NVIDIA GeForce RTX 2060 Review: Reasonably Priced Ray Tracing
GeForce RTX 2060 - Taking Turing Mainstream
Just like its x60-branded predecessors, the GeForce RTX 2060 offers the very same features as its higher-end counterparts, but the GPU is pared down somewhat to lower the power requirements, performance target, and ultimately the price. Here’s what the GeForce RTX 2060’s specifications looks like alongside the GTX 1060 and a couple of its big brothers. Take a gander at the number and then we’ll dive in, check out the card itself, and see how it performs...
|
|
The GeForce RTX 2060’s design language is virtually identical to the higher-end, NVIDIA-built GeForce GTX 2070, 2080, and 2080 Ti. They all feature a silver, curved shroud, with stylized back-plate, and dual axial fans. A matte-black, center section is adorned with the chrome “RTX 2060” logo and a lighted “GeForce RTX” badge is situated at the top.
The GeForce RTX 2060, however, is somewhat shorter than other RTX 20-series cards at only 9.5” (including the case bracket), and its GPU has a few functional blocks disabled. Although it’s packing a TU106 like the 2070, six Streaming Multiprocessors (SMs) have been disabled, along with 20% of its Tensor and RT cores. All told, the RTX 2060 has 1,920 active CUDA cores, with 240 Tensor cores, and 30 RT cores – down from 2,304, 288, and 36 in the RTX 2070. Of course, the GeForce GTX 1060 doesn’t feature any Tensor or RT cores, and it has only 1,280 CUDA cores. Although the GeForce RTX 2060 seems like the next-gen cousin the 1060, the RTX 2060 is significantly more powerful and more in line with the GeForce GTX 1070 Ti and GTX 1080.
The RTX 2060 has 6GB of GDDR6 RAM with the same 14Gbps effective data rate as the RTX 2070, but the 2060’s memory interface has been scaled back to 192-bits wide, which brings memory bandwidth down to “only” 336.1GB/s. The number of active ROPs and Texture units in the RTX 2060 has been reduced as well to 48 and 120, respectively, versus 64 and 144 in the higher-end RTX 2070.
GPU frequencies have also been tuned with the RTX 2060. Without overclocking, its GPU will boost up to 1,680MHz, which is higher than a base RTX 2070, but somewhat lower than a Founder’s Edition card. All told, the changes introduced with the GeForce RTX 2060 bring the TDP down to 160W. As such, it only requires a single supplemental PCI Express power connector, which is situated along the back of the card.
Speaking of connectors, you’ll notice that there is no NVLink bridge on the top of the RTX 2060, so no multi-GPU setups with this card. We know that’s a bummer for some of you, but the multi-GPU attach rate for mid-range cards was much lower than higher-end configurations, so NVIDIA made the decision to just not support SLI with lower-end RTX series cards at this time.
The outputs on the GeForce RTX 2060 have been tweaked versus the GeForce GTX 2070, 2080, and 2080 Ti as well. A DVI connector returns on the 2060, accompanied by dual full-sized DisplayPorts, and HDMI port, and USB-C / VirtualLink connector, which offers a 4 lane HBR3 DisplayPort and USB 3.1 Gen 2 over a single connector.
Over and above the speeds and feeds outlined here, the Turing architecture the RTX 2060 GPU is based on features a number of enhancements versus the previous-gen Pascal architecture used on the GeForce 10-series cards. We aren’t going to dive into the architecture again here though, so we’d strongly suggest checking out our initial coverage of the Turing Architecture and our review of the first-wave of GeForce RTX-series cards.
And now, let’s see what the new GeForce RTX 2060 can do...