Date: Tuesday , November 11, 2014
Continuing our coverage of NVIDIA's new Maxwell architecture and its current flagship the GeForce GTX 980. Today we are going to look at overclocked 2-Way SLI performance. The NVIDIA GeForce GTX 980 and the Maxwell architecture were released on September 18th, 2014 for $549. In our initial evaluation we found competitive performance that offered a better gameplay experience compared to the competition's current flagship video card, the AMD Radeon R9 290X. We also found that the GeForce GTX 980 also offers competitive performance to the GeForce GTX 780 Ti, giving us more performance, in a more efficient package.
We just recently published our NVIDIA GeForce GTX 980 SLI 4K Video Card Review using two GeForce GTX 980 video cards. Two of these video cards are very powerful, and we found that it does help at 4K. We also found that the competition’s AMD Radeon R9 290X CrossFire also does very well at 4K and gave GTX 980 SLI some fierce competition. It left us wanting more out of GTX 980 SLI, so that brings us to this review today.
We are going to overclock both reference GTX 980 video cards in SLI and see how these perform. We are also going to overclock two 290X video cards and two 780 Ti video cards. All three card comparisons overclocked as much as we can with consistent clock speeds and no throttling.
Following the launch evaluation we took an in-depth look at overclocking the GeForce GTX 980. It is important to read that evaluation first, before this one here today. There is something you need to know about in regards to the power limits and TDP that will directly affect the potential overclock on GeForce GTX 980 video cards. On this page read the "Re-thinking Overclocking" section. On this page read the "Overclocking" section. That will get you up to date on how we must overclock video cards in today's dynamic intelligently controlled power management video card world.
In summary, no longer is the focus on raising the voltage to the highest levels to achieve the best overclock. GPUs today have power management controls which cap the clock speeds and performance once the GPU hits a set TDP or power limit wall. Increasing voltage can make the GPU hit this wall faster, lessening your ability to raise the clock speed. It can also make the GPU run hotter, making the GPU hit its thermal throttling sooner with insufficient GPU cooling.
The focus is to first find what the starting GPU clock speed is at while gaming. Then, we raise the clock speed until it starts throttling backwards. We try combinations of voltage and see if it helps, or hinders performance. We find the balance of consistent clock speed that doesn't throttle backwards, but instead remains higher than the starting clock speed and above all remains consistent for long periods of gaming. Only this way can we say that the GPU is truly overclocked.
For this article we are only going to perform apples-to-apples performance comparisons. All video cards are tested at the same settings at 4K - 3840x2160 and NV Surround and AMD Eyefinity in a triple-display setup at 5760x1200 for multi-display gaming.
We are using two reference GeForce GTX 980 video cards for testing.
We are using two reference GeForce GTX 780 Ti video cards for testing.
We are using two XFX Radeon R9 290X Double Dissipation video cards for testing. While these are custom cooled video cards, these aren't the best cooled 290X video cards. If we were to have used two reference R9 290X cards we wouldn't be able to overclock at all since these can clock throttle even at stock settings depending on your cooling setup.