• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

GTX1070: Weird Throttling, Sudden Drop In FPS

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

ronferri

Registered
Joined
Oct 31, 2017
Hello,

I will try to make this as short as possible so I won't lose your kind attention.

Here is a recording of the insane GPU throttling I am experiencing in game:

Specs & Settings:

I have an MSI GT62vr laptop: CPU i7-6700HQ, GPU Nvidia 1070, 16GB RAM.

Vsync is turned off in Nvidia control panel and in all games setting.
In all scenarios, Sports mode is activated in MSI Dragon center which brings the GPU and CPU to their max native clocks without any Overclocking. In Nvidia control panel, power mode is set globally to maximum.

Problem:

My system has been running perfectly for 10 months up until last week where it can no longer achieve the regular high fps in the same games I used to play.
In Ryse son of Rome on ultra, I used to get steady 80-100fps, now it cannot go above 70fps max, while constantly dipping to 25-35fps.
In Crysis 3 on very high, used to get 70-80fps, now 35-65fps.
In Battlefield 3 on max settings, used to get above 150fps, but now down to 70-110fps.
The degree of jitter or throttling of FPS is high in all these games. The FPS ranges significantly while gaming. I tested for overheating and it is not a temperature problem.

Investigation: (6 is the most interesting)

1. Drivers: I updated to the latest nvidia 388 driver using DDU clean install. Problem not solved. I reverted to the factory stock nvidia drivers using DDU. Problem still not solved. I did not "remove the present and nonpresent monitors" (under options of DDU) which is recommended to do, so I will try that again soon although I doubt it.

2. Temperature: I monitored the GPU and CPU using MSI Afterburner and their temps range between 64C and 80C at maximum during Gameplay. Correct me if I am wrong, these are not considered as high temps which might cause voltage throttling.

3. Utilization: This bothers me as MSI Afterburner shows that neither the GPU nor the CPU are being fully utilized up to 100% in any of the games at max settings. They go up to 70%. The CPU is always more relaxed at 40%-60%. I saw recordings on youtube of identical systems playing same games with GPU always on 99% and much higher FPS than my current performance. My GPU does get clocked to around 1700Mhz during gaming while on Sports mode. It can reach a max of 1919Mhz native. Never OC'ed anything.

4. Malware and Cryptomining: Some theories on the web state that malware can cause what i am experiencing so i am running a deep system scan while typing this.

5. Benchmarks: This is the tricky part. All synthetic benchmarks such as Unigine Heaven, Valley, and PC Mark 7 resulted in competitive scores which are comparable to similar systems. My Low FPS and throttling problem is more evident in games.

6. Lowering Settings: I lowered graphics to lowest settings in all the above games and I still got the low fps problem. This is very perplexing. Ryse Son of War wouldn't go above 64fps again even on lowest settings same as on ultra settings! It went over 100fps before this problem started. Same for other games. Lowering settings does not change FPS. What does this tell you?

I also uploaded the logfile from GPU-Z during my gaming session from the video. Note the many "perfcap reason 16". I dont know what that means.

I am out of options, and appreciate your guidance on how to restore my original FPS power.

Thank you
 

Attachments

  • GPU-Z Sensor Log 2.txt
    62.7 KB · Views: 70
Last edited:
Two things to consider here:

1, I also have an MSI laptop (GE62 6QF). I was testing CPU and it was under-performing which I found odd. In short, their laptops can only run at full power if they draw on both the mains adapter and battery. If the battery is low, it will reduce performance. Was the testing performed with a high battery level? Let it charge full and repeat.

2, When I first got my 1070 in a desktop system, I had some weird benchmark results. It is just too long ago for me to remember the details now. I was getting a low score in one of TimeSpy or Firestrike. Monitoring was showing low usage like reported here. I was wondering if somehow it was getting stuck in a low power state. I tried some things in both GPU driver and Windows power that I found in searching, none of them helped. What I did find was that under some loads it works at full power, then after that, the low score test would also give normal scores. I never figured out why that was happening. I haven't noticed anything similar recently, so I guess something had changed.
 
Thanks for reading. Yes battery is full and plugged in.
 
Last edited:
Did you update the graphic drivers?Maybe a driver issue? Worth a clean driver install (remove current with DDU) with a previous version.
 
Thanks. Already covered in original post, part (1) under Investigation section. Sorry it is a long read.
 
Start with the basics. Can you post up some screens of CPUz main , memory and SPD. Also some hwinfo64 shots, while the laptop is under load. Gaming will be fine. See if we can figure out what's going on.
 
