. Choose TestStop to end this test.
[Jun 20 1803] Test 1, 460000 Lucas-Lehmer iterations of M250519 using type-0 FFT length 12K, Pass1=48, Pass2=256.
[Jun 20 1805] Test 2, 460000 Lucas-Lehmer iterations of M245759 using type-0 FFT length 12K, Pass1=48, Pass2=256.
[Jun 20 1807] Test 3, 460000 Lucas-Lehmer iterations of M245281 using type-0 FFT length 12K, Pass1=48, Pass2=256.
[Jun 20 1809] Self-test 12K passed!
[Jun 20 1809] Test 1, 800000 Lucas-Lehmer iterations of M172031 using FFT length 8K.
[Jun 20 1811] Test 2, 800000 Lucas-Lehmer iterations of M163839 using FFT length 8K.
[Jun 20 1814] Test 3, 800000 Lucas-Lehmer iterations of M159745 using FFT length 8K.
[Jun 20 1815] Torture Test completed 5 tests in 12 minutes - 0 errors, 0 warnings.
[Jun 20 1815] Worker stopped.

Images:

Spoiler

Also let me add this other thread i have found with same card and, i think, pretty similar problems:
http://www.tomshardware.co.uk/answers/id-3445778/gtx-nvidia-1060-scc-card-working.html
(Waiting his answer, but Fallout 4, and LoL behavoir makes me think this is same i had on F4 and same the user of the other thread had on LoL).

And i am going to ask more info to other threads that i think are similar.

I think this is a problem to a lot of 1060s (and mine included).

 

[helpstar]

Which cooler are you using?
Reapply thermal paste, e.g. Arctic MX -2

 

[JFST]

Which cooler are you using?
Reapply thermal paste, e.g. Arctic MX -2

Cooler Master Hyper 212 Plus.
Maybe temps are not great but i have almost 40ºC (on winter i never reach these). Should i try with 26.6 version just to see if changes anything?

Is not a bad idea anyway but i don't think it is the problem.

Anyway:
- Could you link to a video to learn how to apply thermal paste the best posible way? (no, haven't done yet 😛); i have read several ways to do it and still don't know the best way.

- If i upgrade RAM it has to be exact same i have now, right? and if i can't find it, it is better to just put 3 new sticks? Even if i find same RAM still is better to just have 3 sticks?

-If i try an ati, a RX 580 would be too much? (is the equivalent to 1060 i think, but maybe they perform different)
-I have seen Gigabyte Aorus and Sapphire Nitro+, which one would be better?
-With a difference of 41€ bettween 4gb and 8gb, which one?

As i said doubt the temps are the problem as i have been using the card on winter with same result, and also, now i don't have the problems on the HD6870.
I am searching for more threads with 1060s as i find it a bit weird that a lot of people are having similar problems.

😉

 

[helpstar]

one thing at a time 😉

i´d recommend the pea method, it´s the easiest way to apply thermal paste and the most promising.

to remove the cooler, just slightly move it clockwise, don´t pull it with force
Remove the old paste with q-tips or similar and clean with isopropyl alcohol (acetone might be too agressive)

https://www.youtube.com/watch?v=gDkz_9AQQ-g

 

[JFST]

one thing at a time 😉

i´d recommend the pea method, it´s the easiest way to apply thermal paste and the most promising.

to remove the cooler, just slightly move it clockwise, don´t pull it with force
Remove the old paste with q-tips or similar and clean with isopropyl alcohol (acetone might be too agressive)

https://www.youtube.com/watch?v=gDkz_9AQQ-g

Thanks for the video and tips (the "don't pull with force" is the most important as i didn't read before and seems basic 😉 ), and thanks to the video i found others that teach you to apply thermal paste on the cooler, maybe this is better to avoid to add a lot and the "extra" to go outside of CPU and break something (what do you think? the "pea" method is best to avoid air, but not sure about the "perfect" size of it and could be a problem if too much or not enough :? ).


On the other hand, i have been thinking and i feel that the best way of testing my problem is to try a different graphics card (AMD-ATI in my case). I think that, without changing anything else, if this gives me same problems too then i have to try to find what else could it be.
Otherwise i think i can spend a lot of money (thermal paste, RAM, SSD) and maybe, at the end, having to buy a full PC anyway. And once i find if is the GPU or not, i can improve everything else.

Unfortunatelly there is no stock of RX580s (no even date yet) or i would be able to test it soon (maybe vega will come even earlier that stock of these xD ).


Also (if you let me add to the questions i made earlier 😛 😉 ) if it would be some kind of bottleneck, what (old) NVIDIA card should be the "safest" option to choose? (i have read that 6XX has no problems).


