Asus Strix GTX970 4GB Massive problem, need help. [SOLVED]

[Foxiol]

Hello everyone, first of all I'll post my specifications because I got new parts for my PC and I'm having problems with my new Asus Strix GTX970 and did a lot of tests, tried some things and now I am in the limit of giving up so I need other opinions and/or suggestions.

PC Specifications:

-Case: Cooler Master HAF 932 Advanced
-CPU: AMD FX8350 @4.0Ghz (stock)
-CPU Cooler: Cooler Master Hyper 412S (CPU never went above 50ºC)
-Motherboard: AsRock FX990 Extreme 3
-Memory (RAM): Kyngston Hyper X 1866mhz 4+4GB (2 slots = 8GB total system RAM)
-GPU: Asus Strix GTX970 4GB
-SSD: AMD R7 240GB (primary where the OS is installed of course)
-HDD's: Western Digital 1TB / Seagate 350GB (just as a backup)
-Sound Card: Asus Xonar DG 7.1
-PSU (power supply): Corsair RM850 (850W) full modular brand new (bought it with the GPU)
-OS: Windows 10Pro/Windows 7 Ultimate 64 bit (made tests in both, and both same results, fully updated)

An image showing current motherboard temps, GPU and CPU:

Advice to all readers:

Also before talking about the issue I did:

-A "memtest" to check my memory and are working as expected.
-Used Prime95 to stress my CPU and everything works fine (all cores) allways running at 100%.
-Used Cinebench R15 and got 86FPS in OpenGL and 650 score in CPU benchmark (compared this with other results and all fine here.
-Used GPU Z "mini" stress test to check if my PCIe port is working in 16x 2.0 and it was okay.
-Used CPU Z stress test for my CPU and got 1200 score in Single thread and 7400/7600 in Multi thread operations.

Never had a single crash error, not a single blue screen in Windows, no drivers crashes, nothing weird at all in both Windows 10Pro and Windows 7 (where I am now since I think it is a more "stable" OS than Windows 10 at the moment).


MY PROBLEM:

At some games like Fifa 16, The Witcher 3, Metro Last Light, Elite Dangerous: Horizons, I play normally until 1 or 2 minutes, then I start to get massive slowdowns for no apparent reasons.

For example in Fifa 16, I start a match against IA's and then very first minute the game works fine at around 170/180FPS and suddenly the game slows down to 40FPS for around 4 or 5 seconds (is not just a drop in frames, I loose control like while having lag).
At the same time CPU usage goes up to 99% for those 4 seconds and GPU usage drops from 99% (100%) usage from 40% more or less and voltages also drop (all the time working at 1.2000V and drops to 0.8XXXX and fluctuates around the max value) close to the actual lower value which is 0.8600V more or less.

So this is something related to Voltage limits or the GPU while regulating Voltages...Or my motherboard can't handle the high Overclock from the GPU or can't handle the voltages or the changes in voltages.


This is the graph in GPU Z while playing Fifa 16:

1st image is once I started the match coming from the main menu, in Fifa 16 GPU usage is higher (99%) in menus than in the actual game while playing (60% GPU usage average while playing):

2nd image shows the first slowdown (right of the graph), check the frames and the CPU (99%) and GPU (less than 40% usage) usage:

3rd image shows all the slowdowns that I got in a minute (3 slowdowns, every 20 seconds):

This picture is from great part of a match in Fifa 16 on Windows 7 Ultimate 64 bit:

This picture shows great part of an entire match in Fifa 16 but playing on Windows 10Pro:

Just one more thing, I tried using VSync and the problem persisted but less frequently, and also tried Adaptive VSync and in Windows 10 I could play some matches without slowdowns. I saw CPU usage being below 50% and can't remember if GPU usage was higher than in here or the same.
But the graph from voltages was almost the same showing me the orange and blue colours and such.

A friend has a GTX980Ti and showed me that GPU usage was all the time at 99% while playing, and Performance Cap Reason for him was all in blue (Reliability Voltage).

A friend with a GTX970 G1 told me that his graph was similar like mine (GPU usage and such) but instead of showing him the voltage as his cap reason, it was Utilization only. (like the GPU being more than enough for the game of course)


So I have no idea what to do...or what can it be. I think that It must be something with Voltage regulation from the GPU or my motherboard as I said before.

