Ray tracing for the masses.
Nvidia has unveiled what will likely be the final RTX GPU in its Turing architecture lineup; the $349 RTX 2060 Founder’s Edition. The appeal of this GPU is that is is the most affordable video card you can buy that supports Nvidia’s new technologies, namely ray tracing and DLSS (super sampling). At roughly $250 less than the next card in the stack, the RTX 2070, that’s a compelling offer assuming it can actually do ray tracing well. Nvidia is sweetening the deal even further with a promotion that offers a copy of Battlefield 5 or Anthem as well for every RTX 2060 it sells, but it’s not clear how long this offer will last. The RTX 2060 goes on sale on January 15th, both in Founder’s Edition trim and from Nvidia’s partners.
RTX 2060 – Design and Features
The RTX 2060 features the exact same dual-fan design we’ve seen on previous Founder’s Edition cards, so there’s not much here to gawk at. The card is the exact same physical size as its big brother, the RTX 2070, and also uses the same TU106 die, albeit cut down somewhat. Here’s a look at the specs and the 2070 alongside it.
Easily the biggest “downgrade” from the 2070 is that it only has 6GB of GDDR6 memory, and 20 percent fewer CUDA cores. It has a narrower memory bus as well, and TDP is down by 25w compared to its big brother, yet it still requires a single eight-pin connector.
Like previous RTX cards this GPU does support ray tracing and DLSS, and Nvidia claims it can hit “nearly” 60fps in Battlefield 5 with Ultra ray tracing enabled, or well beyond 60fps with ray tracing set to Medium, so that’s enticing for some people who want to get in on the ray tracing action without dropping $500. Overall this GPU is designed for 1080p and 1440p gameplay, just like its predecessor, the GTX 1060. However, since the RTX 2060 uses faster 14Gb/s memory (compared to the 8Gb/s memory in the 1060), memory bandwidth has increased by 83 percent, which is quite a leap in just one generation.
Otherwise, there’s nothing really “new” to talk about with the RTX 2060. It’s the same GPU you’ve seen before, just with a bit less power and memory than the previous GPU, and at a lower price point. It’s main selling point is it’s the first “affordable” RTX card, though $350 is still a chunk of change, no doubt.
RTX 2060 – Benchmarks
To see how the RTX 2060 performs I loaded it into our GPU test bench, which consists of an Intel Skylake Core i7-7700K CPU, 16GB of DDR4 RAM, an Asus Z270 motherboard, an Intel SSD, and an EVGA power supply. I tested the GPU at the three most common resolutions, with all settings maxed and anti-aliasing disabled. All tests were run in DX11 mode.
Though Nvidia claims the RTX 2060 performs on par with the GTX 1080 in its press materials, in my tests it was about equivalent to the GTX 1070 Ti, which was certainly not a midrange GPU nor was it a performance slouch. Also, as of press time, the GTX 1070 Ti was selling for $100 more than this GPU, which makes the RTX 2060 look pretty good. This means the “midrange” has now officially moved beyond 1080p and into very playable 1440p gaming. In every gaming test I ran aside from Shadow of the Tomb Raider and Monster Hunter: World, the RTX 2060 was easily beyond 60 frames per second (FPS) for fluid gaming. And in those other two titles, it would not take much tweaking to get the frame rate past 60fps.
As far as the card’s ray tracing abilities go, I am happy to report it can handle Ultra DXR settings in Battlefield 5 at 1080p, as in my tests it averaged about 50fps the entire time with in-game settings on Medium and ray tracing set to Ultra. Personally, I’d settle for less ray tracing and higher FPS, but the good news is this GPU is certainly not a dog in that regard. I’m sure it would be totally fine for single player gaming, actually, but note that you’ll be relegated to 1080p gameplay. Not a huge surprise given this card’s price point.
RTX 2060 – Overclocking
To see how far I could push this GPU I fired up Heaven 4.0 and let it just run for about a half hour to see how high the GPU Boost would go all on its own. As it turns out it did a pretty good job, clocking up to 1,860MHz and running at 70C. That’s essentially expected behavior for an Nvidia GPU, so all in all it was not surprising. Using EVGA’s PreciousXOC, I gently began to nudge the clock offset upwards in small increments, after raising the temperature and power limit to their maximum setting. After about an hour of fiddling with it I ended up with a final clock of 2,025MHz, and a maximum temperature of just 72C.
I ended up with a final clock of 2,025MHz, and a maximum temperature of just 72C.
That is a very good overlock, as most of the Nvidia cards I’ve tested over the years can usually only get into the high 1900s typically. As far as the performance boost afforded by this overclock, I ran a few tests and saw about five percent uplift, which isn’t much but also very typical and also expected. Overall I’d say this GPU is an excellent overclocker, though your results may vary.