NVIDIA GeForce GTX 1070 Ti vs NVIDIA GeForce GTX 1660 SUPER Comparison

Head-to-Head Benchmarks

The benchmark data paints a clear picture: the NVIDIA GeForce GTX 1070 Ti wins 9 of the 10 head-to-head comparisons, with the GTX 1660 SUPER taking a single victory. The most decisive margin comes in PassMark's GPU compute test, where the 1070 Ti scores 7165 against 5076, a 41.2% advantage. That is a massive gap for compute-oriented workloads, reflecting the older card's larger shader array and memory subsystem.

DirectX 10 performance shows the second-largest delta, with the 1070 Ti scoring 90 versus 65 for the 1660 SUPER, a 38.5% lead. This legacy API result suggests the 1070 Ti retains substantial raw throughput that benefits older titles. The 3DMark Steel Nomad DX12 test also favors the 1070 Ti substantially: 1471 versus 1278, a 15.1% edge. PassMark G3D confirms the pattern, with the 1070 Ti at 14674 and the 1660 SUPER at 12699, a 15.6% difference.

The 1070 Ti also wins the Vulkan geekbench test, scoring 61006 against 57102, a 6.8% margin. In DirectX 9, it leads 207 to 189 (9.5%), and in 2D graphics it posts 876 versus 805 (8.8%). The DirectX 11 and DirectX 12 results are closer: 107 to 104 (2.9%) and 51 to 50 (2.0%), respectively. These narrow margins indicate that in modern API workloads, the gap between the two cards shrinks considerably.

The single win for the GTX 1660 SUPER comes in geekbench OpenCL, where it scores 52490 against 49219 for the 1070 Ti, a 6.2% advantage. This is an interesting outlier, as the 1660 SUPER's smaller chip and fewer shading units still manage to outperform in this particular OpenCL workload. However, it is the exception rather than the rule.

Average benchmark scores reinforce the overall picture. The 1070 Ti sits at 14277, while the 1660 SUPER trails at 12986. In the database's percentile ranking, the 1070 Ti places in the 56th percentile of all GPUs, compared to the 53rd percentile for the 1660 SUPER. The 1070 Ti's nearest rivals in the database include the Intel Iris Xe MAX Graphics (14315, only 0.3% higher) and the AMD Radeon Vega 11 (14352, 0.5% higher), meaning it sits in a tightly contested cluster. The 1660 SUPER, by contrast, is closest to the NVIDIA GeForce RTX 3050 Ti Mobile (12940, 0.4% higher) and the AMD Radeon RX 580 (12928, 0.4% higher).

The Verdict

From the data, the GTX 1070 Ti is the stronger performer in nearly every measured category. Its wins span legacy APIs, modern DX12, Vulkan, and compute workloads. The 41.2% advantage in GPU compute and the 38.5% lead in DirectX 10 are not marginal; they represent a fundamentally higher throughput ceiling. For anyone prioritizing raw frame rates or compute tasks, the 1070 Ti is the clear choice.

The GTX 1660 SUPER, however, has its place. Its OpenCL win (6.2% ahead) shows that in certain general-purpose compute scenarios, the newer Turing architecture can punch above its weight. The card also draws less power, as seen in its TDP of 125 W against 180 W for the 1070 Ti, and it requires a smaller suggested PSU (300 W versus 450 W). The 1660 SUPER is also physically shorter at 229 mm versus 267 mm, making it easier to fit in compact cases.

The verdict depends on the use case. For maximum performance across a broad range of benchmarks, the 1070 Ti wins. For a lower-power, shorter card that still delivers competitive results in modern APIs, the 1660 SUPER is the pragmatic pick. The data does not support calling the 1660 SUPER a replacement for the 1070 Ti; it is a different class of product with different trade-offs.

Architecture Differences

The two cards come from different NVIDIA generations. The GTX 1070 Ti uses the GP104 chip built on the Pascal architecture, fabricated on TSMC's 16 nm process. The GTX 1660 SUPER uses the TU116 chip with the Turing architecture, on a 12 nm process from the same foundry. This process shrink is significant: the 1070 Ti's die measures 314 mm², while the 1660 SUPER's die is smaller at 284 mm², despite the newer node. Transistor counts differ accordingly: 7,200 million for the 1070 Ti versus 6,600 million for the 1660 SUPER. Transistor density is slightly higher on the newer part, at 23.2M per mm² versus 22.9M per mm².

The 1070 Ti has a much larger shader configuration: 2432 shading units, 152 texture mapping units, and 64 render output units. The 1660 SUPER is cut down to 1408 shading units, 88 TMUs, and 48 ROPs. This explains much of the performance gap, particularly in fill-rate-bound scenarios. The 1070 Ti's pixel rate is 107.7 GPixel/s versus 85.68 GPixel/s for the 1660 SUPER, and its texture rate is 255.8 GTexel/s versus 157.1 GTexel/s. FP32 compute also favors the 1070 Ti: 8.186 TFLOPS against 5.027 TFLOPS.

Memory is another major divergence. The 1070 Ti uses 8 GB of GDDR5 on a 256-bit bus, yielding 256.3 GB/s of bandwidth. The 1660 SUPER uses 6 GB of GDDR6 on a narrower 192-bit bus, but the faster memory clocks (14 Gbps effective versus 8 Gbps) push bandwidth to 336.0 GB/s. So the newer card has 31% more memory bandwidth, which helps explain its competitive showing in some modern workloads despite fewer cores.

FP16 performance is a stark contrast: the 1070 Ti runs at 127.9 GFLOPS with a 1:64 ratio, while the 1660 SUPER achieves 10.05 TFLOPS with a 2:1 ratio. This is a Turing architecture feature that Pascal lacks, giving the 1660 SUPER a massive advantage in FP16 workloads, though the benchmark suite does not include a direct FP16 test.

