Nvidia GTX 1660 Ti
Turing GTX 16 card without Tensor Cores, for basic compute and display work.
- Arch
- Turing
- Memory
- 6 GB GDDR6
- Bandwidth
- 288 GB/s
- Released
- Q1 2019
At a glance
We don't currently track any GTX 1660 Ti offers. See alternative GPUs.
Technical Specifications
Nvidia GTX 1660 Ti · Per GPU
| Compute · dense | |
|---|---|
| FP16 / BF16 | 10.9 TFLOPS |
| FP32 | 5.4 TFLOPS |
| Precision support | FP16FP32 |
| Memory | |
|---|---|
| Capacity | 6 GB GDDR6 |
| Bandwidth | 288 GB/s |
| Bus width | 192-bit |
| ECC | No |
| Silicon | |
|---|---|
| Architecture | Turing |
| Process | TSMC 12 nm FinFET |
| Transistors | 6.6 billion |
| Shader cores | 1,536 CUDA cores |
| Compute units | 24 SMs |
| Fabric and host | |
|---|---|
| Host interface | PCIe 3.0 x16 |
| Power | |
|---|---|
| Board power | 120 W |
| Cooling | Active |
Frequently Asked Questions
Why choose the GTX 1660 Ti?
6GB of GDDR6 at 288 GB/s and the highest shader count in the GTX 16 series. A low-cost card for cloud desktops, video encoding and CUDA code that does not need tensor cores.
When is the GTX 1660 Ti not a good fit?
No tensor cores and 6GB of VRAM, so mixed-precision training and most current models do not fit or run slowly. For tensor-core inference on small models, the RTX 2060 has the same 6GB with Turing tensor cores. Subject to Nvidia's consumer EULA, which may restrict data center use.
Alternatives to Nvidia GTX 1660 Ti
-
Nvidia RTX 3060
12 GB GDDR6 · 25% more bandwidth than the GTX 1660 Ti
On-demand from $0.08 /GPU/hr
-
Nvidia RTX 2060
6 GB GDDR6 · 17% more bandwidth than the GTX 1660 Ti
On-demand from $0.06 /GPU/hr
-
Nvidia GTX 1660 Super
6 GB GDDR6 · 17% more bandwidth than the GTX 1660 Ti
On-demand from $0.04 /GPU/hr
-
Nvidia GTX 1660
6 GB GDDR5 · 33% less bandwidth than the GTX 1660 Ti
On-demand from $0.04 /GPU/hr
Last updated