Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Drivers and Overclocking Software › GTX 770 halved Framte/Benchmark score
New Posts  All Forums:Forum Nav:

GTX 770 halved Framte/Benchmark score

post #1 of 13
Thread Starter 
Hey Guys,

I know my Gigabyte Windforce 3XOC GTX 770 is a bit old now. But it still does the job quite well. Or at least most of the time.

First to my System.
CPU: i5 3570k @ 4400Mhz
MB: Gigabyte Z77X-D3H (BIOS: F18i but I had the same problem with BIOS F16)
RAM: Kinston Hyper X Blu 2x4GB 1333 @ 1600 9/11/11/29 (I know could be better but it's stable right now.)
PSU: LC-Power Hyperion Fury 8700 v2.2 700w
Two 2TB HDD's
Two SSD's
Two ODD's
One Soundcard

Monitors:
1. QNIX QX2710LED 256ßx1440 @ 108Hz
2. Acer FHD @ 60Hz

Now to my Problem.
Since I upgraded my Monitor I try to OC my GPU as much as possible to get ever last frame I can get. And while tinkering around with MSI Afterburner I encountered a small problem. When I crank the settings to high, not as high as the driver would crash, I will lose half of my performance.

I ran a few times Vally Benchmark 1.0
valleybenchmarkl3oxr.png

I didn't change the settings during the test.

1. Run with my "stable" setting. There are higher settings I can run but it's summer so with 30-35 °C ambient it's getting to hot.
valleystable1gnold.png

2. I cranked the settings as high as possible befor the driver crashed. That wasn't necessary it also happens with lower settings. But I didn't want to wait 30-40 minutes running Benchmark before I encounter the problem.
In the beginning it was fine but after 60% my performance dropped by arround 50%.
valleyfirstrun.unstabhvq6h.png

3 + 4. I ran the same settings again. Twice.
valleysecondrun.unstaxoof9.png

valleythirdrun.unstabizpnm.png

5. I ran the setting from the first run. I forgot to overlay Afterburner in this screen. But I ran the same settings. You may be able to see it in the Monitoring.
valleystablesettings2ahomo.png

I also had HWMonitor running in the Background the whole time.
hwmonitordzpwk.png


It only bothers me a bit that I can't run the higher settings. It's more that I hate it not to know what's going on here. Maybe someone of you encountered a simular problem befor, if you did... I coudn't find anything for this Topic in the whole observable Internet. (Mabye I just don't know what to search for)

Does someone have a clue what's going on here?
I would love to get some help.


Best regards,
DesAnderes


EDIT: I run DriverVersion: 368.81 but I had the problem also 1 year ago. I also patched my Driver with THIS TOOL found in THIS THREAD.
Edited by DesAnderes - 7/19/16 at 4:33am
post #2 of 13
I just want to get this completely clear:

After a period of time, your 770 benchmarks get progressively worse and bottom out at around 50% of what they initially did. This is irrespective of overclock, you just showed it happening on a higher overclock to make the issue occur faster.

If the above is true, then you are experiencing a deterioration of performance over time that I have not seen with a graphics card unless it was thermally throttling. Looking at your temperatures, it appears that you are hitting roughly 84C while benchmarking, I am not sure if you are consistently hitting that high of a temperature or not just by looking at your graphs, but it appears that way. Despite this, it also appears that your card is still running at the clockspeed it was set to, which indicates no thermal throttling.

To make sure your temperatures are not causing this, I would run the benchmark with your fans set to run at maximum, and see whether 1. the temperatures do not exceed 80C and 2. whether you still see a decrease in performance. Thermal throttling with GPUs typically occurs around 80C, with some cards it happens regardless of what temperature limit you have set.

Having said all of that, here are some potential causes if the above statement isn't entirely correct.

Bad overclocks can still pass benchmarks, but with worse scores than normal. In other words, they are unstable, but not unstable enough to crash. Benchmarks can be affected by bad overclocks even when you set the card back to stock - you have to restart the computer after you change clocks to ensure that no safety checks have been put in place from previous overclocks during the same session.

These were all the things I can think of - if none of these things help/work, try running the benchmark with your GPU at stock, after a fresh restart of your computer. If you still see a performance decrease then, and temperature has nothing to do with it, then this is a unique situation that I have not seen before. See if it happens in other benchmarks/games as well if you can't work out the reason.

