The launch of the RTX 2000 family by NVIDIA has had a bittersweet reception: we have more power and a striking commitment to ray tracing and artificial intelligence, but the price of the RTX 2080 and 2080 Ti with which the family debuted has been very criticized by users.
NVIDIA nevertheless has been completing the range: first it did with the RTX 2070 and now it has just announced the NVIDIA GeForce RTX 2060, the most affordable model of this family. We have had the opportunity to analyze the performance of the RTX 2060 Founders Edition for a few days, and then we tell you everything we have discovered.
The advantages of Turing at a more affordable price
The NVIDIA GeForce RTX 2060 are one step below the RTX 2070 that precede them, and again represent an important jump in power and performance over the previous generation of NVIDIA graphics.
In fact, in general, the RTX 2060 pose a performance similar to that of the GeForce GTX 1070 Ti. In our tests we did not have this specific model, but we could compare them with the GTX 1070 which, as we will see, exceeded in an appreciable way in all the tests.
As you can see in this specification table, we have a graph that thanks to its TU106 12-nanometer GPU achieves more power without compromising efficiency: the 160 W it consumes is similar to the GTX 1070, and although the latter exceeds to the RTX 2060 in the amount of graphic memory, the greater bandwidth of the (very expensive) GDDR6 helps the performance increase with respect to those models.
The clock frequency of the GPU is reduced compared to the RTX 2070: from the 1680 MHz of the latter we pass to the 1365 MHz of the RTX 2060, a gap not too wide. The GTX 1070 is clearly superior to the GTX 1060 in performance, but in this generation the RTX 2060 is not “so different” from the RTX 2070, and of course its balance between price and performance is more promising.
NVIDIA and the problematic nomenclatures
We will see how in the tests that balance is certainly striking, especially taking into account that the RTX 2070 had a launch price of 639 dollars and the RTX 2060 are located in a much more attractive 369 dollars.
Maybe that price has been influenced by the numerous criticisms of the user community, but the truth is that we have some good candidates for many users who are looking for new graphics for their desktop PCs.
Here NVIDIA has a challenge in its way of communicating who directs these cards. Until this generation arrived, graphics with that number ending in xx60 were always a great option for those users with less budget: the GTX 1060 appeared at $ 249 in July 2016. Two and a half years later we find that its successors appear to prices of 349 dollars that were reserved to GTX 1070 in that previous family. What happened?
The answer is in those model numbers and those prices that do not seem well adjusted. The RTX 2060 is comparable to a GTX 1070 Ti, the RTX 2070 to a GTX 1080 and the RTX 2080 to a GTX 1080 Ti, but perhaps one would expect the price of an RTX 2060 not to be very different from that of a GTX 1060 when it appeared, and the same goes for the rest of the RTX 2000 family, which has taken a significant leap in price.
We already explained in our analysis of the RTX 2080 and RTX 2080 TI how there are strong arguments for prices to rise: the new graphs are more powerful, but above all they are more expensive to manufacture because they integrate both the nuclei for traditional areas and the dedicated to ray tracing (RT Cores) and artificial intelligence (Tensor Cores).
For many users these last two facets are an addition they had not asked for, and probably many would have settled for a pure power increase without more.
That would have made it possible to lower the prices with which all the graphics of the RTX family have appeared, but here NVIDIA wanted to take risks and try to make the leap towards a new era of graphic computing by bringing closer the ray tracing and that power of the Tensor Core to all audiences.
The initial reception of that decision has not been good in the comments of our topics, but as we said in all those topics, here it might be good to give the benefit of the doubt to NVIDIA. Perhaps in a few years the tracing of lightning has become the norm in the market and these chips could have been at fault.
If that bet fails or we are not interested we have options, of course: AMD is still a valid alternative although not so much if we demand more performance, but that could change in the coming months when the firm is expected to arrive with its new AMD Navi 7 nm. In 2020 the thing will be encouraged even more because we already know that Intel is preparing its own dedicated graphics to compete with those of NVIDIA and AMD.
Sacrifices on the RTX 2060
We spoke before the potential improvement, but the other part of the equation is added to it: the advantages of Turing also reach these graphs, although logically they do so with certain sacrifices.
