NVIDIA GeForce GTX 1070 Ti vs NVIDIA GeForce GTX 1080 Comparison

Head-to-Head Benchmarks

The benchmark data shows a clear overall winner, but the one loss for the GTX 1080 is so dramatic that it deserves attention first. In the Geekbench Vulkan test, the GTX 1070 Ti scores 61,006 against the GTX 1080's 30,398, a 100.7% advantage. That is not a small margin; it is a doubling of performance in a single API workload. This is the only head-to-head win for the 1070 Ti, but it is a massive one.

Every other recorded test goes to the GTX 1080. The largest margin comes in Passmark DirectX 11, where the 1080 scores 124 versus 107, a 13.7% lead. That is the biggest conventional gap between the two cards in the entire dataset. The DirectX 12 Passmark test shows a 7.3% advantage for the 1080 (55 vs. 51), and the DirectX 10 test shows a smaller 3.2% lead (93 vs. 90). DirectX 9 is nearly a tie: 211 vs. 207, a 1.9% edge for the 1080.

In 3DMark Steel Nomad DX12, the 1080 scores 1,560 against 1,471, a 5.7% lead. Geekbench Metal shows the 1080 ahead by 6.9% (23,824 vs. 22,180), while Geekbench OpenCL shows a 3.9% margin (51,204 vs. 49,219). Passmark G3D, the general 3D graphics score, puts the 1080 at 15,586 and the 1070 Ti at 14,674, a 5.9% difference. The GPU compute test shows the same 5.9% gap (7,614 vs. 7,165). Passmark G2D, a 2D graphics test, is close: 888 vs. 876, a 1.4% lead for the 1080.

Across eleven tests, the GTX 1080 wins ten. The average benchmark score in the database tells a more complex story: the 1070 Ti averages 14,277, while the 1080 averages 11,960. That seems backwards given the head-to-head results, but it reflects the different nearest rival sets. The 1070 Ti sits at the 56th percentile of all GPUs, while the 1080 sits at the 51st percentile. The 1070 Ti's nearest rivals include the Intel Iris Xe MAX Graphics (14,315, only 0.3% higher) and the AMD Radeon Vega 11 (14,352, 0.5% higher), all within 0.7%. The 1080's nearest rivals are lower-scoring cards: the GTX 960A (11,998, 0.3% lower), the RX 6500 XT (11,842, 1% higher), and the GTX 1660 (11,680, 2.4% higher). The average score is dragged down by the 1080's extremely low Vulkan result, which is not representative of its overall strength.

Architecture Differences

Both cards are built on the same GP104 chip, the Pascal architecture, and the same 16 nm TSMC process. Transistor count is identical at 7,200 million, and die size is identical at 314 mm², giving both a transistor density of 22.9M per mm². The fundamental silicon is the same; the differences come from how NVIDIA configured it.

The GTX 1080 has 2,560 shading units, 160 texture mapping units, and 64 ROPs. The GTX 1070 Ti has 2,432 shading units, 152 TMUs, and the same 64 ROPs. That is 128 more shading units and 8 more TMUs on the 1080, with ROP count unchanged. These extra units translate directly into higher peak rates: the 1080 reaches 8.873 TFLOPS of FP32 compute, 277.3 GTexel/s texture fill, and 110.9 GPixel/s pixel fill. The 1070 Ti sits at 8.186 TFLOPS, 255.8 GTexel/s, and 107.7 GPixel/s.

Memory is a major split. The 1070 Ti uses 8 GB of GDDR5 on a 256-bit bus, with memory clocked at 2002 MHz (8 Gbps effective) and bandwidth of 256.3 GB/s. The 1080 uses 8 GB of GDDR5X on the same 256-bit bus, clocked at 1251 MHz (10 Gbps effective), delivering 320.3 GB/s. The GDDR5X memory gives the 1080 a 25% bandwidth advantage, which matters in texture-heavy and high-resolution workloads.

Clock behavior also differs. Both have a base clock of 1607 MHz, but the boost clock is higher on the 1080: 1733 MHz versus 1683 MHz. That 50 MHz boost difference compounds with the extra shading units. FP16 performance follows the same pattern: 138.6 GFLOPS on the 1080 versus 127.9 GFLOPS on the 1070 Ti, both at a 1:64 ratio.

Power and physical characteristics are identical. Both are rated at 180 W TDP, dual-slot cards, use a single 8-pin power connector, and recommend a 450 W PSU. Dimensions match exactly: 267 mm length, 112 mm height, 40 mm width. Display outputs are the same: 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a. Both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. Neither has ray tracing cores or tensor cores, as expected for the Pascal generation.

Where Each One Wins

The GTX 1070 Ti wins in exactly one scenario: the Geekbench Vulkan test. That result is extraordinary, doubling the 1080's score. If a workload is heavily optimized for Vulkan and behaves like that benchmark, the 1070 Ti is the better choice. The data does not explain why the 1080 scores so low in Vulkan, but the recorded result is what it is.