So, what do you think of the "plan"? makes sense? 😛 😉

 

[helpstar]

Thermal paste is like 6$, try this first. It really doesn´t matter that much how you apply it. not too less and not too much is the thing. The middle of the CPU is the most important thing, you won´t have to spread it all over the top of the CPU.

official Intel application of TIM:
https://www.intel.com/content/www/us/en/support/processors/000005576.html

 

[JFST]

Thermal paste is like 6$, try this first. It really doesn´t matter that much how you apply it. not too less and not too much is the thing. The middle of the CPU is the most important thing, you won´t have to spread it all over the top of the CPU.

official Intel application of TIM:
https://www.intel.com/content/www/us/en/support/processors/000005576.html

I know but as i don't have the 1060 anymore i won't know if any change i make have any effect until i rebuy a 1060 (doubt i will) or until i try an equivalent ATI (and in this case it would be ""worse"" as i won't know what fixed it).
That's is why i think it is better to know if any GPU gives me problems, and if they all do, then i will try to fix other components and, maybe, find a "weaker-older" card that won't be bottlenecked by CPU or motherboard or whatever.

Also, can i ask what temps did you expected the CPU to have? (just to know how bad are they 😉 ).

Talking about the intel link you posted: THANK YOU! I had no idea these existed and in a couple of videos i watched i learned a lot of things i was a bit worried about (not only about thermal paste but also for CPUs installations). I'm saving this link! 😉 .

 

[helpstar]

TCASE 67.9°C for this CPU, but I think it´s fine for prime strest test to get higher

 

[JFST]

TCASE 67.9°C for this CPU, but I think it´s fine for prime strest test to get higher

Thanks, that is actually what i have always had as maximun on normal usage (games): 68ºC, unless i remember it wrong. But it's been a lot of months since i checked these so maybe they changed (i will take a look some days just in case).

If it makes any change it started at 78ºC and stayed there, i mean they were not slowly increasing or changing, it was stable at 78ºC (at the end i think it decreased just a couple of ºC).
Maybe i will try Prime 95 26.6 just in case it affects my CPU too (just for curiosity 😛 😉 ).


Well, as i said i will try to find an RX580 (damn miners xD ) and see how it performs without any other change just to discard it or point to my other hardware.
Sooner or later i will end up changing my thermal paste so thanks helpstar for the (more than you think) useful links and tips 😉 .

I will report back once i have done this 😉 .


Of course you all can keep comenting and asking me tests, but i will try (key word 😛) not to bother a lot as i think i have abused a bit of your help (thanks again 😉 ).

Except (sorry xD ) for these questions:
-RX580 ok or a waste on my system?
-Gigabyte Aorus or Sapphire Nitro+? 4gb or (+41€) 8gb (on my system)?


And one last favour that is to take a look on 1060s problems theads to see if they look like mine and if there is something in common, as i find it pretty weird the coincidence of behavoir on a lot of games.


Thank you all again for your great help 😉 .

P.S. again, keep commenting as i love as much info as posible 😉 .

 

[helpstar]

the more video RAM the better, more detailed textures can be loaded.
Like in Rise of the Tomb Raider, about 4GB would be not enough to run this game at full graphics details, so 8GB would be preferable in all cases.

The RX580 is a bit better than the 1060 and normally cost a bit less, so should be worth. Which card you choose is not that important.

eventually this one is available
https://pcpartpicker.com/product/9ZmxFT/asus-radeon-rx-580-8gb-rog-strix-video-card-rog-strix-rx580-o8g-gaming

 

[JFST]

the more video RAM the better, more detailed textures can be loaded.
Like in Rise of the Tomb Raider, about 4GB would be not enough to run this game at full graphics details, so 8GB would be preferable in all cases.

The RX580 is a bit better than the 1060 and normally cost a bit less, so should be worth. Which card you choose is not that important.

eventually this one is available
https://pcpartpicker.com/product/9ZmxFT/asus-radeon-rx-580-8gb-rog-strix-video-card-rog-strix-rx580-o8g-gaming

Checked the asus but it is a bit too big for my case (and i don't like asus after a bad warranty experience 😛 ).

Ok, 8gb then (hopefully bottlenecks won't waste them 😛), and being more powerful than a 1060 i think problems will show soon once installed (if the problem is not the GPU). I think at the end i will choose Sapphire as seems the more reliable brand and has good temps and as it seems people says ATI is "not as good as NVIDIA on drivers and hardware" (not for my 1060 😛 -will see xD -) i prefer to choose the best i can.

But maybe if they take too long to have stock i will wait for Vega that would be just 1-2 months more with good new features (hopefully it will be compatible).

I'll report back when i can test something.

Thanks 😉 .

 

[JFST]

I have found i have like half syringe of thermal paste from what i got with my Cooler Master Hyper 212 Plus (close to 10 years ago), do you know if would it be fine or it is better to buy a new one? Also would it be enough? have tried to find somewhere how much had it but i couldn't; i think i could even have for 2 times but maybe it is way less than it looks like (never applied it so close to 0 idea).

