Did I overclock my GPU correctly?

[Samhfx]

First of all sorry if this is the wrong section. Wasn't sure where else to post this. I'll do my best to explain the steps I done in order.

Put 50% Fan Speed on MSI Afterburner to ensure I won't have any GPU temp problems.

Core Clock - Increased the core clock from 0 - +25% and opened Heaven Benchmark 4.0 and done a whole playthrough. After every successful benchmark playthrough I was increasing the core clock by +25 up until I hit a point where my PC/Driver would crash. My crashing point was +120 on the core clock so I immediately reduced the core clock back down to +110 then I done x2 benchmark playthroughs and everything seemed to be ok. - I set the core clock back to 0% to start on the memory clock separately

Memory Clock - I started at +100 on the mem clock and after every successful playthrough on the benchmark I kept adding another +100. When I got to +600 I crashed, so I took away 50+ and crashed again. Took away +25s until I stopped crashing. Managed to get +500 on the memory clock with no crashes.

Core and Mem clock settings - https://gyazo.com/2ef27eaffb3f733ef1a3fc5128ebef9e

I then put my highest core and memory clock back into MSI Afterburner together and managed to do x3 benchmark playthroughs without any problems. After benchmarking at default and overclocked these are the results, just wanted to check if this was normal and good?

Default - https://gyazo.com/dd77652368a6b5135b5a26d86d50c689
Overclock - https://gyazo.com/8fceacfe62257db5c8a3719e535c8ee4
Windows 10 Pro 64-bit Intel Core i7 @ 4.00GHz 16.0GB MSI Z170A GAMING M5 (MS-7977) (U3E1) BenQ XL2411Z (1920x1080@144Hz) 2047MB NVIDIA GeForce GTX 980 Ti (EVGA) 465GB Samsung SSD 850 EVO 500GB (SATA (SSD))
931GB Western Digital WDC WD1003FZEX-00MK2A0 (SATA ) EVGA Supernova 850w

My question now is should I take it to the next level and increase the core clock by +1 up until I crash to try and get the most out of my gpu or just leave it how it is? Also how safe or reliable would it for me to run the benchmark with the stable clocks I got above and go AFK for a while (for example go and shower and leave benchmark running in a loop)

Sorry for the huge thread this is my first time actually putting in the time to overclock a GPU. Do you think I will notice a better performance if I keep the overclock?

 

[mattyfez]

Assuming temperatures are ok it sounds fine.
You could try and push a little higher using smaller iterations until you hit a ceiling again..but your probably close to the limit by the sounds of it..

In terms of increased performance there will be some . You should notice higher scores in the heaven benchmarks for example.

Although looking at your screen shots it doesn't look to be much .. What your looking for is a higher minimum and average frame rate rather than max.

Although it's worth trying out some other graphics intensive benchmarks too, 3d Mark being probably the next obvious choice..

 

[bJN]

The only other options you've got left are power limits and core voltage adjustment which would aid getting a bit higher. Provided your current settings are stable and you're happy with the performance then I wouldn't worry about tinkering too much. As above, provided the temperature seems reasonable then no reason to not stick with what you've found.

 

[LuckyBenski]

In terms of overclock it sounds like you've found something pretty close to the sensible limits. Do a more thorough stability check and then leave it at that IMO.

Personally I would have checked the maximum temperature at stock clocks with default fan curve. Then the same with fans fixed at your chosen speed. This gives a baseline.

I don't feel comfortable with fans at a fixed speed unless I know the absolute limit temperature wise. By that, I mean running Furmark for maximum heat output until temperatures stop increasing. It can take a while to soak the case and saturate your entire system. You may choose to be less thorough but 2-3 runs of a benchmark won't show what happens after an hour of gaming at moderate loads. You need to know what happens when CPU and GPU are loaded for several hours, how does the case air situation respond, etc.

That is, unless you're finding GPU temperatures are stupidly low (<50° at all times). In that case maybe you can get away with not monitoring too closely.