NVIDIA GeForce GTX 1070 vs NVIDIA GeForce GTX 650 Ti Boost Comparison

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA GeForce GTX 1070 records an average benchmark score of 9780, while the GTX 650 Ti Boost sits at 8067. That is a gap of roughly 1,700 points in favor of the 1070.

Q: How does the GTX 1070 compare to its nearest rivals in overall average score?

A: The GTX 1070's nearest rival is the AMD FirePro W5000 at 9803, which is only 0.2% ahead. The Quadro M2000M is 0.5% ahead, while the Tesla M10 and Tesla C2070 trail by 0.6% and 0.7% respectively. The 1070 sits in a tight cluster around the 9,700 to 9,800 mark.

Q: What about the GTX 650 Ti Boost's position among its peers?

A: The GTX 650 Ti Boost averages 8067, putting it 0.2% behind the GRID K2 at 8080, and 0.2% ahead of the GTX 650 Ti at 8053. It sits 0.3% ahead of both the GTX 880M and the Quadro P5000, the latter scoring 8039. The delta between all four rivals is within a third of a percent.

Q: Which card has more memory and bandwidth?

A: The GTX 1070 carries 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s. The GTX 650 Ti Boost has 2 GB of GDDR5 on a 192-bit bus, yielding 144.2 GB/s. The 1070 has quadruple the capacity and roughly 78% more bandwidth.

Q: Do both cards support modern APIs?

A: Both support DirectX 12 and OpenGL 4.6, but at different feature levels. The GTX 1070 reaches DirectX 12 (12_1) and Vulkan 1.4, while the GTX 650 Ti Boost is limited to DirectX 12 (11_0) and Vulkan 1.2.175.

Q: What is the launch MSRP for each card?

A: The GTX 1070 had a launch MSRP of 379 USD. The GTX 650 Ti Boost launched at 169 USD.

Architecture Differences

The GTX 1070 is built on the GP104 chip using the Pascal architecture, manufactured on a 16 nm process at TSMC. The GTX 650 Ti Boost uses the GK106 chip with the older Kepler architecture, fabricated on a 28 nm process, also at TSMC. The process node difference is substantial: 16 nm versus 28 nm, which directly impacts transistor density. The GP104 packs 7,200 million transistors into a 314 mm² die, giving a density of 22.9M transistors per mm². The GK106 contains 2,540 million transistors on a 221 mm² die, for a density of 11.5M per mm². The Pascal chip nearly doubles the transistor density of the Kepler part.

The GTX 1070 has 1,920 shading units, 120 texture mapping units, and 64 ROPs. The GTX 650 Ti Boost has 768 shading units, 64 TMUs, and 24 ROPs. That is 2.5 times the shader count, nearly double the TMUs, and over 2.5 times the ROPs. The 1070's pixel rate is 107.7 GPixel/s versus 16.51 GPixel/s on the older card, and its texture rate is 202.0 GTexel/s versus 66.05 GTexel/s. FP32 compute is 6.463 TFLOPS against 1.585 TFLOPS, a 4x difference. The GTX 1070 also lists FP16 at 101.0 GFLOPS (1:64), while the GTX 650 Ti Boost has no recorded FP16 capability.

Memory architecture differs significantly. The 1070 uses a 256-bit bus with 8 GB of GDDR5 at 2002 MHz (8 Gbps effective). The 650 Ti Boost uses a 192-bit bus with 2 GB at 1502 MHz (6 Gbps effective). Bandwidth figures are 256.3 GB/s versus 144.2 GB/s. Clock speeds also favor the newer card: the 1070 runs at 1506 MHz base and 1683 MHz boost, while the 650 Ti Boost runs at 980 MHz base and 1032 MHz boost.

Power and physical specifications differ as well. The 1070 has a TDP of 150 W and requires a single 8-pin power connector, with a suggested 450 W PSU. The 650 Ti Boost has a TDP of 134 W, uses a single 6-pin connector, and suggests a 300 W PSU. Both are dual-slot cards, but the 1070 is longer at 267 mm (10.5 inches) versus 241 mm (9.5 inches). The 1070 is 112 mm tall and 40 mm wide; the 650 Ti Boost has no recorded height or width. Display outputs differ: the 1070 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a, while the 650 Ti Boost offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

Where Each One Wins

The GTX 1070 wins every head-to-head benchmark recorded in the database. There are no tests where the GTX 650 Ti Boost comes out on top. The wins span Metal, OpenCL, and Vulkan workloads, which indicates the 1070 holds the advantage across different compute and graphics APIs.

For the GTX 650 Ti Boost, the data shows no winning scenario in the direct comparisons. Its only competitive position is within its own nearest rival group, where it sits within 0.3% of the GRID K2, GTX 650 Ti, GTX 880M, and Quadro P5000. That makes it a middle-of-the-pack performer among its own contemporaries, but it cannot challenge the GTX 1070 in any recorded test.

The GTX 1070's percentile ranking of 47 among all GPUs places it slightly below the midpoint of the entire database. The GTX 650 Ti Boost sits at the 42nd percentile. Both cards are near the middle of the overall distribution, but the 1070 is clearly above the 650 Ti Boost in absolute scores.

