The Turing-based Nvidia GTX 1660 Ti GPU is out. The all-new budget graphics processor incorporates Turing’s improved shaders with a full complement of video decode & encode acceleration features.
The graphics card packs a GDDR6 VRAM memory, just like the higher-end GeForce RTX 20-series.
However, the GTX 1660 Ti doesn’t pack the RTX ray tracing accelerating features. As a result, these TU116-based cards offer a cheaper processor chip with still top-notch specs.
Performance-per-dollar
The Nvidia GeForce GTX 1660 Ti GPU came out on February 22 with a starting price of $280. The company says the unit will completely replace the old GeForce GTX 1060 6GB. Keep in mind the price of the newer GPU is $30 higher than the older hardware. Meanwhile, Nvidia claims the 1660 Ti is 1.5 times faster than the 2016’s model -and with the same 120W power consumption.
In summary, the new card is 60 percent more expensive for a unit that’s 50 percent more efficient. Improved performance per dollar isn’t a thing we’ve seen much from Turing generations so far. At least we’re now getting a more budget-friendly alternative for us PC build enthusiasts out there.
Even so, is the GTX 1660 Ti GPU a good option for Full HD (1080p) gaming?
The TU116 without RT and Tensor Cores
Nvidia has launched four different GPUs down the Turing hierarchy. With each one, the company keeps eliminating resources to hit lower prices.
However, it’s clear Nvidia is trying to maintain balance along the way. It means there are no bottlenecks lowering peak performance.
The GeForce RTX 2060 is packed with 44 percent the power of the 2080 Ti’s CUDA cores and texture units. Furthermore, it has 54 percent its memory bandwidth and ROPs, and 50 percent of its L2 cache.
The new TU116 unit doens’t feature Nvidia’s RT and tensor cores, which lowers its efficiency but also its price.
In the TU116, Nvidia has also replaced Turing’s Tensor cores with 128 dedicated Fp16 cores per SM. IT gives the GeForce GTX 160 Ti the ability to process at 2X speed than its predecessor.
Other Turing-based improvements allows the 1660 Ti graphics card to be more precise and gast than Pascal-based GTX 1060 GPU at 1080p gaming.
Nvidia GeForce GTX 1660 Ti specs
Let’s put it in perspective:
| EVGA GeForce GTX 1660 Ti XC Black Gaming | GeForce RTX 2060 FE | GeForce GTX 1060 FE | GeForce GTX 1070 FE |
|
|---|---|---|---|---|
| Architecture (GPU) |
Turing (TU116) | Turing (TU106) | Pascal (GP106) | Pascal (GP104) |
| CUDA Cores |
1536 | 1920 | 1280 | 1920 |
| Peak FP32 Compute |
5.4 TFLOPS | 6.45 TLFOPS | 4.4 TFLOPS | 6.5 TFLOPS |
| Tensor Cores |
N/A | 240 | N/A | N/A |
| RT Cores |
N/A | 30 | N/A | N/A |
| Texture Units |
96 | 120 | 80 | 120 |
| Base Clock Rate |
1500 MHz | 1365 MHz | 1506 MHz | 1506 MHz |
| GPU Boost Rate |
1770 MHz | 1680 MHz | 1708 MHz | 1683 MHz |
| Memory Capacity |
6GB GDDR6 | 6GB GDDR6 | 6GB GDDR5 | 8GB GDDR5 |
| Memory Bus |
192-bit | 192-bit | 192-bit | 256-bit |
| Memory Bandwidth |
288 GB/s | 336 GB/s | 192 GB/s | 256 GB/s |
| ROPs |
48 | 48 | 48 | 64 |
| L2 Cache |
1.5MB | 3MB | 1.5MB | 2MB |
| TDP |
120W | 160W | 120W | 150W |
| Transistor Count |
6.6 billion | 10.8 billion | 4.4 billion | 7.2 billion |
| Die Size |
284 mm² | 445 mm² | 200 mm² | 314 mm² |
| SLI Support |
No | No | No | Yes (MIO) |