The RTX 2080 Ti have a combined power of 78 trillion RTX-OPS, 60 for the RTX 2080 or 45 for the RTX 2070. In the case of the RTX 2060 that power is reduced to 37 trillion operations of this type, which as you know, is dedicated to rendering with ray tracing and artificial intelligence operations performed by the Tensor Core.
What does that mean? Well, we can enjoy the photorealism of ray tracing in games like ‘Battlefield V’, but we will do it with joy only in 1080p resolutions, since the impact in major resolutions is appreciable.
As we will see in the tests the benefits of this technology again impose a burden on the rate of frames per second, but thanks to the efforts of NVIDIA that impact is now lower and there are already several titles that will also offer that photorealism in the coming months.
We must pay attention to another interesting element of these graphs, and it is the absence of SLI support, something that as we already knew in the launch of the family was only reserved for the RTX 2080 and the RTX 2080 Ti.
In the design that connector disappears, and also changes the position of the 8-pin power connector, which is located on the back of the graph. On the front, as always, the connectors with two ports DisplayPort, 1 HDMI port, 1 DVI and a USB-C connector with VirtualLink technology, which will connect virtual reality glasses with a single cable.
The RTX 2060 in tests
To evaluate the performance of the NVIDIA GeForce RTX 2060 we used the equipment we already used to analyze the RTX 2080 and the RTX 2080 Ti. Thus, the processor is a Core i5-8400, accompanied by 16 GB of 3200 MHz DDR4 memory mounted on an ASRock Z370M-ITX / ac board. The main storage unit is a Samsung 970 EVo M.2 NVMe, while the power supply is a Be Quiet BN282 650 W.
To have two reference points and given that we did not have an RTX 2070 or a GTX 1070 Ti we used the results we already obtained with the aforementioned RTX 2080 and also with the GTX 1070 that also served as a comparative element in the previous analysis.
They are not the ideal options to draw even clearer conclusions, but we believe they are valid enough to understand how far these new NVIDIA graphics can go. It is important to clarify that in all the tests we have selected the maximum level of detail when we could do it, while the resolutions that we have passed the games are 1080p, 1440p and 4K.
We started with Cinebench, a benchmark that confirms that the CPU is much more relevant: the differences between these three graphics are practically null since the workload falls on the processor, which in this case is a relatively modest Core i5- 8400.
The thing changes with 3DMark, which curiously behaves better in the GTX 1070 in Ice Storm. The impact of the new RTX 2060 is not too noticeable in our tests.
The first of the benchmarks with games we do is with Dirt Rally: here we start to see how the RTX 2060 shows its virtues against the GTX 1070. It is far from the RTX 2080 in 1080p resolutions, but that difference disappears in 4K, something that it is especially interesting.
In Shadow of the Tomb Raider those differences are evident in all resolutions. The NVIDIA graphics behave much better here in DirectX 12 mode, and the improvement over the GTX 1070 is remarkable.
With The Division curiously the difference between the RTX 2060 and the GTX 1070 is almost negligible, and it is the RTX 2080 that gives a very remarkable jump with respect to these graphics not in one, but in all the resolutions with which we perform the tests.
In World of Tanks the performance of the RTX 2060 returns to be very similar to that of the GTX 1070 although it surpasses it, especially when we do not play in 4K. Still the RTX 2080 again proves to be a great distance from both proposals.
The Final Fantasy XV demo allows you to evaluate a very important section of these graphs: the impact of DLSS antialiasing technology. In the RTX 2060 is present and the benefits over the traditional TAA is evident. In the GTX 1070 this option is not supported, so it is only possible to get the performance with TAA antialiasing.
Finally we did our usual test with Battlefield V in which we made a one minute tour in story mode in a repetitive way. In these numbers we only compare the RTX 2060 with the RTX 2080 to know the impact of the RTX technology that is not present in the GTX 1070.
The results show how even activating the ray tracing it is possible to play more than 60 FPS in 1080p mode with the low level of rendering with RTX technology. In fact we are close to that barrier in 1440p mode, which is excellent news and shows that even with these graphics it is possible to enjoy that photorealistic experience without problems.