I would start with some of the basics:

1) Do you have GeForce Experience Installed? If so, check to make sure that the DVR functionality is not turned on (I forget what they call it, but if you go into the options you will see something with a name that resembles this). If it's on, it can cause this wacky behavior.

2) Are you running Windows 10? If so, check to make sure that the Xbox DVR option is not enabled. Go to start menu, run the Xbox setup and check to see if it's enabled. If it's on, it can cause this wacky behavior.

3) In the NVIDIA control panel, do you have the GPU power set to "Prefer Maximum Performance"? In my experience, the "optimal" or "adaptive" setting was less than stellar.

4) There was a Windows 10 update which was causing some wacky framerate experiences. I read about it on the EVGA forums...but cannot find the post. The solution there was to install the update to the update to fix the problem.

5) Check the frame-sync setting in the NVIDIA control panel. I set my to "fast sync" which causes much less stutter compared to the typical sync setting.
 
Two things to consider here:

1, I also have an MSI laptop (GE62 6QF). I was testing CPU and it was under-performing which I found odd. In short, their laptops can only run at full power if they draw on both the mains adapter and battery.
That's garbage and MSI is taking shortcuts and cutting corners if that's the case. I have a laptop with an overclocked processor and overclocked 1070 and I can run on max screen brightness with gaming for hours and not drain the battery 1%. MSI needs to use a larger power supply and stop being a cheap a** then if that's the case. Buy a better brand next time.
 
Start with the basics. Can you post up some screens of CPUz main , memory and SPD. Also some hwinfo64 shots, while the laptop is under load. Gaming will be fine. See if we can figure out what's going on.

Thanks. Here are CPU-Z and HWInfo shots. The latter shows CPU throttling during gaming (Core 0 and Core 1) at the bottom of the sensor screenshot. Does this explain the wild fluctuation in FPS in my previously uploaded gaming video? I am interested in your ideas for solutions: Is a simple cleaning and repaste enough? Or shall i return the whole laptop because it is only 10months old and this is a deal breaking problem to be that can keep on happening.

HWinfo Load.jpg CPU-Z Main.jpg CPU-z Memory.jpg CpU-Z SPD.jpg
 
Last edited:
Furmark bench results:

furmark.jpg

Observations:
-Under limits, I had Power:1, voltage:0, temp:0, OV:0 (does that mean i have power limitation?)
-GPU Clock: stayed at ~1290Mhz whereas it does go up to 1900MHz in games as seen by MSI AB
-Voltage: ~0.7V whereas it does go up to 1.05V in Games as seen by MSI AB
-Temperature: 70C at the end, benchmark ended faster than I thought. Less than 5mins.
-FPS: min 113, avg 119

Given all the above, does that mean I only have CPU thermal throttling and nothing wrong with the GPU? Except for that power limit observation in furmark?
 
Last edited:
That's garbage and MSI is taking shortcuts and cutting corners if that's the case. I have a laptop with an overclocked processor and overclocked 1070 and I can run on max screen brightness with gaming for hours and not drain the battery 1%. MSI needs to use a larger power supply and stop being a cheap a** then if that's the case. Buy a better brand next time.

It isn't well documented or that widely known, so I found out long after I could do anything about it. In practice it doesn't have much impact unless you have a flat battery and you want to game plugged in. Keep it charged and no problem. For gaming for longer periods of time you'll be plugged in anyway. This is primarily a gaming laptop, not one you use mobile on battery much.

When it comes to replacing it I will certainly look at other brands, but at the time its spec was the best available in the price range and it does perform as such.

Thanks. Here are CPU-Z and HWInfo shots. The latter shows CPU throttling during gaming (Core 0 and Core 1) at the bottom of the sensor screenshot. Does this explain the wild fluctuation in FPS in my previously uploaded gaming video? I am interested in your ideas for solutions: Is a simple cleaning and repaste enough? Or shall i return the whole laptop because it is only 10months old and this is a deal breaking problem to be that can keep on happening.

Throttling certainly wont help. I assume they have a "warranty void if broken" sticker on the bottom, so it is up to you if you want to gamble with opening it. Depends on your local laws what return policy is. Here at that age you would not have a legal right to return/replacement, but have to go through manufacturer warranty/repair process.

Power limiting is not a problem. GPUs will adjust their clocks by themselves, even without OC, to maximise performance until some limiting factor is hit. Furmark generates above average power, so you can hit the power limiting. Lower loads may limit clock due to voltage and it can't clock itself higher.
 
