NVIDIA GeForce GTX 670MX vs NVIDIA GeForce GTX 765M Comparison

The NVIDIA GeForce GTX 670MX and GTX 765M are both end-of-life Kepler mobile GPUs, but they target different performance profiles despite sharing the same 28 nm process node and 75 W TDP. The GTX 765M, released later as part of the GeForce 700M generation, wins both head-to-head benchmarks decisively, while the GTX 670MX counters with a larger memory capacity and a wider memory bus. This analysis separates the two based on benchmark results, architectural differences, and specification sheets.

Where Each One Wins

The GTX 765M is the outright winner in raw compute benchmarks. In the Geekbench OpenCL test, it scores 7176 against the GTX 670MX’s 6125, a 14.6% advantage. In the Vulkan test, the gap widens: the GTX 765M scores 6714 versus 5316, a 20.8% lead. These are the only two head-to-head tests available, and the GTX 765M wins both, giving it a clean 2-0 record.

However, the GTX 670MX is not without its own strengths. It carries 3 GB of GDDR5 memory compared to the GTX 765M’s 2 GB, and it uses a 192-bit memory bus versus the GTX 765M’s 128-bit bus. While the GTX 765M has slightly higher memory bandwidth (64.13 GB/s vs 67.20 GB/s for the GTX 670MX — wait, the GTX 670MX actually has higher bandwidth), the 670MX’s larger frame buffer is an advantage for workloads that exceed 2 GB, such as high-resolution textures or multi-monitor setups. For pure compute throughput, the GTX 765M wins; for memory capacity, the GTX 670MX wins.

The GTX 670MX also has a higher transistor density (12.0M / mm² vs 11.5M / mm²) and more shading units (960 vs 768), TMUs (80 vs 64), and ROPs (24 vs 16). But the GTX 765M’s higher clock speeds (base 797 MHz vs 601 MHz, boost 863 MHz vs 601 MHz) more than compensate, resulting in higher pixel rate (13.81 GPixel/s vs 12.02 GPixel/s), texture rate (55.23 GTexel/s vs 48.08 GTexel/s), and FP32 throughput (1,325.6 GFLOPS vs 1,153.9 GFLOPS).

Architecture Differences

Both GPUs are built on NVIDIA’s Kepler architecture, but they use different chips. The GTX 670MX uses the GK104 chip, which is physically larger (294 mm² die size) and packs more transistors (3,540 million) than the GTX 765M’s GK106 chip (221 mm², 2,540 million). The GK104 is a higher-tier chip in the Kepler lineup, which explains the 670MX’s superior core counts: 960 shading units, 80 TMUs, and 24 ROPs versus the 765M’s 768 shading units, 64 TMUs, and 16 ROPs.

The GTX 670MX is a GeForce 600M generation part, released on 2012-09-30, while the GTX 765M belongs to the GeForce 700M generation, released on 2013-05-29. The 765M’s predecessor is the GeForce 600M, which means it is a direct successor to the 670MX’s generation. The 670MX’s predecessor is the GeForce 500M, and its successor is the GeForce 700M — so the 765M is the newer architecture revision.

Both use the same API support: DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. Both are manufactured by TSMC on a 28 nm process. Neither has RT cores or tensor cores. The memory type is GDDR5 for both, but the 670MX has 3 GB on a 192-bit bus while the 765M has 2 GB on a 128-bit bus. The effective memory clock differs: the 670MX runs at 2.8 Gbps effective, while the 765M runs at 4 Gbps effective. Despite the 765M’s faster memory clock, the 670MX’s wider bus gives it slightly higher bandwidth (67.20 GB/s vs 64.13 GB/s).

The 765M uses an MXM-B (3.0) bus interface and is listed as an MXM Module, while the 670MX uses PCIe 3.0 x16. Both have no power connectors and draw 75 W. Both have display outputs described as "Portable Device Dependent."

Head-to-Head Benchmarks

In the Geekbench OpenCL test, the GTX 765M scores 7176 against the GTX 670MX’s 6125, a 14.6% delta. This is a substantial lead in a compute-oriented workload. The GTX 765M’s higher FP32 throughput (1,325.6 GFLOPS vs 1,153.9 GFLOPS) likely drives this result, despite the 670MX having more cores.

The Geekbench Vulkan test shows an even larger gap. The GTX 765M scores 6714, while the GTX 670MX scores 5316, a 20.8% delta. Vulkan workloads often benefit from higher clock speeds and better memory efficiency, both of which favor the 765M. The 765M’s boost clock of 863 MHz is 43.6% higher than the 670MX’s fixed 601 MHz, which explains why the 765M wins despite having fewer cores.

There are no other head-to-head benchmarks in the data. The GTX 670MX has a higher average benchmark score (5721) than the GTX 765M (5501), but this is misleading because the 670MX only has two benchmark entries (OpenCL and Vulkan) while the 765M includes a Metal score (2612) that drags its average down. In the two tests where both are measured, the 765M wins both.

