NVIDIA GeForce GTX 650 Ti vs NVIDIA GeForce GTX 670M Comparison

NVIDIA GeForce GTX 650 Ti vs NVIDIA GeForce GTX 670M

The benchmark data records a single direct comparison between these two NVIDIA mobile and desktop parts, and the result is decisive. The GeForce GTX 650 Ti wins the only head-to-head test, the Geekbench OpenCL workload, by 20.9 percent. That margin, combined with the architectural gap between the two, places the GTX 650 Ti clearly ahead in raw compute performance. The GTX 670M, however, is not without its own advantages, primarily in memory capacity and power envelope, which we detail below.

Where Each One Wins

The GeForce GTX 650 Ti wins the only recorded benchmark comparison. In the Geekbench OpenCL test, it scores 7,877 points against the GTX 670M's 6,513 points. That is a 20.9 percent lead for the desktop card. Beyond that single test, the broader database scores reinforce this margin: the GTX 650 Ti averages 8,053 points across its recorded benchmarks, while the GTX 670M averages 6,513 points. The GTX 650 Ti also has a second recorded benchmark, Geekbench Vulkan, where it scores 8,229 points, a figure the GTX 670M cannot match because it has no Vulkan result in the database.

The GTX 670M does not win any recorded benchmark. Its sole advantage in the data is memory capacity: it has 1536 MB of VRAM compared to the GTX 650 Ti's 1024 MB. That larger frame buffer could matter in texture-heavy workloads at higher resolutions, but no benchmark result in the database confirms this advantage. In every measured metric, the GTX 650 Ti is the faster card. The GTX 670M, however, is a mobile part with a lower power draw, which we address in the architecture section.

Architecture Differences

The two cards come from different NVIDIA architectures and process nodes. The GTX 650 Ti uses the GK106S chip, built on the Kepler architecture, fabricated by TSMC on a 28 nm process. It packs 2,540 million transistors into a die size of 221 mm². The GTX 670M, in contrast, uses the GF114 chip, based on the older Fermi 2.0 architecture, also made by TSMC but on a 40 nm process. That older node means the GTX 670M has fewer transistors, 1,950 million, but a larger die, 332 mm². Transistor density tells the story clearly: the GTX 650 Ti has 11.5 million transistors per square millimeter, while the GTX 670M has only 5.9 million.

Compute resources also differ substantially. The GTX 650 Ti has 768 shading units, 64 texture mapping units, and 16 raster operation units. The GTX 670M has 336 shading units, 56 TMUs, and 24 ROPs. The GTX 650 Ti therefore has more than double the shader count and more TMUs, while the GTX 670M has more ROPs. Pixel rate reflects this: the GTX 650 Ti delivers 14.85 GPixel/s versus the GTX 670M's 8.372 GPixel/s. Texture rate is similarly lopsided: 59.39 GTexel/s for the GTX 650 Ti, 33.49 GTexel/s for the GTX 670M. FP32 compute is 1,425.4 GFLOPS for the GTX 650 Ti versus 803.7 GFLOPS for the GTX 670M.

Memory architectures also diverge. The GTX 650 Ti uses a 128-bit bus with GDDR5 memory running at an effective 5.4 Gbps, yielding 86.40 GB/s of bandwidth. The GTX 670M has a wider 192-bit bus but slower GDDR5 memory at an effective 3 Gbps, resulting in 72.00 GB/s of bandwidth. The GTX 670M's extra memory, 1536 MB versus 1024 MB, comes at a bandwidth cost. Power and physical design differ too: the GTX 650 Ti has a TDP of 110 W, is single-slot, and requires a 1x 6-pin power connector with a suggested PSU of 300 W. The GTX 670M has a TDP of 75 W, comes as an MXM module, and uses no external power connectors.

Head-to-Head Benchmarks

The database records one head-to-head benchmark: Geekbench OpenCL. The GTX 650 Ti scores 7,877 points, and the GTX 670M scores 6,513 points. The delta is 20.9 percent in favor of the GTX 650 Ti. That is a substantial margin for a compute workload, reflecting the GTX 650 Ti's higher shader count, faster clocked memory, and newer architecture.

Putting that score into context with nearest rivals helps. The GTX 650 Ti's average benchmark score of 8,053 places it within 0.2 percent of the NVIDIA GeForce GTX 880M (8,040 points) and within 0.2 percent of the NVIDIA Quadro P5000 (8,039 points). It sits 0.2 percent behind the GTX 650 Ti Boost (8,067 points) and 0.3 percent behind the NVIDIA GRID K2 (8,080 points). This means the GTX 650 Ti is essentially at parity with a high-end mobile GPU from a later generation, despite being a desktop part from 2012.