Also no need to remove CPU to clean it, right?
I have been checking my temps and never reach these 78ºC when gaming (62-64ºC), if the CPU temps are the culprit, would i get any stutter even if it don't reach these temps?
If it is some kind of bottleneck this will do nothing i guess (but thought bottlenecks didn't cause stutters).


And helpstar once thermal paste is applied, what was going to be you next steps? (just to know if there is some method).

I know i said i was going to "try" to not ask ( 😛 ) but as i am waiting to get stock of GPUs i have time to think on all things i will do once i get it and test it (thermal compound is a thing i will almost sure do once i find if there is something wrong with GPU or my computer, but as i will do it once i test GPU i am planing on adding some TP syringe to the order to save some shipping xD 😛 ).

😉

 

[helpstar]

no need to remove the CPU, but be gentle while cleaning.

after applying thermal paste, reinstall the cooler and plug in its fan

62-64 while gaming is fine and wouldn´t require reapplying thermal paste.

 

[mdd1963]

Hours upon countless hours of research, reading etc....

Can often be avoided with a good 'ole 'Nuke and Pave'! (Delete Partitions, reformat, reinstall)

Jumping from one GPU to a competing type can leave little driver remnant booby traps that are not easily detected or corrected....

One ironclad/rock-solid way to eliminate this: NUke it!

 

[helpstar]

like recommended before, reinstallation of windows would be my first choice after all too

 

[JFST]

Thanks for the answers guys 😉 .

So by now if the games keep CPU under 68ºC it is safe, so no need to reapply unless it pass it (i am a bit surprised it keeps "good" temps after so long 😛 ).

About the half syringe of thermal paste i found, i have read that maybe it is safe to use even after all these years if it looks fine, just to keep an eye on temps once applied.

About Windows 10 reinstall maybe it is what i have to do if the new GPU acts weird too and CPU temps keep fine, so 'Nuke and Pave' pointed 😛 (but i'd hate it xD ) .

Now i am reading (no videos found xD ) what happens if you put too much paste and it goes to sides of CPU, motherboard and other components and mainly how to clean it. Maybe this is too dificult to happen but i have always been a bit worried if it could happen :/ xD .

😉

 

[helpstar]

It depends on your thermal paste, if it´s electrically conductive or not. If not, that´s no problem at all.

 

[JFST]

It depends on your thermal paste, if it´s electrically conductive or not. If not, that´s no problem at all.

Yep, that is what i see (i have read big messes that get it even under CPU, hope not to be that unlucky xD).
Seems what the Hyper 212+ has is not conductive, as the Arctic MX -2 you recomended 😉 .

Well, as by now temps seem safe i will try to focus in my plan of testing a different GPU xD (when i have to wait i tend to think all options and things that can happen and how to solve them 😛 ).

😉

 

[JFST]

Guys, i know, but i need your help one more time:

I still have stutters on Killing Floor 2 (not sure if same or just the 6870 can't "give more" because in Solarix, performance was perfect) so i have found this:
https://forums.tweaktown.com/gigabyte/51776-please-help-solve-stutter-4.html

Spoiler

After 3 years if you wannt the solution.:

When u pick optimised option u have : CPU Clock Drive= 800mv
pcie clock drive= 900mv

This is the problem. If u put the both 700 mv the problem now is not exist. I had the same problem And the only thing i dont change was motherboard.
The oficial specs for cpu clock drive is 620mv and pci e 700mv. I dont know why they used so big 800 and 900 overshoots, and corrupt the signal.
IF the big number has no bad efeects intel will used for defaults 1000 and 1000 mv. So they knew what they done.

So, does that make sense? (i can't really understand it).
Is it safe if i do it?
I got into bios and found these 800 and 900 values, what i have to do is just to change them directly or do it choosing a setting somewhere? (if you have a guide i'd appreciate it).

Maybe it is the problem (and others with different hardware have these same problems too because they have similar settings on their bios?), or maybe it has nothing to do.

 

[helpstar]

you can safely try undervolting the settings to 700mV. If it crashes or won´t boot anymore, just reset the BIOS by jumper (clearCMOS while power is unplugged from the wall socket)

 

[JFST]

you can safely try undervolting the settings to 700mV. If it crashes or won´t boot anymore, just reset the BIOS by jumper (clearCMOS while power is unplugged from the wall socket)

Ok, i will try to do it today (i have to do some things and if it gives me problems i prefer to have these done 😛 ).

I will set directly both, CPU and PCIe, at 700.

Also what can happen by doing it, even if it becomes more stable will i get less performance o what?

And can others have same problems because of similar settings on their bios? (if it can how has nobody found that yet?)

And it is true that the user says that "The oficial specs for cpu clock drive is 620mv and pci e 700mv"?

Well, i will report back once i have tested it just to see if makes any change on KF2 (or just i am thinking it too much xD ).

Thank you for the help 😉 .