NVIDIA GeForce GTX 1660 Ti Review: Turing Under $300
GeForce GTX 1660 Ti - A New Breed Of Turing GPU
What the GeForce GTX 1660 Ti does do, however, is incorporate all of the architectural efficiencies of Turing into an affordable graphics card that significantly outperforms its Pascal-based namesake. We’ve got the main features and specification for the GeForce GTX 1660 Ti outlined in the table below, alongside a few of NVIDIA’s other mainstream GPUs, to show you where the card fits in the line-up. As we move forward, we’ll dig into the card itself, see how it performs in a myriad of benchmarks and games, as well as how it overclocks. So let’s get moving...
![]() |
|
The GeForce GTX 1660 Ti is based on the TU116 GPU, which is a different slice of silicon in comparison to any of the other Turing-based offerings in NVIDIA’s line-up. The TU116 is comprised of roughly 6.6B transistors and has a die size of 284mm2, which for reference is 161 square millimeters smaller than the TU106 GPUemployed in the GeForce RTX 2060 and 2070.
All of those transistor and die size savings result from the removal of Tensor and RT cores from the TU116, which is why the GPU doesn’t support NVIDIA’s RTX Technology, i.e. hardware accelerated Ray Tracing and DLSS (Deep Learning Super Sampling). Those features, however, weren’t the only new things to be introduced with Turing.
For a more in-depth breakdown of what NVIDIA brings to the table with the Turing architecture, we’d suggest checking out some of our early coverage here. To quickly re-iterate some of the features most pertinent to today’s launch of the GeForce GTX 1660 Ti, however, we will cover a few of them again here.
NVIDIA TU116: What Does It Support?
It may lack RT and Tensor cores, but the TU116 GPU, like the other Turing-based GPUs, supports concurrent integer and floating point instructions (rather than serializing integer and FP instructions), and it also has a redesigned cache structure relative to Pascal. NVIDIA’s Turing-based GPUs have double the amount of L2 cache versus their predecessors and the L1 cache has been outfitted with a wider memory bus that ultimately doubles the bandwidth. There is more total L1 cache and shared memory as well, and the overall cache configuration has been changed to be more symmetrical. These changes improve utilization and efficiency, and ultimately increase performance with many workloads.The TU116 also supports new shading technologies, like Variable Rate Shading. VRS offers developers finer-grained control of shading rates in any given scene. For example, each 16x16 pixel region of a scene can have a different level of shader detail and a different shading rate as a result. Other examples of VRS shading optimization techniques include Foveated Rendering, where shading is done at variable rates depending on what area is most visible to the viewers eyes, and Texture Space shading, where game developers can reuse shader output in areas of a scene that are identical, rather than having to re-shade those areas and unnecessarily increasing the workload.
The GeForce GTX 1660 Ti and TU116 GPU also have NVIDIA’s latest NVENC video engine. The video engine in Turing offers HEVC 8K30 HDR encoding in real time, along with VP9 10/12b HDR and HEVC 444 10/12b HDR decoding. It is also more efficient and results in lower CPU utilization than previous-gen architectures for streaming video and media sites like YouTube and Twitch, which is ideal for creators who may be gaming, as well as processing and managing their stream from a single system.
In terms of its actual configuration, the TU116 has 1,536 active CUDA cores, which is a decent uptick from the GTX 1060, but less than the RTX 2060. The TU116 in the GeForce GTX 1660 Ti has 48 ROPs (like the 1060 and 2060) and 96 texture units, which also falls somewhere in between the GTX 1060 and RTX 2060. The memory interface to the 6GB of GDDR6 on the card is 192-bits wide and runs at an effective data rate of 12Gbps, for peak memory bandwidth slightly north of 288GB/s. The GPU’s base clock is 1,500MHz and it’ll boost up to 1,770MHz. The 1660 TI's TDP is roughly 120W, and as such, the cards only require a single, supplemental power connector. In the case of the EVGA card we’ll be showing you here, it sports a single 8-pin connector, to offer some additional headroom.
NVIDIA’s reference specifications may not be all that common, however, since there will be no Founder's Edition or NVIDIA-branded GeForce GTX 1660 Ti cards. All 1660 Tis will be coming from NVIDIA’s partners; we’ve got an EVGA GeForce GTX 1660 Ti XC Black here. This particular card does conform to NVIDIA’s reference specs, but there will be an array of other custom and overclocked models with higher boost clocks, and also higher prices, of course.
Like the EVGA GeForce RTX 2060 XC we recently evaluated, the EVGA GeForce GTX 1660 Ti XC Black is a smaller form factor card and is only about 7.5” long. The cooler on the card, however, is relatively thick at 2.75 slots wide, so the EVGA GeForce GTX 1660 Ti XC Black will technically consume 3 slots in a system. That thicker cooler also means that cooling performance isn’t hampered by the shorter heatsink, however, because surface area isn’t sacrificed. The EVGA GeForce GTX 1660 Ti XC Black has a deep, dense heatsink, with a single, centrally-mounted cooling fan. And that fan is outfitted with a new blade design, and hydro-dynamic bearings that not only reduce noise and increase the fan’s lifespan, but improve cooling performance as well.
The EVGA GeForce GTX 1660 Ti XC Black features the same webbed fan shroud and angular design as most of EVGA’s recently-released graphics cards, though it does not have any RGB lighting to speak of. The outputs on the card are also somewhat different than many newer GeForce cards, though. The EVGA GeForce GTX 1660 Ti XC Black is rocking one full-sized DisplayPort, an HDMI port, and a DVI port – and all three can be used simultaneously. The USB Link connector and second DisplayPort available on NVIDIA’s Turing-based GeForce RTX cards are not present here.