EDIT

I would also use the Heaven 4.0 benchmark over the Valley one for stability. I had a 770 once and found that it crashed much faster in Heaven than it did in Valley, and so was more useful for finding maximum overclocks quickly. Especially for the core clock. For memory clock I used Firestrike's test 1 in a loop, if the men clock was set too high it would show green artifacts pretty quickly. I still use Firestrike for that.

Oh, and gratz on your first post! smile.gif
Edited by d0mini - 7/19/16 at 5:03am
My baby
(21 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.9GHz, 1.36v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) EK Supremacy Evo & VGA Ek xres 140 pump 
CoolingCoolingCoolingOS
4x Gentle Typhoons 1850rpm intake 4x Gentle Typhoons 1450rpm exhaust PE-240, XT45 120, SR-2 360, G2 240 slim rads windows 10 
MonitorKeyboardPowerCase
Acer Predator XB270HU Corsair Strafe RGB EVGA G2 850 PSU Suppressor f51 
MouseMouse PadAudioAudio
Logitech G502 XXL neoprene mousemat goodness Bose companion 20 Sennheiser HD 650 
Audio
GSX 1000 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
My baby
(21 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.9GHz, 1.36v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) EK Supremacy Evo & VGA Ek xres 140 pump 
CoolingCoolingCoolingOS
4x Gentle Typhoons 1850rpm intake 4x Gentle Typhoons 1450rpm exhaust PE-240, XT45 120, SR-2 360, G2 240 slim rads windows 10 
MonitorKeyboardPowerCase
Acer Predator XB270HU Corsair Strafe RGB EVGA G2 850 PSU Suppressor f51 
MouseMouse PadAudioAudio
Logitech G502 XXL neoprene mousemat goodness Bose companion 20 Sennheiser HD 650 
Audio
GSX 1000 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
post #3 of 13
Thread Starter 
Quote:
Originally Posted by d0mini View Post

I just want to get this completely clear:
After a period of time, your 770 benchmarks get progressively worse and bottom out at around 50% of what they initially did. This is irrespective of overclock, you just showed it happening on a higher overclock to make the issue occur faster.

No. It does not progressively get worse over Time. I have 100% performance to a certain point and my performace suddenly drops by 50%.
Quote:
Originally Posted by d0mini View Post

Bad overclocks can still pass benchmarks, but with worse scores than normal. In other words, they are unstable, but not unstable enough to crash. Benchmarks can be affected by bad overclocks even when you set the card back to stock - you have to restart the computer after you change clocks to ensure that no safety checks have been put in place from previous overclocks during the same session.

I didn't restart my PC betwen the runs. After a restart everything works normal again. So a safety check could be a possiblity
Quote:
Originally Posted by d0mini View Post

These were all the things I can think of - if none of these things help/work, try running the benchmark with your GPU at stock, after a fresh restart of your computer. If you still see a performance decrease then, and temperature has nothing to do with it, then this is a unique situation that I have not seen before. See if it happens in other benchmarks/games as well if you can't work out the reason.

After a restart everything is fine. I encounter the problem also with the CineBench R15 openGL test as well while playing Battlefield 3/4. The Driver never crashed on me with these Settings. The problem also seems to lower my Frames outputet to my Displays by 50%.
Quote:
Originally Posted by d0mini View Post

I would also use the Heaven 4.0 benchmark over the Valley one for stability. I had a 770 once and found that it crashed much faster in Heaven than it did in Valley, and so was more useful for finding maximum overclocks quickly. Especially for the core clock. For memory clock I used Firestrike's test 1 in a loop, if the men clock was set too high it would show green artifacts pretty quickly. I still use Firestrike for that.

I use Valley cause I got i few paper chart with settings and scores from the last 3 years. But I will do a run with Heaven 4.0 to check if it's fine.




I left Valley running a few minutes before i started benching. So the performacedrop already happened.

1.
01x8s1j.png

2.
024gsoj.png

HW Monitor:
hw9vsly.png
post #4 of 13
Thread Starter 
So I just benched Heaven a few times. And the results do not match my expectations.

These are my settings and again I didn't change them during the runs. I also did not change any settings. I ran all runs straight throught leaving my system just a few seconds to cool down.
heavensettingsyiub4.png

I first made to runs with my "stable" OC to get a baseline
heavenstable13fuli.png
heavenstable29cuj8.png

Seems allright so far.

Then I cranked up my settings.
I ran 3 runs because I was surprised with the results.
1. run seemed fine and gave more or less the performance the cards should give at this speed.
heavenunstable1seugl.png

2. run had a decreasing score.
heavenunstable2f9uy0.png

3. one bottemed out.
heavenunstable3h0uqw.png


This time I closed HWMonitor without making a screenshot. But The GPU Max was at 78 °C. I also noticed that that was the peak and the card went down 8-9° after hitting this point. So a safty "feature" could be possible. Have to check if the problem also occures if i'm at my Stable OC with 80°C+.

