Screwed Up Overclocking.. HELP!

Shaan5

Reputable
Mar 1, 2015
577
0
4,990
I Have A G1 Gaming GTX 970 and I was overclocking. I pushed it too far and Valley Crashed.

All my overclocks were reset. But my driver is still up to date
Everything in the Nvidia Control Panel are the same and the temps are also the same.
Just benchmarked it at stock and and the FPS Is the same as it was before.
It only crashed when I turned on V Sync though. It was fine with it off.
I also noticed that the FPS Would CAP At 30 in some scenes and at 60 in others.
I'M FREAKING OUT RIGHT NOW
I'm Worried that My PC is Going to break or something.. HELP!
Is This A big deal or am I Fine
Help!
 
OK this is because you pushed the Card to far and it became unstable. This is common when pushed to far.

TO help save you some time Download OCCT. Use just teh GPU testing part of this program as teh rest of it is unreliable. Run the GPU test with Error Checking enabled and shaders set to 3 For Nvidia and 7 for AMD cards.

Up one clock at a time by 20's and test for 30 mins each time till it errors then back it down by 10's till it stops erroring. THen up it by 1's till errors again then back it down 2 and that will be teh highest stable OC that card will have.

write that number down and reset to default. and follow the same for the rest of the clocks.

Once all max clocks are found input them and test for 1 hour to be sure it wont error. If it does error back teh clock down by 2 and retest. THis will get you your highest OC possible.
 


Some programs will limit scenes to 30 FPS and others 60 and ones with V-sync off will go over "if" the GPU and CPU are powerful enough. I think you are just worried because your PC crashed due to a hardware instability. Once your clocks were reset to default the instability went away. Follow the way I mentioned for OCing your GPU and you will be fine.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


So I'm Fine..

Is there any chance that this screwed up my PC or anything like that
 
Very very unlikely that you hurt anything. Crashes often happen out of self preservation. The PC is stopping itself to prevent damage. To help prevent this in the future please follow the guide I laid out for you on OCing your GPU. It is the safest way to get the max out of your GPU when OCing.
 

gokitty199

Honorable
Feb 4, 2013
1,398
0
11,660


doubt it, it crashed because you pushed the card to far as stated above so it became unstable, all it did was reset itself to its default speeds, you would need to reapply the settings of the overclock to get it overclocked again
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990




3 Things
1. Whenever I Turned up the Core Clock. It would only very slightly increase performance. Maybe by about 0.1-0.3 FPS Which seems Low. After Pushing it by +120 To The Core. I Only saw an increase of 2 FPS or so

2. In GPU-Z. It says the Card is Clocked at 1329Mhz but, In MSI Afterburner, It says the card is Clocked at 1380Mhz

3. I Did Not notice any Graphical errors before the Crash Which also seems weird

Any Ideas

I Just Ran Benchmarks and Tested Temps and It Seems Like Everything is fine But I'm still really Worried about it

Thanks

 

gokitty199

Honorable
Feb 4, 2013
1,398
0
11,660


download gpu-z and see what your boost clock speed is. by default the boost clock should be around 1350ish for the gigabyte g1 i think so you might not be increasing it by much. how far did you go up in your msi afterburner? like is your core clock at +150 and is your memory at +300 or what
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


In GPU-Z The Default is 1329 however, in MSI Afterburner. it says 1380

I Increased by about 150 on the Core and I didn't touch the Memory Clock

 
1. Whenever I Turned up the Core Clock. It would only very slightly increase performance. Maybe by about 0.1-0.3 FPS Which seems Low. After Pushing it by +120 To The Core. I Only saw an increase of 2 FPS or so

That is about normal for some programs. Some programs respond to OC's better than others. OCCT will check if you have errors that you can not physically see. THis is why I recommend this program. Having errors can also hurt the performance so a stable OC is always the goal UNLESS you are just trying to push for te hbest benchmark score possible from your system .

2. In GPU-Z. It says the Card is Clocked at 1329Mhz but, In MSI Afterburner, It says the card is Clocked at 1380Mhz

I do not care for GPU-Z myself as I have hound it can give errors from time to time. In a situation like you have use a third program to check the speed to see which one is correct. Since Afterburner is the program adjusting the clocks my guess would be that it the correct one.

3. I Did Not notice any Graphical errors before the Crash Which also seems weird

Just because you do not physically see errors does not mean that they are not there. This is where OCCT's error check is the best way to find stable.

I Just Ran Benchmarks and Tested Temps and It Seems Like Everything is fine But I'm still really Worried about it

If the clocks are at stock you are fine, If you have not tested with OCCT then I recommend doing so to be sure the OC is stable before proceeding.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


So I Have OCCT and Have No Idea what it means

I Chose DirectX as well in the Installation

Does the on Button Start the Test

What should I put in for be MB When I hit error Checking


 
once OCCT is installed run the program. There will be tabs, like windows tabs, select the GPU test tab. In teh middle of the program screen are the test options. Select Direct X 11, make sure the error check box is ticked, Shaders set to 3 and then press the ON button and let it alone for 30 mins.


the way you discribe it sounds like you have not installed it, If this is the case install teh program.