Both cards share the same API support: DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. Neither has ray tracing cores or tensor cores. Both use a PCIe 3.0 x16 interface and a single 8-pin power connector. Display outputs differ slightly: the 1070 Ti offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a, while the 1660 SUPER has 1x DVI, 1x HDMI 2.0, and just 1x DisplayPort 1.4a.

FAQ

Q: Which card has higher average benchmark scores?

A: The GTX 1070 Ti has an average benchmark score of 14277, while the GTX 1660 SUPER averages 12986. The 1070 Ti also ranks higher in the database, at the 56th percentile of all GPUs versus the 53rd percentile for the 1660 SUPER.

Q: Is the GTX 1660 SUPER faster in any benchmark?

A: Yes, in the geekbench OpenCL test the 1660 SUPER scores 52490 against 49219 for the 1070 Ti, a 6.2% advantage. This is the only head-to-head test the 1660 SUPER wins.

Q: What is the biggest performance difference between the two?

A: The largest gap is in PassMark GPU compute, where the 1070 Ti leads by 41.2% (7165 versus 5076). The second-largest is DirectX 10, with a 38.5% margin (90 versus 65).

Q: How do the memory systems compare?

A: The 1070 Ti has 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s. The 1660 SUPER has 6 GB of GDDR6 on a 192-bit bus, but delivers 336.0 GB/s due to faster effective memory clocks (14 Gbps versus 8 Gbps).

Q: Do either cards support ray tracing?

A: No. Both cards have no ray tracing cores and no tensor cores, according to the database specifications.

Q: Which card has a higher boost clock?

A: The GTX 1660 SUPER has a boost clock of 1785 MHz, while the GTX 1070 Ti boosts to 1683 MHz. The 1070 Ti has a higher base clock at 1607 MHz versus 1530 MHz.

Where Each One Wins

The GTX 1070 Ti wins in almost every category that matters for traditional gaming and compute. Its 15.1% lead in 3DMark Steel Nomad DX12 shows strength in modern DirectX 12 titles. The 38.5% margin in DirectX 10 makes it the better choice for older games that rely on that API. The 41.2% lead in GPU compute means it is the superior card for tasks like rendering, physics simulations, or any workload that stresses raw shader throughput. The 9.5% edge in DirectX 9 and the 8.8% lead in 2D graphics further solidify its position for legacy software and desktop acceleration. Its 6.8% Vulkan advantage also matters for games and applications that use that API.

The GTX 1660 SUPER wins in OpenCL compute, with a 6.2% margin. This makes it the better option for specific applications that are optimized for OpenCL, though the broader compute picture still favors the 1070 Ti. The 1660 SUPER also offers superior memory bandwidth (336.0 GB/s versus 256.3 GB/s), which could benefit bandwidth-sensitive workloads even if the raw compute is lower. Its lower TDP of 125 W versus 180 W makes it the better choice for systems with modest power supplies or limited cooling. The shorter physical length (229 mm versus 267 mm) gives it an advantage in small form factor builds.

For practical use, the 1070 Ti is the pick for high-refresh gaming, heavy compute, and anyone who wants maximum performance from an older-generation card. The 1660 SUPER is the pick for those who need a smaller, cooler, and less power-hungry card that still handles modern games competently, especially when memory bandwidth is the bottleneck. The data does not show the 1660 SUPER as a straight upgrade; it is a sidegrade with different strengths.

Specification Differences

The two cards differ across nearly every core specification. The 1070 Ti uses the GP104 chip on Pascal architecture and a 16 nm process, while the 1660 SUPER uses the TU116 chip on Turing architecture and a 12 nm process. The 1070 Ti has a larger die at 314 mm² versus 284 mm², and more transistors at 7,200 million versus 6,600 million. Transistor density is slightly higher on the 1660 SUPER, at 23.2M per mm² versus 22.9M per mm².

Clock speeds differ: the 1070 Ti has a 1607 MHz base and 1683 MHz boost, while the 1660 SUPER has a 1530 MHz base and a higher 1785 MHz boost. Memory clocks are different too: 2002 MHz (8 Gbps effective) for the 1070 Ti versus 1750 MHz (14 Gbps effective) for the 1660 SUPER.

Memory capacity, type, bus width, and bandwidth all differ. The 1070 Ti has 8 GB GDDR5 on a 256-bit bus with 256.3 GB/s; the 1660 SUPER has 6 GB GDDR6 on a 192-bit bus with 336.0 GB/s. Shader resources are heavily in the 1070 Ti's favor: 2432 shading units, 152 TMUs, and 64 ROPs versus 1408 shading units, 88 TMUs, and 48 ROPs. Pixel rate is 107.7 GPixel/s versus 85.68 GPixel/s, and texture rate is 255.8 GTexel/s versus 157.1 GTexel/s. FP32 compute is 8.186 TFLOPS versus 5.027 TFLOPS, and FP16 is 127.9 GFLOPS (1:64) versus 10.05 TFLOPS (2:1).

Power and physical specs also separate them. The 1070 Ti has a 180 W TDP and a suggested 450 W PSU; the 1660 SUPER has a 125 W TDP and a suggested 300 W PSU. Both are dual-slot with a single 8-pin connector, but the 1070 Ti is longer (267 mm versus 229 mm) and thicker (40 mm versus 35 mm). Height is nearly identical at 112 mm versus 111 mm. Display outputs differ: the 1070 Ti has 3 DisplayPorts, while the 1660 SUPER has only 1. Both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The 1070 Ti was released in November 2017 with a launch MSRP of 399 USD; the 1660 SUPER was released in October 2019 with a launch MSRP of 229 USD.