Specification Differences

| Specification | GTX 1070 | GTX 650 Ti Boost |

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

| Architecture | Pascal | Kepler |

| Process node | 16 nm | 28 nm |

| Transistors | 7,200 million | 2,540 million |

| Die size | 314 mm² | 221 mm² |

| Transistor density | 22.9M / mm² | 11.5M / mm² |

| Base clock | 1506 MHz | 980 MHz |

| Boost clock | 1683 MHz | 1032 MHz |

| Memory clock | 2002 MHz (8 Gbps effective) | 1502 MHz (6 Gbps effective) |

| Memory size | 8 GB | 2 GB |

| Memory bus width | 256 bit | 192 bit |

| Memory bandwidth | 256.3 GB/s | 144.2 GB/s |

| Shading units | 1920 | 768 |

| TMUs | 120 | 64 |

| ROPs | 64 | 24 |

| Pixel rate | 107.7 GPixel/s | 16.51 GPixel/s |

| Texture rate | 202.0 GTexel/s | 66.05 GTexel/s |

| FP32 | 6.463 TFLOPS | 1.585 TFLOPS |

| FP16 | 101.0 GFLOPS (1:64) | null |

| TDP | 150 W | 134 W |

| Power connectors | 1x 8-pin | 1x 6-pin |

| Suggested PSU | 450 W | 300 W |

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

| Vulkan | 1.4 | 1.2.175 |

| Display outputs | 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 |

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

| Height | 112 mm (4.4 inches) | null |

| Width | 40 mm (1.6 inches) | null |

| Launch MSRP | 379 USD | 169 USD |

Head-to-Head Benchmarks

The database records three direct head-to-head tests between these two cards, and the GTX 1070 wins all three by wide margins.

In Geekbench Metal, the GTX 1070 scores 18801 against 4858 for the GTX 650 Ti Boost. That is a delta of 287%, meaning the 1070 delivers nearly four times the Metal performance. This is the smallest relative gap of the three tests, but it is still a dominant result.

Geekbench OpenCL shows an even larger gap. The GTX 1070 posts 44700, while the GTX 650 Ti Boost manages 9302. The delta is 380.5%, the largest margin in any recorded test. OpenCL is a common compute workload, and the 4.8x difference reflects the massive shader and memory advantages of the Pascal card.

Geekbench Vulkan narrows the gap somewhat, but the 1070 still wins decisively. The score is 22121 versus 10041, a delta of 120.3%. The 650 Ti Boost's Vulkan score is actually its best relative showing in the head-to-head, likely because the workload scales differently, but it still trails by more than double.

The GTX 1070's average benchmark score of 9780 comes from a broader set of tests, including several Passmark entries where it scores 13498 in G3D, 846 in G2D, 197 in DirectX 9, 100 in DirectX 11, 82 in DirectX 10, 48 in DirectX 12, and 6102 in GPU compute. The 650 Ti Boost only has three recorded benchmarks, all Geekbench tests, so its average of 8067 is computed from a much smaller sample. That sample difference matters when interpreting the averages, but the head-to-head results are unambiguous.

The Verdict

The GTX 1070 is the clear choice in every recorded metric. It wins all three head-to-head benchmarks with deltas ranging from 120.3% to 380.5%, and it holds a roughly 1,700 point advantage in average benchmark score. Its architecture advantages are substantial: a 16 nm process versus 28 nm, 7,200 million transistors versus 2,540 million, 1920 shading units versus 768, and 8 GB of memory versus 2 GB. Every spec sheet comparison favors the 1070, and the benchmark data confirms it.

The GTX 650 Ti Boost is not without merit within its own generation. Its average score of 8067 places it within 0.3% of four closely matched rivals, meaning it performs on par with cards like the GRID K2 and GTX 880M. For a user constrained to that performance tier, the 650 Ti Boost is a reasonable middle-ground option. But it cannot compete with the GTX 1070 in any direct test.

The percentile rankings tell a similar story. The GTX 1070 sits at the 47th percentile of all GPUs, while the 650 Ti Boost sits at the 42nd. Both are near the median, but the 1070's position reflects a much faster card in absolute terms, especially considering the 650 Ti Boost's percentile is drawn from a much smaller benchmark pool.

For practical purposes, the GTX 1070 is the card to pick if the workload involves modern APIs, high-resolution textures, or compute tasks. The 8 GB frame buffer and 256.3 GB/s bandwidth give it headroom that the 2 GB, 144.2 GB/s configuration of the 650 Ti Boost simply cannot match. The 650 Ti Boost is only relevant if the system is limited to a 300 W PSU and a single 6-pin connector, or if the workload is strictly tied to the older Kepler feature set. Even then, the 1070's higher TDP of 150 W versus 134 W is modest, and its single 8-pin connector is a minor trade for the performance gain.

The data does not support any scenario where the GTX 650 Ti Boost outperforms the GTX 1070. Buyers should treat the 1070 as the superior option across the board.