Is not software related and not my CPU or anything else since as I said at the top, tried all my hardware and is working fine, no issues at all with memory, crashes or blue screens.

This is weird and I need to know what is happening.


Thanks in advance for taking your time to read this and hope I or we can find a solution.

 

[mstenholm]

You got 8 wire power supply. Look at any of the other GTX 970s - they have 6+8. I got 3 MSI and one ASUS. Same problem when I fold. Live with it or get a MSI.

 

[Foxiol]

You got 8 wire power supply. Look at any of the other GTX 970s - they have 6+8. I got 3 MSI and one ASUS. Same problem when I fold. Live with it or get a MSI.

Well I have another power supply...a cheaper Tacens Radix V 850W Plus Bronze with 2 +12V 36A and have the same problem with it. Actually that's why I got the RM850. Just in case, this PSU (RM850) comes with 8 wire and 6 wire cords so I can put anything with it and it uses a single +12V 70.2A as the main power. I have more power than I need.

Thanks for your reply.

 

[flmatter]

no he means the power connector on you video card, top one is your video card the second image is the other gtx 970's
as you can see yours only has an 8 pin connector and others have a 6 & 8 pin connector. Your psu has nothing to do with it.


edit also update your profile with your system specs makes it easier for others to see your machine without constantly having to go to start of thread.

 

[Foxiol]

top one is your video card the second image is the other gtx 970's
as you can see yours only has an 8 pin connector and others have a 6 & 8 pin connector. Your psu has nothing to do with it.


edit also update your profile with your system specs makes it easier for others to see your machine without constantly having to go to start of thread.

I know that now because the one I'm using now is new...I bought it thinking that my old one was causing the issue.

The old one was a Tacens Radix V 850W Plus Bronze (2x +12V 36A) which I have since 2011 and never gave me problems. (and if it wasn't for me it will still be here on my PC)

I got the Corsair RM850 one day after getting this GPU.


PS: I am going to build a 2nd PC with old components not meant for gaming, so is not a waste of money since the old one works great.

PS2: Profile updated with system specs.

 

[DarthBaggins]

very odd issue, I have a Strix 970 as well (but w/ /Samsung Memory) and I have mine clocked to 1500. What driver set are you running?

 

[Foxiol]

very odd issue, I have a Strix 970 as well (but w/ /Samsung Memory) and I have mine clocked to 1500. What driver set are you running?

Hi, well bios from motherboard and GPU are both up to date (motherboard is 1.90 and GPU is 84.04.36.00.5E), and tried latest drivers from Nvidia 361.43 on Windows 10 Pro and Windows 7 which have issues with Open GL (artifacts and stuff like that with all Nvidia Maxwell GPU's).

Now I'm using 359.06 which work great in all aspects as far as my tests were made.

 

[DarthBaggins]

359.06 is the one I'm running and have had zero issues w/ it, and I have yet to try that newer one (I don't use Nvidia Experience either as it can cause issues in w10 or has in the past)

 

[xvi]

Odd that GPU utilization goes down, but there's no change in perfcap reason. I'd almost say it's an issue with something besides the GPU.

Edit: Image below shows perfcap reason "Util". Card is knowingly throttling down because it's not seeing utilization. I'd almost be curious to say something is stressing CPU.

 

[Foxiol]

359.06 is the one I'm running and have had zero issues w/ it, and I have yet to try that newer one (I don't use Nvidia Experience either as it can cause issues in w10 or has in the past)

Hi again, uninstalled GeForce Experience just in case and played a match in Fifa 16. Got a little bit higher frames (around 190/200) and it seemed stable until it started to happen again.

Here is the graph of the entire match with slowdowns:

For the record, this also happens in The Witcher 3, Metro Last Light and some other games. In for example Helldivers (top down shooter) this doesn't happen and the graph where voltages are, appear all in grey colour which means that the GPU is not being stressed at all. (this game barely uses vRAM and get above 400FPS withouth VSync).

I'm using Fifa as the reference because it is the one in where I can reproduce this error quick. Also I had a Gygabite GTX560Ti OC 1GB and it worked great at 140FPS stable, no slowdowns or such. (never had to use GPU Z since I never had problems with the GPU at all so I could check voltages and such...don't have it here with me anymore sadly...).


