NVIDIA GeForce GTX 1070 Ti vs NVIDIA GeForce GTX 660 Ti Comparison

Head-to-Head Benchmarks

The benchmark data shows a decisive performance gap between the NVIDIA GeForce GTX 660 Ti and the NVIDIA GeForce GTX 1070 Ti, with the newer card winning both head-to-head tests. In Geekbench OpenCL, the GTX 1070 Ti scores 49,219 points against the GTX 660 Ti's 15,113 points, a delta of -69.3% from the perspective of the older card. This means the GTX 1070 Ti delivers more than triple the compute throughput in this workload. The margin grows even larger in Geekbench Vulkan, where the GTX 1070 Ti posts 61,006 points versus 15,012 points for the GTX 660 Ti, representing a -75.4% delta. The Vulkan result is particularly telling because it reflects modern API efficiency, an area where the older Kepler architecture struggles significantly.

The GTX 660 Ti's average benchmark score of 15,063 places it in the 57th percentile of all GPUs, while the GTX 1070 Ti's average of 14,277 sits in the 56th percentile — a seemingly contradictory pair of figures given the head-to-head results. This anomaly is explained by the different benchmark suites each card was tested with; the GTX 660 Ti only has OpenCL and Vulkan Geekbench scores recorded, whereas the GTX 1070 Ti was subjected to a broader range including 3DMark Steel Nomad DX12, PassMark DirectX 9/10/11/12, G2D, G3D, and GPU Compute tests. The GTX 1070 Ti's PassMark G3D score of 14,674 and GPU Compute score of 7,165 show strong rasterization and compute performance in legacy APIs, while its Geekbench Metal score of 22,180 indicates capable cross-platform performance.

Looking at nearest rival data, the GTX 660 Ti's closest competitor is the AMD Radeon Pro 560X with an average score of 15,082 (-0.1% delta), followed by the AMD Radeon RX 7600 at 15,171 (-0.7%), the NVIDIA GeForce RTX 3050 OEM at 15,199 (-0.9%), and the AMD Radeon 680M at 15,270 (-1.4%). This clustering within roughly 1.5% suggests the GTX 660 Ti's compute capability is competitive with much newer entry-level hardware, despite its age. The GTX 1070 Ti's nearest rivals include the Intel Iris Xe MAX Graphics at 14,315 (-0.3%), AMD Radeon Vega 11 at 14,352 (-0.5%), NVIDIA GeForce GTX TITAN at 14,373 (-0.7%), and AMD Radeon RX Vega 11 at 14,385 (-0.7%). The GTX 1070 Ti's rivals are all within 1% of its average score, indicating that its overall benchmark average is tightly grouped with these parts, though the head-to-head wins show its Geekbench advantage is substantial.

FAQ

Q: Which card wins in Geekbench OpenCL?

A: The NVIDIA GeForce GTX 1070 Ti wins decisively with a score of 49,219 versus 15,113 for the GTX 660 Ti, a -69.3% delta. The GTX 1070 Ti's score is more than triple that of the older card.

Q: How much faster is the GTX 1070 Ti in Vulkan?

A: The GTX 1070 Ti scores 61,006 in Geekbench Vulkan compared to 15,012 for the GTX 660 Ti, a -75.4% delta. This is the largest relative gap between the two cards across all head-to-head tests.

Q: What percentile rank does each card hold among all GPUs?

A: The GTX 660 Ti ranks in the 57th percentile of all GPUs, while the GTX 1070 Ti ranks in the 56th percentile. Despite the GTX 1070 Ti's dominant head-to-head wins, the percentile figures are nearly identical due to different benchmark suites being used for each card's average score.

Q: What is the GTX 1070 Ti's strongest benchmark result?

A: The GTX 1070 Ti's highest recorded score is 61,006 in Geekbench Vulkan, followed by 49,219 in Geekbench OpenCL. Its PassMark DirectX 9 score of 207 is its strongest legacy API result.

Q: How does the GTX 660 Ti compare to its nearest rivals?

A: The GTX 660 Ti's average score of 15,063 is within 1.4% of all four of its nearest rivals: AMD Radeon Pro 560X (-0.1%), AMD Radeon RX 7600 (-0.7%), NVIDIA GeForce RTX 3050 OEM (-0.9%), and AMD Radeon 680M (-1.4%). This indicates its compute performance is closely matched with these newer parts.

Q: Does the GTX 1070 Ti have any weak benchmark results?

A: The GTX 1070 Ti shows relatively low scores in PassMark DirectX 12 (51) and DirectX 10 (90), while its DirectX 11 score is 107 and DirectX 9 reaches 207. Its G2D score of 876 suggests stronger 2D performance than its 3D legacy scores.

Architecture Differences

The two cards represent fundamentally different GPU architectures from NVIDIA. The GTX 660 Ti uses the GK104 chip built on the Kepler architecture, fabricated on TSMC's 28 nm process node. It packs 3,540 million transistors into a 294 mm² die, yielding a transistor density of 12.0M per mm². The GTX 1070 Ti, by contrast, uses the GP104 chip with the Pascal architecture on a 16 nm process from the same foundry. It contains 7,200 million transistors across a 314 mm² die, achieving a density of 22.9M per mm² — nearly double the transistor density of the older chip. The smaller process node allows the GTX 1070 Ti to pack more than twice the transistors into a similar die area.