The GTX 670MX’s percentile rank is 33rd among all GPUs, while the GTX 765M sits at the 32nd percentile. The GTX 765M’s nearest rivals include the NVIDIA GeForce MX130 (deltaPct -0.1), AMD Radeon R7 M440 (0.3), and NVIDIA Quadro M4000 (0.6). The GTX 670MX’s nearest rivals include the Intel Iris Pro Graphics P6300 (0.2), NVIDIA GeForce GTX 550 Ti (-0.2), and AMD Radeon HD 8790M (0.5). Both GPUs sit in a similar performance tier, but the 765M is faster in the tests that matter.

FAQ

Q: Which GPU is faster in compute benchmarks?

A: The GTX 765M wins both head-to-head tests. It scores 7176 in Geekbench OpenCL versus 6125 for the GTX 670MX (14.6% faster), and 6714 in Geekbench Vulkan versus 5316 (20.8% faster).

Q: Does the GTX 670MX have any advantages over the GTX 765M?

A: Yes. The GTX 670MX has 3 GB of GDDR5 memory compared to 2 GB on the GTX 765M, and a 192-bit memory bus versus 128-bit. It also has more shading units (960 vs 768), TMUs (80 vs 64), and ROPs (24 vs 16).

Q: Why does the GTX 765M win if the GTX 670MX has more cores?

A: The GTX 765M operates at much higher clock speeds. Its base clock is 797 MHz and boost clock is 863 MHz, while the GTX 670MX runs at a fixed 601 MHz. This results in higher pixel rate (13.81 GPixel/s vs 12.02 GPixel/s) and FP32 throughput (1,325.6 GFLOPS vs 1,153.9 GFLOPS).

Q: What are the memory bandwidth figures for these GPUs?

A: The GTX 670MX has a bandwidth of 67.20 GB/s, slightly higher than the GTX 765M’s 64.13 GB/s. The 765M uses faster memory (4 Gbps effective vs 2.8 Gbps effective) but a narrower 128-bit bus.

Q: Are these GPUs from the same architecture generation?

A: Both use NVIDIA’s Kepler architecture and a 28 nm TSMC process, but the GTX 670MX is from the GeForce 600M generation (released 2012-09-30) while the GTX 765M is from the GeForce 700M generation (released 2013-05-29). The 765M is the direct successor to the 600M line.

Q: What is the TDP of each GPU?

A: Both have a TDP of 75 W and require no power connectors.

Specification Differences

| Specification | GTX 670MX | GTX 765M |

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

| Chip | GK104 | GK106 |

| Generation | GeForce 600M | GeForce 700M |

| Transistors | 3,540 million | 2,540 million |

| Die Size | 294 mm² | 221 mm² |

| Transistor Density | 12.0M / mm² | 11.5M / mm² |

| Base Clock | 601 MHz | 797 MHz |

| Boost Clock | 601 MHz | 863 MHz |

| Memory Clock | 2.8 Gbps effective | 4 Gbps effective |

| Memory Size | 3 GB | 2 GB |

| Memory Bus Width | 192 bit | 128 bit |

| Memory Bandwidth | 67.20 GB/s | 64.13 GB/s |

| Shading Units | 960 | 768 |

| TMUs | 80 | 64 |

| ROPs | 24 | 16 |

| Pixel Rate | 12.02 GPixel/s | 13.81 GPixel/s |

| Texture Rate | 48.08 GTexel/s | 55.23 GTexel/s |

| FP32 | 1,153.9 GFLOPS | 1,325.6 GFLOPS |

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

| Slot Width | — | MXM Module |

| Release Date | 2012-09-30 | 2013-05-29 |

| Predecessor | GeForce 500M | GeForce 600M |

| Successor | GeForce 700M | GeForce 800M |

The Verdict

The GTX 765M is the better choice for anyone prioritizing raw compute performance. It wins both available benchmarks by margins of 14.6% and 20.8%, and its higher clock speeds give it superior pixel, texture, and FP32 rates despite having fewer cores. If you are running OpenCL or Vulkan workloads, the 765M is clearly the faster part.

The GTX 670MX is the better choice if memory capacity is your bottleneck. Its 3 GB frame buffer and 192-bit bus provide more headroom for large textures or datasets, and its bandwidth is slightly higher (67.20 GB/s vs 64.13 GB/s). It also has more shading units, TMUs, and ROPs, which could matter in specific driver or workload scenarios where parallelism outweighs clock speed.

Both GPUs are end-of-life and sit at the 32nd-33rd percentile of all GPUs, meaning neither is a modern performer. The GTX 765M’s nearest rivals include the NVIDIA GeForce MX130 and AMD Radeon R7 M440, while the GTX 670MX trades blows with the Intel Iris Pro Graphics P6300 and NVIDIA GeForce GTX 550 Ti. In practical terms, the 765M is the stronger all-around chip, but the 670MX offers a larger memory pool that some users may find indispensable. Choose the 765M for speed; choose the 670MX for memory.