I don't have the time now to do more runs. But later I will get the performance drop with valley and will than run heaven again. Just to finde out what happens.
post #5 of 13
So after a restart things were as normal and then after a while they went back to being weird..?

Test the benchmarks with no overclock and after a restart. If there is a problem then, then there is a serious issue with either your drivers or your graphics card.

EDIT

And who knows, it could be your PSU like Duality is saying! I couldn't find a review on it, but I didn't look particularly hard.
Edited by d0mini - 7/19/16 at 11:08am
My baby
(21 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.9GHz, 1.36v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) EK Supremacy Evo & VGA Ek xres 140 pump 
CoolingCoolingCoolingOS
4x Gentle Typhoons 1850rpm intake 4x Gentle Typhoons 1450rpm exhaust PE-240, XT45 120, SR-2 360, G2 240 slim rads windows 10 
MonitorKeyboardPowerCase
Acer Predator XB270HU Corsair Strafe RGB EVGA G2 850 PSU Suppressor f51 
MouseMouse PadAudioAudio
Logitech G502 XXL neoprene mousemat goodness Bose companion 20 Sennheiser HD 650 
Audio
GSX 1000 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
My baby
(21 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.9GHz, 1.36v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) EK Supremacy Evo & VGA Ek xres 140 pump 
CoolingCoolingCoolingOS
4x Gentle Typhoons 1850rpm intake 4x Gentle Typhoons 1450rpm exhaust PE-240, XT45 120, SR-2 360, G2 240 slim rads windows 10 
MonitorKeyboardPowerCase
Acer Predator XB270HU Corsair Strafe RGB EVGA G2 850 PSU Suppressor f51 
MouseMouse PadAudioAudio
Logitech G502 XXL neoprene mousemat goodness Bose companion 20 Sennheiser HD 650 
Audio
GSX 1000 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
post #6 of 13
I'm pointing the finger at your sub-par quality PSU that cannot deliver an adequate load to your GPU.

Try with another PSU?
post #7 of 13
Thread Starter 
Quote:
Originally Posted by d0mini View Post

So after a restart things were as normal and then after a while they went back to being weird..?

Test the benchmarks with no overclock and after a restart. If there is a problem then, then there is a serious issue with either your drivers or your graphics card.

With no Overclock or even with my +50 Core and +150 Memory everything is fine. And it's stil fine after several hours of gaming, etc.
The GPU also keeps it's performance by 80°C+ 50 Core and +150 Memory so it doesn't seem to be a thermal problem.
Quote:
Originally Posted by Duality92 View Post

I'm pointing the finger at your sub-par quality PSU that cannot deliver an adequate load to your GPU.

Try with another PSU?

I thought about it but I don't know enough about PSU's to nail down the Problem. I also know that it's far away from beeing a good PSU but Money is a Problem right now.
My PSU got 3 PCIe Powerconnectors. Do you think if it could help using a y-Adapter from 2x6 PCIe to a PCIe8 connector to spread the load for the PSU?
post #8 of 13
Quote:
Originally Posted by DesAnderes View Post

With no Overclock or even with my +50 Core and +150 Memory everything is fine. And it's stil fine after several hours of gaming, etc.
The GPU also keeps it's performance by 80°C+ 50 Core and +150 Memory so it doesn't seem to be a thermal problem.
I thought about it but I don't know enough about PSU's to nail down the Problem. I also know that it's far away from beeing a good PSU but Money is a Problem right now.
My PSU got 3 PCIe Powerconnectors. Do you think if it could help using a y-Adapter from 2x6 PCIe to a PCIe8 connector to spread the load for the PSU?

I doubt it very much, It goes to the same PSU. It's not the connection that's the problem, it's the power supply itself. Maybe you have a freind or a shop close that could help you test it?
post #9 of 13
Thread Starter 
Du to the fact that I work in a PC Store I could probably test my card there. biggrin.gif

Did you experience something like that before?
post #10 of 13
Quote:
Originally Posted by DesAnderes View Post

Du to the fact that I work in a PC Store I could probably test my card there. biggrin.gif

Did you experience something like that before?

a lot of people have experience this, I have not personally.
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Drivers and Overclocking Software › GTX 770 halved Framte/Benchmark score