@xvi, Well check back in that graph made a couple of minutes, you can see that PerfCap Reason goes to grey when slowdowns occur. Weird for me is to see both Voltage values at the same time and even "Power" (the green colour) also appearing sometimes during the gameplay and mostly at main menu where the GPU Load is higher than while playing.

So confused.


PS: Maybe I'll go and change this GPU tomorrow (not sure if the very same model or MSI model) and get rid of the mistery if it is the GPU itself, or it will be the motherboard. (I was looking for an Asus Sabertooth R2.0)

 

[newtekie1]

The second picture shows the CPU only running at 1.3GHz during the slowdown. I believe your CPU might be throttling.

It very likely could be the VRMs on your motherboard overheating. The CPU might not be getting too hot, but the motherboard power delivery for the CPU might be. That motherboard isn't the best for 8-Core CPUs, it is only a 4+1 Phase power design. It would probably be fine for a 4 or 6 core CPU, but 8-Cores is probably overloading it.

If you can, try going into the BIOS and disabling 2 cores to make your CPU into a 6-Core. That should reduce the power usage. Then see if the throttling still happens.

 

[Foxiol]

The second picture shows the CPU only running at 1.3GHz during the slowdown. I believe your CPU might be throttling.

I was thinking the same but after trying Cinebench R15, Prime95 for half an hour I saw nothing weird at all with my CPU.

The pictures were taken in a pause so I could copy them by the way, that's why values are not the same in all pictures.

And CPU goes up to 99% usage when slowdowns happen.

 

[xvi]

The second picture shows the CPU only running at 1.3GHz during the slowdown. I believe your CPU might be throttling.

I was thinking the same, but dismissed it because temps were good. Could be throttling for.. uh.. VRM temps maybe?

@Foxiol Perhaps grab OpenHardwareMonitor (or anything that will graph CPU core speeds), run it, click "View", click "Show plot", then check the boxes for CPU Freq on all cores. Run the game again, see if CPU speed dips down when you see the game/GPU-utilization dip too.
Edit: Might as well check CPU load to see if/when speeds dip, if the cores max out at 100% too.

 

[okidna]

To prove what @newtekie1 wrote, try to disable Turbo and Cool n Quiet from your BIOS, and then play your game in full screen with HWMonitor running in the background. Make sure you reproduce the issue for a couple of times then quit the game. And then post the HWmonitor screenshot (make sure it shows CPU Clocks) here.

 

[Foxiol]

I was thinking the same, but dismissed it because temps were good. Could be throttling for.. uh.. VRM temps maybe?

@Foxiol Perhaps grab OpenHardwareMonitor, run it, click "View", click "Show plot", then check the boxes for CPU Freq on all cores. Run the game again, see if CPU speed dips down when you see the game/GPU-utilization dip too.

Going to do that now. Stay tuned...I use Open Hardware Monitor as well.

 

[okidna]

I was thinking the same, but dismissed it because temps were good. Could be throttling for.. uh.. VRM temps maybe?

@Foxiol Perhaps grab OpenHardwareMonitor (or anything that will graph CPU core speeds), run it, click "View", click "Show plot", then check the boxes for CPU Freq on all cores. Run the game again, see if CPU speed dips down when you see the game/GPU-utilization dip too.
Edit: Might as well check CPU load to see if/when speeds dip, if the cores max out at 100% too.

It could be the VRM temperature. I used to have a similar board (970 Ext3), different chipset but very similar VRM configuration. And it throttles all the time with my old FX8120 (even on stock clock) with high VRM temp, up to 65 C.
However, all is fine when I downgraded the CPU to FX4100 and then upgraded to FX6300.

 

[xvi]

I have an 8350 overclocked on a ASRock 990FX Extreme9 and haven't had any VRM issues, but those Extreme3 VRMs don't look as strong.

If it does turn out to be VRM temps, @Foxiol since you're running at stock clock speeds, you might be able to undervolt the CPU a fair bit which should reduce temps considerably and (hopefully) resolve your issue.

 

[newtekie1]

And CPU goes up to 99% usage when slowdowns happen.

If it is using 60-70% at 4.0GHz, it is definitely going to be 99% at 1.3GHz, and the CPU will very quickly become the limiting factor.