Clock speeds differ substantially. The GTX 660 Ti runs at a 915 MHz base clock and 980 MHz boost, while the GTX 1070 Ti operates at 1,607 MHz base and 1,683 MHz boost. That is a 692 MHz advantage at base and 703 MHz at boost for the newer card. The memory clocks show a similar story: the GTX 660 Ti uses 1,502 MHz memory with 6 Gbps effective speed, while the GTX 1070 Ti runs at 2,002 MHz with 8 Gbps effective. Memory configuration also diverges significantly. The GTX 660 Ti has 2 GB of GDDR5 on a 192-bit bus with 144.2 GB/s bandwidth. The GTX 1070 Ti comes with 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s — nearly 78% more bandwidth.

Compute resources scale accordingly. The GTX 660 Ti has 1,344 shading units, 112 texture mapping units, and 24 ROPs. The GTX 1070 Ti increases these to 2,432 shading units, 152 TMUs, and 64 ROPs. Pixel fill rate jumps from 27.44 GPixel/s to 107.7 GPixel/s, and texture rate rises from 109.8 GTexel/s to 255.8 GTexel/s. FP32 throughput expands from 2.634 TFLOPS to 8.186 TFLOPS, more than a threefold increase. The GTX 1070 Ti also has a native FP16 capability rated at 127.9 GFLOPS (1:64 ratio), while the GTX 660 Ti has no listed FP16 specification.

API support differs in important ways. Both cards support DirectX 12 and OpenGL 4.6, but the GTX 660 Ti's DirectX 12 implementation is limited to the 11_0 feature level, while the GTX 1070 Ti supports the full 12_1 feature level. Vulkan support also differs: the GTX 660 Ti supports Vulkan 1.2.175, whereas the GTX 1070 Ti supports Vulkan 1.4. Display outputs are another point of divergence. The GTX 660 Ti offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the GTX 1070 Ti provides 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. The newer card's display capabilities are more modern, supporting higher resolutions and refresh rates through newer standards.

The Verdict

The data makes the choice clear for performance-oriented users: the GTX 1070 Ti wins every head-to-head benchmark and does so by overwhelming margins. Its OpenCL score is 326% higher, and its Vulkan score is 406% higher than the GTX 660 Ti. For anyone running compute-intensive workloads or modern games that leverage Vulkan, the GTX 1070 Ti is the only rational option based on the benchmark evidence. The GTX 1070 Ti's 8 GB memory capacity, 256-bit bus, and 256.3 GB/s bandwidth give it a massive advantage in memory-heavy scenarios, while its 8.186 TFLOPS FP32 throughput dwarfs the GTX 660 Ti's 2.634 TFLOPS.

However, the GTX 660 Ti still holds relevance in specific contexts. Its 57th percentile ranking is actually one point higher than the GTX 1070 Ti's 56th percentile, and its average benchmark score of 15,063 is within 5.5% of the GTX 1070 Ti's 14,277 average. This suggests that in the specific workloads covered by its benchmark suite, the GTX 660 Ti remains competitive with much newer hardware. Its nearest rivals include the AMD Radeon RX 7600 and NVIDIA GeForce RTX 3050 OEM, both modern entry-level cards, and the GTX 660 Ti trails them by less than 1%. For legacy applications, older DirectX titles, or scenarios where the Kepler architecture's characteristics are sufficient, the GTX 660 Ti can still serve adequately.

The production status of both cards is end-of-life, so neither is a current purchase recommendation for new builds. But for those comparing used or existing hardware, the GTX 1070 Ti's benchmarks are categorically superior in every head-to-head measurement. The GTX 1070 Ti also offers better power efficiency per unit of performance despite its higher 180 W TDP versus 150 W for the GTX 660 Ti, given that it delivers roughly three times the FP32 throughput. The GTX 660 Ti's 2x 6-pin power connectors versus the GTX 1070 Ti's single 8-pin connector further illustrate the architectural efficiency gains of Pascal over Kepler. In summary, the GTX 1070 Ti is the superior performer by every metric in the head-to-head data; the GTX 660 Ti's only advantage is its lower launch MSRP of 299 USD versus 399 USD for the newer card.

Specification Differences

| Specification | GTX 660 Ti | GTX 1070 Ti |

|---|---|---|

| Architecture | Kepler | Pascal |

| Chip | GK104 | GP104 |

| Process Node | 28 nm | 16 nm |

| Transistors | 3,540 million | 7,200 million |

| Die Size | 294 mm² | 314 mm² |

| Transistor Density | 12.0M / mm² | 22.9M / mm² |

| Base Clock | 915 MHz | 1607 MHz |

| Boost Clock | 980 MHz | 1683 MHz |

| Memory Clock | 1502 MHz / 6 Gbps | 2002 MHz / 8 Gbps |

| Memory Size | 2 GB | 8 GB |

| Memory Bus | 192 bit | 256 bit |

| Memory Bandwidth | 144.2 GB/s | 256.3 GB/s |

| Shading Units | 1344 | 2432 |

| TMUs | 112 | 152 |

| ROPs | 24 | 64 |

| Pixel Rate | 27.44 GPixel/s | 107.7 GPixel/s |

| Texture Rate | 109.8 GTexel/s | 255.8 GTexel/s |

| FP32 | 2.634 TFLOPS | 8.186 TFLOPS |

| FP16 | Not specified | 127.9 GFLOPS (1:64) |

| TDP | 150 W | 180 W |

| Power Connectors | 2x 6-pin | 1x 8-pin |

| DirectX | 12 (11_0) | 12 (12_1) |

| Vulkan | 1.2.175 | 1.4 |

| Display Outputs | 2x DVI, 1x HDMI 1.4a, 1x DP 1.2 | 1x DVI, 1x HDMI 2.0, 3x DP 1.4a |

| Length | 241 mm (9.5 inches) | 267 mm (10.5 inches) |

| Release Date | 2012-08-15 | 2017-11-01 |

| Launch MSRP | 299 USD | 399 USD |