The GTX 1080 wins everywhere else. It is ahead in DirectX 9, 10, 11, and 12 tests, in Metal, in OpenCL, in 3DMark Steel Nomad DX12, in both G3D and G2D Passmark tests, and in GPU compute. The margins range from 1.4% (G2D) to 13.7% (DirectX 11). For most gaming and compute workloads, the 1080 is the faster card, typically by 4% to 7% in the modern API tests.

The compute result is worth noting: the 1080 leads by 5.9% in Passmark GPU compute, matching its G3D lead. The 1080's extra shading units and higher memory bandwidth give it a consistent edge in raw throughput. The 1070 Ti is not a weak card; its average score of 14,277 places it at the 56th percentile, above the 1080's 51st percentile, but that is due to the Vulkan anomaly. In head-to-head testing, the 1080 is the better performer in ten of eleven tests.

Specification Differences

| Specification | GeForce GTX 1070 Ti | GeForce GTX 1080 |

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

| Boost Clock | 1683 MHz | 1733 MHz |

| Memory Clock | 2002 MHz (8 Gbps effective) | 1251 MHz (10 Gbps effective) |

| Memory Type | GDDR5 | GDDR5X |

| Memory Bandwidth | 256.3 GB/s | 320.3 GB/s |

| Shading Units | 2432 | 2560 |

| TMUs | 152 | 160 |

| Pixel Rate | 107.7 GPixel/s | 110.9 GPixel/s |

| Texture Rate | 255.8 GTexel/s | 277.3 GTexel/s |

| FP32 Performance | 8.186 TFLOPS | 8.873 TFLOPS |

| FP16 Performance | 127.9 GFLOPS (1:64) | 138.6 GFLOPS (1:64) |

| Release Date | 2017-11-01 | 2016-05-26 |

Identical specifications include base clock (1607 MHz), memory size (8 GB), bus width (256 bit), ROPs (64), TDP (180 W), slot width (dual-slot), power connector (1x 8-pin), suggested PSU (450 W), bus interface (PCIe 3.0 x16), display outputs, APIs, dimensions, process node (16 nm), transistors (7,200 million), die size (314 mm²), and transistor density (22.9M / mm²). Both are end-of-life products from the GeForce 10-series, with the GeForce 900 as predecessor and GeForce 20 as successor.

FAQ

Q: Which card is faster overall?

A: The GTX 1080 wins ten of eleven head-to-head benchmarks. Its largest lead is 13.7% in Passmark DirectX 11, and it leads by 5.9% in both Passmark G3D and GPU compute.

Q: Why does the GTX 1070 Ti have a higher average benchmark score?

A: The 1070 Ti averages 14,277 versus 11,960 for the 1080. This is because the 1080's Geekbench Vulkan score is only 30,398, while the 1070 Ti scores 61,006, a 100.7% difference that heavily pulls down the 1080's average.

Q: Is there any test where the GTX 1070 Ti beats the GTX 1080?

A: Yes, the Geekbench Vulkan test. The 1070 Ti scores 61,006 against 30,398, a 100.7% advantage. No other recorded test favors the 1070 Ti.

Q: What are the memory differences?

A: The 1070 Ti uses 8 GB of GDDR5 with 256.3 GB/s bandwidth. The 1080 uses 8 GB of GDDR5X with 320.3 GB/s bandwidth. Both have a 256-bit bus, but the 1080's memory clock runs at 10 Gbps effective versus 8 Gbps.

Q: Do both cards use the same power and cooling requirements?

A: Yes. Both are rated at 180 W TDP, use a single 8-pin power connector, recommend a 450 W PSU, and are dual-slot cards with identical dimensions of 267 mm x 112 mm x 40 mm.

Q: Which card has more shading units?

A: The GTX 1080 has 2,560 shading units, 128 more than the 1070 Ti's 2,432. It also has 160 TMUs versus 152, and the same 64 ROPs.

The Verdict

The data points to the GTX 1080 as the superior card for almost every task. It wins ten of eleven head-to-head benchmarks, with leads from 1.4% in Passmark G2D to 13.7% in DirectX 11. The 1080's higher boost clock (1733 MHz vs. 1683 MHz), extra shading units (2,560 vs. 2,432), and faster GDDR5X memory (320.3 GB/s vs. 256.3 GB/s) all contribute to its consistent margins. The launch MSRP for the 1080 was 599 USD, while the 1070 Ti launched at 399 USD.

The only reason to pick the 1070 Ti is if your workload is dominated by the specific Vulkan scenario captured in the Geekbench Vulkan test, where it doubles the 1080's score. That is a real, recorded result, and it is not a small effect. But it is one test out of eleven. For DirectX gaming, OpenCL compute, Metal, and general 3D performance, the 1080 is measurably ahead.

For a builder choosing between these two end-of-life Pascal cards, the 1080 is the safer pick for broad compatibility and raw speed. The 1070 Ti is a niche choice for Vulkan-focused use cases. The percentile rankings (56th for the 1070 Ti, 51st for the 1080) are misleading due to the Vulkan outlier; head-to-head data is the more reliable guide. If you want the fastest card in the most tests, take the GTX 1080. If you specifically need Vulkan performance as measured here, the GTX 1070 Ti is the one.