Have you removed the gigabyte oc guru ii software or are you running that and afterburner
I never installed it. Only Afterburner. My card wasn't from Gigabyte, but I would only use Afterburner anyway.
downloaded afterburner what do i do now
Uninstall OC Guru, install Afterburner.
Run a benchmark at stock settings to (a) confirm everything is working and (b) get a baseline to compare any other results to. I used the Heaven benchmark from Unigine - it's a free download and good for DirectX 11.
Afterburner has some very useful monitoring settings in Settings (gearwheel icon on the main display) under the monitoring tab. I think the key ones for you will be GPU temp and power limit (if it's available for your card, which I'm not sure of). When I did my 7950 with an older version of Afterburner, I had to monitor several settings to see if the power limit was being hit, but that power limit setting does it all for you.
I also used HWINFO64 to measure the VRM temp on the graphics card because Afterburner doesn't measure that and it can be an issue with some coolers. I'd disable monitoring in Afterburner at that point, so I can't say for sure if you can run both simultaneously.
After the benchmark run, bring up Afterburner and look at those settings. GPU temp max will show as a figure in addition to the graph of GPU temp, but power limit is something you'll need to look at the graph for because it only has two values. 0 means your card didn't hit the power limit, 1 means that it did. So you'll need to look at the graph right after the benchmark finishes to see how much it was hitting the power limit during the benchmark because even a fraction of a second at the limit will show as a max of 1.
If you are hitting the power limit a lot (and I expect you will be), gains from overclocking will be less than you'd expect because your card is being power throttled and further overclocking will increase power draw and thus increase throttling.
One thing you can do is to increase the power limit, which is a simple slider on Afterburner's main screen. That increases the amount of power your card can draw without being power throttled. It won't increase clock speeds or voltages, only the maximum power draw. Doing so will probably cause your card to exceed its specified power draw, so it counts as overclocking in terms of risk and warranty (if your card had one). It might cause over-spec power draw from your motherboard's PCI-E slot and/or your PSU's power cable too. Do it at your own risk, like all overclocking. It will
probably be fine. It's not much over spec. But its at your own risk.
Run the same benchmark again, note the results again, especially the power throttling and GPU temp.
If you do that and you're still being power throttled, I'd suggest
decreasing voltages a bit while keeping clock speeds the same.
Repeat the above two steps until you either stop being power throttled more than rarely or your card fails to complete the benchmark. If the run fails, nudge the voltages back up to the last successful settings, then add a little more for stability if you're intending to use those settings for more than one benchmark. I got mine down to 1.0V for GPU and 1.5V for VRAM, but it will vary from card to card.
If you get a stable run without power throttling, start increasing clocks a little at a time, monitoring GPU temp and power throttling for each run.
You should change GPU and VRAM settings seperately, benchmarking after each change.
In short, it's about getting max clocks while not exceeding power draw limits or temperature limits. In my case, the optimal settings were 950/1275 (a tiny overclock on the GPU only) with GPU at 1.05V and VRAM at 1.5V. But your card will probably be different.
You'll see references to 7950s running over-volted and with GPU clocks as high as 1200 and VRAM as high as 1600.
I got mine to 1100 and 1575, which were the max settings available using the version of Afterburner I was using, and it did that with stock volts...and the benchmark scores were lower than they were from undervolting instead of overclocking and the temps were much too high for my liking and the noise was much too high for my liking.
But if you want to be thorough, you could try both approaches and see which works best for you. Different hardware might get different results. Or try a mix of both rather than max overclocking or max undervolting at stock clocks.