The GTX 670M's average score of 6,513 puts it in a lower tier. It is 0.3 percent ahead of the NVIDIA GeForce GT 555M (6,493 points) and 0.5 percent ahead of the NVIDIA Quadro M5000M (6,481 points). It also leads the AMD Radeon Vega 10 Mobile (6,476 points) by 0.6 percent. However, it trails the Intel UHD Graphics P750 (6,554 points) by 0.6 percent. The GTX 670M's nearest rivals are a mix of older mobile parts and integrated graphics, confirming its position as a mid-range mobile GPU of its era.

In the only direct comparison, the GTX 650 Ti leads by a wide margin. The GTX 670M has no recorded Vulkan score, so its compute potential is limited to the OpenCL result. The GTX 650 Ti's Vulkan score of 8,229 points is higher than its own OpenCL score, suggesting the Kepler architecture handles modern APIs well, but no comparable data exists for the GTX 670M.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The GeForce GTX 650 Ti has an average benchmark score of 8,053 points, while the GeForce GTX 670M averages 6,513 points.

Q: How much faster is the GTX 650 Ti in OpenCL compute?

A: The GTX 650 Ti scores 7,877 points in Geekbench OpenCL, which is 20.9 percent higher than the GTX 670M's 6,513 points.

Q: Does the GTX 670M have any advantage in memory?

A: Yes, the GTX 670M has 1536 MB of VRAM compared to the GTX 650 Ti's 1024 MB, although the GTX 650 Ti has higher memory bandwidth at 86.40 GB/s versus 72.00 GB/s.

Q: What are the power requirements for each card?

A: The GTX 650 Ti has a TDP of 110 W and requires a 1x 6-pin power connector with a suggested PSU of 300 W. The GTX 670M has a TDP of 75 W and uses no external power connectors.

Q: Which card has more shading units?

A: The GTX 650 Ti has 768 shading units, while the GTX 670M has 336 shading units.

Q: Are there any benchmark tests where the GTX 670M wins?

A: No, the database records no benchmark wins for the GTX 670M. The GTX 650 Ti wins the sole head-to-head test.

The Verdict

The data is unambiguous. The GeForce GTX 650 Ti outperforms the GeForce GTX 670M in every recorded benchmark metric. Its OpenCL score is 20.9 percent higher, its average score is 23.7 percent higher (8,053 versus 6,513), and it has a Vulkan result where the GTX 670M has none. The GTX 650 Ti also has superior compute resources: more shading units, more TMUs, higher pixel and texture rates, and nearly double the FP32 throughput.

The GTX 670M's strengths are narrower. It has 50 percent more VRAM (1536 MB versus 1024 MB) and a lower TDP (75 W versus 110 W). Those traits make it suitable for portable systems where power draw and memory capacity are priorities. But the GTX 670M's older Fermi 2.0 architecture, larger die on a 40 nm process, and lower bandwidth put it at a clear disadvantage in raw performance.

For users choosing between these two, the decision depends on the use case. If the goal is maximum compute or gaming performance in a desktop context, the GTX 650 Ti is the obvious choice. Its benchmark scores place it at parity with the GTX 880M, a much later mobile flagship, and its Vulkan support extends its relevance. If the goal is a low-power mobile solution with more VRAM, the GTX 670M fits that niche, but the data shows it is simply a slower card. The GTX 650 Ti wins the comparison on merit of measured performance.

Specification Differences

| Specification | NVIDIA GeForce GTX 650 Ti | NVIDIA GeForce GTX 670M |

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

| Chip | GK106S | GF114 |

| Architecture | Kepler | Fermi 2.0 |

| Process Node | 28 nm | 40 nm |

| Transistors | 2,540 million | 1,950 million |

| Die Size | 221 mm² | 332 mm² |

| Transistor Density | 11.5M / mm² | 5.9M / mm² |

| Memory Size | 1024 MB | 1536 MB |

| Memory Type | GDDR5 | GDDR5 |

| Memory Bus Width | 128 bit | 192 bit |

| Memory Clock | 1350 MHz, 5.4 Gbps effective | 750 MHz, 3 Gbps effective |

| Memory Bandwidth | 86.40 GB/s | 72.00 GB/s |

| Shading Units | 768 | 336 |

| TMUs | 64 | 56 |

| ROPs | 16 | 24 |

| Pixel Rate | 14.85 GPixel/s | 8.372 GPixel/s |

| Texture Rate | 59.39 GTexel/s | 33.49 GTexel/s |

| FP32 | 1,425.4 GFLOPS | 803.7 GFLOPS |

| TDP | 110 W | 75 W |

| Slot Width | Single-slot | MXM Module |

| Power Connectors | 1x 6-pin | None |

| Suggested PSU | 300 W | None |

| Bus Interface | PCIe 3.0 x16 | MXM-B (3.0) |

| Display Outputs | 2x DVI, 1x mini-HDMI 1.4a | Portable Device Dependent |

| DirectX Support | 12 (11_0) | 12 (11_0) |

| OpenGL Support | 4.6 | 4.6 |

| Vulkan Support | 1.2.175 | None |

| Release Date | 2012-10-08 | 2012-03-21 |

| Launch MSRP | 149 USD | None |