GTX 1070 vs. GTX 1660

In this article, we’re going to pit the GTX 1070 and the GTX 1660 against each other in a battle of GPUs to determine which one will provide you with the best gameplay experience

Last Updated:

In this article, we’re going to pit the GTX 1070 and the GTX 1660 against each other in a battle of GPUs to determine which one will provide you with the best gameplay experience. We compare their different specifications and features, such as the brand, the architecture of each GPU, cooling, and dimensions.

MSI GeForce GTX 1660 Gaming X 6G

MSI GeForce GTX 1660 Gaming X 6G

Clock Speed

1860 MHz (Boost)



Memory Bus Width

192 bit



When you reach the higher end of the GPU spectrum, the choice becomes less about Nvidia and AMD, and more about which GeForce graphics cards are going to best suit your needs. In this particular battle, we’re talking about the GTX 1070 and the GTX 1660.

Released shortly after the GTX 1660 Ti, the GTX 1660 offers less of the bells and whistles but provides a more affordable option to the Ti of its series, as does the GTX 1070 in its own group of GPUs. Compared to the GTX 1080, it’s more easily in reach of the average gamer.

We’re going to take a look at both in close comparison to determine which performs better in each area so you can decide which is going to be the better graphics card for you, considering factors such as architecture, cooling, dimensions, and VRAM.


There are some core differences between the GTX 1070 and the GTX 1660, despite having the same prefix to their name. The GTX 1070 is based on Nvidia’s Pascal microarchitecture which is what the GTX 1070 Ti uses, but with a reduced number of cores.

The newer GeForce 1660 GPU features the more advanced Turing architecture, although one of the main advantages of this is that it brings with it specialized RT and Tensor cores which are absent in the GTX 1660. It is also supposed to boost performance by about 50% per CUDA core for an enhanced gaming experience.

To compare them, the GTX 1070 has 1,920 CUDA cores whereas the GTX 1660 has just 1,408, but this is where that boost might come in handy. If Nvidia is right, then their Turing architecture could see the GTX 1660 performing at a standard more in line with 2,816 CUDA cores.

But how does it translate to clock speeds and actual benchmarks? The GTX 1070 has a base clock speed of 1,506 MHz and it can be boosted to 1,683 MHz when overclocked. The GTX 1660 takes the lead in terms of speed, then, with a base clock speed of 1,530 MHz and a maximum boosted clock speed of 1,785 MHz.


The GTX 1070 can run pretty hot despite its blower cooling system, with a power draw of 150 watts. It does tend to settle around the 70℃ mark after reaching highs of over 80 when it’s not overclocked, but once you start altering these settings to squeeze even more from your GPU you’ll be pushing closer toward the 94℃ limit.

There’s one degree of difference between the maximum safe temperature as the limit for the GTX 1660 is 95℃. However, it’s less likely to get near this level of heat as it cools much more efficiently and it has a reduced power draw of 120 watts, so it won’t require as much energy to run and therefore maintains a cooler temperature.


Already looking a little crowded on your board? You might want to pay some attention to the dimensions of each GPU just to check that it will definitely slot in with your existing set-up. Really, there’s not a huge difference in it between the GTX 1070 and the GTX 1660.

The measurements for the GTX 1660 are 4.37” (H) x 5.7” (L) x 2-slot (W) which is a fairly standard size for GPUS. Comparably, the GTX 1070 is almost double the size at 4.376” (H) x 10.5” (L) x 2-slot (W) so it’ll take up a lot more space. This is fine for bigger boards, but you might want to downsize by upgrading to the GTX 1660 if you have a smaller motherboard.

In terms of die area, the GTX 1070 has a 314 mm² chip with 7,200 million transistors compared to the GTX 1660’s smaller Pascal chip, which is 284 mm² with 6,600 million transistors by comparison.

Resolution and Frames Per Second

From an initial glance, it appears that the GTX 1070 delivers better frames per second across the different resolution qualities. At your 1080p performance settings, it delivers on average around 93.6 frames per second which is a huge improvement on the GTX 1660.

Games like Fortnite and World of Tanks push the gap between these graphics card’s performances even further, but it’s less noticeable when you’re playing games like Grand Theft Auto V and practically non-existent for the less demanding games, such as Minecraft.

The difference in frames per second also becomes less noticeable when the settings are increased, so there’s a smaller gap at 1440p and it narrows again when you reach 4k graphics where there’s as little as 7% between them.

While you can’t altogether discount the Turing architecture, it’s clear to see that the GTX 1070 still delivers faster frame rates and improved gaming performance when compared to the GTX 1660.

Ray Tracing?

If you’re looking for a GPU that can utilize all that the new Turing technology has to offer, including ray tracing ability and improved CUDA cores, unfortunately, you won’t find it in either of these GPUs so you’ll need to redirect your search toward the RTX series instead.

With that said, the very fact that the GTX 1660 is equipped with Turing architecture means that you can use the Nvidia Driver that supports ray tracing on GTX cards to experience this new technology, but it’s an incredibly heavy load for a card of this quality to handle, so there will be a noticeable drop in frame rate speeds and quality.


It’s a shame that the GTX 1660 didn’t upgrade the memory type along with the architecture, but like the GTX 1070, it features GDDR5. They do, however, have different capacities. The former has 6 GB of memory whereas the latter has a larger 8 GB, therefore it stores more.

This may not seem like a lot, but GPUs with a better memory capacity will perform better in general and you’ll be able to play your favorite games on higher graphics settings. The GTX 1070 also has a 256-bit bus interface and a 256.3 Gbps bandwidth compared to the GTX 1660, which has a mere 193-bit but interface and just a 192.1 Gbps bandwidth.

Despite this, the Turing architecture comes into play once again and brings the GTX 1660 up to the same memory clock speed of 8 Gbps as the GTX 1070.


If you’re looking for a new graphics card for your gaming PC then the GTX 1070 offers faster frame rates, improved benchmarking results, and in most cases, there’s very little in the price difference between this and the GTX 1660.

It’s certainly not worth upgrading from the GTX 1660 to the GTX 1070 if you already have this GPU as part of your existing gaming rig as you’ll only notice around a 7% improvement, but it can be a good choice for anyone with an older or less advanced graphics card to upgrade your rig.

However, although it doesn’t feature the advantage of the Turing architecture’s new Tensor and Ray Tracing cores, the GTX 1660 still offers much better value for money than the older GTX 1070 graphics card and the Turing architecture does seem to offer a certain boost.