EDIT: the program will not register more than 2000 MB no matter what you do so set it to that.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


I Have OCCT. There is no "GPU Test Tab" but there is a GPU 3D Tab
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


I've messed around a bit in OCCT and I think I get it

Should I cap the frame rate or no. It's hard to monitor temps if only 27% Of my GPU is being used or should I Run a program like Heaven after for temps
 
Do not limit the frame rate, You want the GPU to run flat out full bore. IF you have set the settings properly, including the screen resolution, it should load the GPU fully and not to just 27%. There is a monitoring tab also, it's on the side, that you can set up to monitor temps in OCCT. It actually uses HW monitor and is fairly accurate.

IF the error check is enabled the fuzzy donut will not spin. This is normal for the maxium stress test.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


It is limited to 27% Because the FPS Is limited to 60
Screen Resolution is set to 1920x1080
Should I run in fullscreen
When I enable Error Check, It asks me about Memory used
Default is 60MB

I also noticed that my GPU Temps hit 70C In OCCT in under a minute and When Running OCCT, it says 1367MHz in MSI Afterburner compared to Heaven where it says 1380

I also noticed that my FPS Is Marginally better since I was reverted back to Stock

Before I would get 54.4 FPS and now I get 55.3 FPS In Heaven

Am I Just overthinking this

Thanks
 
Do not limit the FPS.
Do not run full screen because it could cause an error or cause the program to crash, Not sure why but I have noticed this and it does not effect the results.
The difference in clock speed is due to the boost/turbo option the Chip has. the boost only kick in "if" there is enough temp head room.
Do not worry about FPS in OCCT just let the program run for a half hour to see if you error. Also make sure you do not exceed 95°C during the test run.

If you believe you have enough cooling and temp head room you can up teh Power Limit. this adjusts the temp at which the GPU starts to throttle the core speed to prevent overheating.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


I've heard some people saay that OCCT is responsible for killing a lot of GPU's and that it is unreliable.

Is OCCT Safe?
 
Yes it is safe as long as you follow normal common sense things. Any time you stress test a part watch your temps. DO NOT let it overheat. Most cards that die from stress testing is caused by the VRM overheating and melting. This has long been cured and does not happen as much since the 200 series from Nvidia so you do not need to worry about this.

The only reliable test from OCCT is the GPU test. This is why I said earlier not to use the other tests. Quoted directly from that post "Use just teh GPU testing part of this program as the rest of it is unreliable." . This test is much more reliable than any other GPU test out there because it will tell you is there are errors. Other programs you need to watch for graphic anomalies and test the OC in many different games doing the same thing. OCCT is a one stop test that is sure to get you to a stable OC.

I understand that you are new to OCing and are concerned about what your doing. This is normal. OCing teh GPU the way I mentioned in this thread will get you to your max OC safely with out damage as long as you follow common sense things like watch your temps, no big clock jumps, no large voltage jumps if applicable and test thoroughly.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


Ok I think I'm ready to give it another shot

1. Record FPS In Ungine Heaven and Temps(Done)
2. Max The power limit
3. Increase Core by 30 and run OCCT Test(Like you mentioned for 20-30 mins while monitoring temps)
4. Run Unigine Heaven and note FPS and Temps
5. Rinse and Repeat until OCCT Notes errors or other issues occur and decrease clock by 10 until issues stop, then up by 5 and then down by 1 until issues stop
6. Record Core increase and reset
7. Increase Mem by 25 and do the same as with the core(Monitoring temps and issues)
8. When unstable Mem Speed is found, Decrease by 10 until there are no errors, then up by 5 and then down by 1
9. Record numbers and reset
10. Input Numbers into Afterburner and Run Benchmarks/Play Games!

Note: I will most likely also run a few other benchmarks through the process such as the Metro Last Light Redux Benchmark just to make sure everything is as it should be

Is This Correct

This is my first OC so is it okay if I ask you if I have any issues

Thanks

EDIT: Will the test in OCCT stop if any errors are found

 
No the test will not stop if it find errors. IN the upper left corner when you start the test you will see a green ERROR written out. IF this turns red and has numbers after it then it found the OC to be unstable.

Do Not max out the power limit. Doing so will stop the card from throttling the speed to prevent over heating. Upping it as necessary is a better way to do it. IF you find that the card is coming out of boost mode OR the frequency is dropping dramatically and bouncing back up then up the power limit and watch the temps. Dont go over 95°C. IF you hit this or over lower the power limit so that it starts to throttle the core speed around 88-92°C. this will prevent overheating during game play.

There is no need to test with benchmarks while OCing using OCCT because it will catch the errors before the benchmarks will be effected. This is not to say you can't do it if you want. Just that it is not necessary to do it.

Do increases by 20's till you error because much more you can easily push to far and the system OR driver will crash. This also is not to say that it wont happen even doing it by 20's but there is less chance of crashing the whole system and not just the driver.
 

Shaan5

Reputable
Mar 1, 2015
577
0
4,990


The G1 Gaming GTX 970 only goes to +112 on the power limit and I Don't think that I Could lower it without causing instability seeing as I have already hit my Max OC to the core(+130) With the Max power limit

also, Whenever I push the GPU Memory past +270, OCCT will run fine for 5 Mins but then 16 Errors will appear out of Nowhere,
+270 Seems rather low so are there any issues with OCCT and GPU Memory OC

Thanks

 

TRENDING THREADS