Furmark bench results:

View attachment 194516

Observations:
-Under limits, I had Power:1, voltage:0, temp:0, OV:0 (does that mean i have power limitation?)
-GPU Clock: stayed at ~1290Mhz whereas it does go up to 1900MHz in games as seen by MSI AB
-Voltage: ~0.7V whereas it does go up to 1.05V in Games as seen by MSI AB
-Temperature: 70C at the end, benchmark ended faster than I thought. Less than 5mins.
-FPS: min 113, avg 119

Given all the above, does that mean I only have CPU thermal throttling and nothing wrong with the GPU? Except for that power limit observation in furmark?

Stop using furmark. Gpus throttle when running that test. Run 3dmark fire strike or something.

No clue on the cpu. I dont see any information which show you are cpu throttling.

Edit: i see it in hwinfo...hmm...

Edit2: that said, its what 400 mhz throttle? That xould cause 6 fps swing from 113 to 119... bjt that difference is normal, really.
 
I'm with E_D on this. From what I see in HWinfo there isn't much throttling going on. Your minimum clocks dip to ~ 3100 which is still 500 MHz over base, average is 3150 very close to all core turbo of 3200 and max at ~ 3500 which is single core max turbo. Looks like it's running correctly to me. This would lead me to believe the problem is likely SW related ie. drivers or the fall update/ compatibility issue as has been mentioned. I would check the MSI website and see if there have been any updates for that laptops drivers.
 
I'm with E_D on this. From what I see in HWinfo there isn't much throttling going on. Your minimum clocks dip to ~ 3100 which is still 500 MHz over base, average is 3150 very close to all core turbo of 3200 and max at ~ 3500 which is single core max turbo. Looks like it's running correctly to me. This would lead me to believe the problem is likely SW related ie. drivers or the fall update/ compatibility issue as has been mentioned. I would check the MSI website and see if there have been any updates for that laptops drivers.

Thanks Earthdog and Johan45. Regardless of everything, just the fact that the CPU cores are reaching a max temp above 85C while gaming, and the fact that my GPU temp reaches 81C, this warrants cleaning and repasting. Even returning the unit. These are fresh screenshots from a short gaming session just now, my CPU cores reached 91C and GPU reached 81C. The CPU throttling is around 600Mhz, and occurring in 3 CPU cores.

High CPU Temp.jpg High GPU Temp.jpg

But the real thought provoking issue here is that I start experiencing low and wildly fluctuating FPS in games before the temps reach 75C. Even at launch where temps are still around 45C, I experience lag and stuttering as you can watch in the youtube video i posted earlier. That's why I need your help to figure out if repasting and cleaning is enough to solve the problem. To me, it sounds more than that.
 
Last edited:
It's funny, I just saw a similar thread in the geforce forums and they said what I said what I just said https://forums.geforce.com/default/...ming-fps-gpu-throttling/post/5236525/#5236525
If you are running Windows 10 and have recently been updated to Creators edition, then this might explain your problems. I and a lot of others have suffered exactly the same symptoms with random FPS drops. Microsoft has admitted there is a problem but no fix as yet.

It's odd the new update was supposed to correct issues that arose from their previous "creatos" update but ........

Your laptop is getting warm. Be sure to use it on a hard surface or better yet on a cooler never on a bed or anywhere else that can reduce airflow and introduce more dust and hair than necessary. They have a lot of hot HW packed into that tight space and it needs as much cooling as it can get. Whatever you do don't open that laptop if it's still under warranty.
 
Guys, from a technical perspective, something is not right. I understand that many of you are diagnosing this as thermal throttling which requires repasting the CPU\GPU and cleaning the fans but here's what is actually happening: (nvidia GTX1070, i7 6700HQ)

When I bring the fans up to max RPM using Cooler Boost, the CPU cores do not go beyond 69C and the GPU is stable at 63C. However, the games are not running at their max possible framerate like they used to and the framerate keeps fluctuating severely in the lower range (30 to 70 fps on old games that should ideally run at stable 90-100fps).

How do you explain that? Yes, thermal throttling is happening at normal fan mode, but not at max fan mode, yet the drop in framerate is still there. Even if I send it to MSI for repasting I do not think the problem will be solved. Yes I tried a clean install of many nvidia drivers including the one which was working fine all year long but that did not fix the problem. I can't find a reasonable explanation. And no it is not a creator's update issue because i was gaming fine after the update. This started to happen only 1 week ago.
 
I thought perhaps it was thermal throttling but I have said from the beginning I think it's a SW issue. Did you read my last post? Many have been having issues since the last MS Win 10 update. If this just started happening recently that's likely the issue. You have two options wait until MS fixes it OR check the updates and remove that one if possible. Might not be that easy with a large up date but there should be a restore point made on your system and that should work.
 
That is all I see in Windows update history. Nothing goes back to October 26-27-28, the date of the problem. Should I look somewhere else for installed updates?

up2.jpg up.jpg
 
Back