 

[Foxiol]

Well guys here it is the picture with Open Hardware Monitor and GPU Z Graph, and again a Full match in Fifa 16:

Edit: I put it back again...at the end is with the game closed and the apps I'm using for monitoring stuff:

Tell me what you think or see. @xvi , @newtekie1 , @okidna .


PS: Those drops could mean one thing...or I mounted my CPU cooler like my face in the morning...or something is wrong.

 

[okidna]

Well guys here it is the picture with Open Hardware Monitor and GPU Z Graph, and again a Full match in Fifa 16:


Edit: I put it back again...at the end is with the game closed and the apps I'm using for monitoring stuff:


Tell me what you think or see.

Those CPU clocks does not look right at all. IMO your CPU is throttling and causing your GPU to stutter.

Can you try what newtekie and xvi suggested next? Disabling 2 or 4 of your cores and/or undervolting your CPU (since you're running at stock clock).

EDIT : for a comparison, I did a short (8 minutes) Witcher 3 run and here's the result :

Spoiler: hwmonitor

I still enable my Cool n Quiet but you can see that the CPU (red and all those colors) and GPU clock (blue) stay stable all the time. Two dips around 0:04 mark on the CPU clocks was because I paused the game to enter crafting menu.

 

[newtekie1]

Edit: I put it back again...at the end is with the game closed and the apps I'm using for monitoring stuff:

Tell me what you think or see. @xvi , @newtekie1 , @okidna .

Your CPU is definitely throttling.

 

[Foxiol]

Those CPU clocks does not look right at all. IMO your CPU is throttling and causing your GPU to stutter.

Can you try what newtekie and xvi suggested next? Disabling 2 or 4 of your cores and/or undervolting your CPU (since you're running at stock clock).

Sure I can "undervolt" but tell me more or less how much shold I put.

And have no idea how to disable the cores in this motherboard since I never messed so much with the Bios aside from checking RAM and other values for the SSD and such. (I can search in how to do it, sure I'll find a tutorial in the net)


@newtekie1 When I bought my new GPU I also bought a Cooler Master Hyper 412S CPU Cooler. First time I put one (non stock one) and I bet I did something wrong (I removed the plastic protecting it at least...I'm sure about that, and also checked if it covered the whole surface -centered nicely-...not sure if I put it with the right "pressure" against the CPU tho).

The stock Cooler Master CPU cooler that came with the FX8350 worked quite nice but its noise was too loud. I set its target for 50ºC and it was always working at about 5000+rpm.

This one works at around 600/700 rpm in idle and 1200/1300rpm max speed in load. Is too quiet that's for sure but I don't like the low rpm fan.

Maybe I know where my problem is now... (probably going to remove this garbage and put the loud one back again...)

I'm really pissed...

 

[newtekie1]

I believe, in our BIOS, under the OC Tweaker heading there is an option called CPU Active Core Control. Change that from Disabled to either 4 or 6 cores.

 

[okidna]

Sure I can "undervolt" but tell me more or less how much shold I put.

And have no idea how to disable the cores in this motherboard since I never messed so much with the Bios aside from checking RAM and other values for the SSD and such. (I can search in how to do it, sure I'll find a tutorial in the net)

Never owned FX8350 but once I set my friend's FX8350 with 1.275v to run on stock clock & stock cooling and it did ran fine until he went for AIO water cooling and OCed his CPU.

Go to your BIOS and enter AsRock OC Tweaker, find "CPU Active Core Control" option and you should see how many cores you want to activate.

 

[Foxiol]

I believe, in our BIOS, under the OC Tweaker heading there is an option called CPU Active Core Control. Change that from Disabled to either 4 or 6 cores.

Okay, I saw that feature. But if it is the temps going up and down like crazy which is causing the CPU to throttle...Should I try by changing the cooler again to the old one (later today) and see what happens?

@okidna Great, going to do that now. Not sure about the clocks. I'll try this first and if the problem persists I'll do the cores thing.

Going to reboot and do the test again with Hardware monitor and such.


Just in case, If I don't see you guys here again THANK YOU VERY MUCH FOR TAKING YOUR TIME! And really you all have no idea how bad I feel...I have this problem since the 25th and had to change OS, downlaod stuff with my not so good network, and do a lot of things...and all of this in the last week of the year. Incredible the time I wasted instead of being with my family....

So just in case I wish you all a Happy New Year and all the best for you guys. Hope we can solve this issue before the day ends!!