NVIDIA GeForce GTX 670MX vs NVIDIA Quadro K4000 Comparison

The NVIDIA Quadro K4000 and NVIDIA GeForce GTX 670MX are both Kepler-generation parts, but they target entirely different segments of the market. The data shows a clear performance hierarchy between them, with the Quadro K4000 winning both shared benchmark tests, though the GTX 670MX counters with a higher transistor budget and a more modern bus interface. These are end-of-life products, but their benchmark scores and architectural specifications still reveal distinct design priorities.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA Quadro K4000 leads with an average benchmark score of 5982, while the NVIDIA GeForce GTX 670MX trails at 5721. The K4000 also holds a higher percentile rank, sitting in the 34th percentile of all GPUs compared to the GTX 670MX's 33rd percentile.

Q: How do the two compare in the Geekbench OpenCL test?

A: The Quadro K4000 scores 6816 in Geekbench OpenCL, which is 11.3% ahead of the GTX 670MX's 6125. This is one of the two head-to-head benchmark wins for the K4000.

Q: What is the biggest performance gap between these two cards?

A: The largest delta is in the Geekbench Vulkan test, where the Quadro K4000 scores 6964 against the GTX 670MX's 5316. That represents a 31% advantage for the K4000, making it the decisive victory in the head-to-head comparison.

Q: Do both cards use the same process node?

A: Yes, both are built on TSMC's 28 nm process. However, their dies differ significantly: the GTX 670MX uses a larger 294 mm² die with 3,540 million transistors, while the Quadro K4000 uses a 221 mm² die with 2,540 million transistors.

Q: Which card has more shading units and texture mapping units?

A: The GeForce GTX 670MX has 960 shading units and 80 TMUs, whereas the Quadro K4000 has 768 shading units and 64 TMUs. Despite this hardware advantage, the GTX 670MX still loses in both shared benchmark tests.

Q: Are there any benchmark results where the GTX 670MX wins?

A: No. The head-to-head data shows zero wins for the GTX 670MX, with the Quadro K4000 taking both Geekbench OpenCL and Geekbench Vulkan. The GTX 670MX's only published benchmark scores are its OpenCL and Vulkan results, both of which are lower than the K4000's corresponding scores.

Architecture Differences

Both GPUs share the Kepler architecture and TSMC's 28 nm manufacturing process, but they diverge in nearly every other physical and functional aspect. The Quadro K4000 is built on the GK106 chip, while the GTX 670MX uses the larger GK104 chip. This difference in chip selection explains the transistor gap: the K4000 packs 2,540 million transistors on a 221 mm² die, yielding a transistor density of 11.5M per mm². The GTX 670MX, by contrast, carries 3,540 million transistors on a 294 mm² die, with a slightly higher density of 12.0M per mm².

The compute resource allocation tells a fascinating story. The GTX 670MX has 960 shading units and 80 texture mapping units, outpacing the K4000's 768 shading units and 64 TMUs. Both cards have 24 ROPs, so that part is equal. However, the K4000 compensates with higher clock rates. Its memory runs at 1404 MHz (5.6 Gbps effective), while the GTX 670MX's memory is clocked at 700 MHz (2.8 Gbps effective). This directly impacts bandwidth: the K4000 delivers 134.8 GB/s versus the GTX 670MX's 67.20 GB/s, despite both using a 192-bit bus and 3 GB of GDDR5.

The FP32 performance figures reflect these trade-offs. The Quadro K4000 achieves 1,244.2 GFLOPS, whereas the GTX 670MX produces 1,153.9 GFLOPS. Pixel and texture rates follow a similar pattern, with the K4000 hitting 12.96 GPixel/s and 51.84 GTexel/s, compared to the GTX 670MX's 12.02 GPixel/s and 48.08 GTexel/s. Neither card features RT cores or tensor cores, and both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

Power and physical design differ markedly. The K4000 has a TDP of 80 W, requires a single 6-pin power connector, and comes in a single-slot form factor measuring 241 mm in length. The GTX 670MX has a lower 75 W TDP and draws power through internal connectors, but its dimensions are not specified in the data, and its display outputs are listed as "Portable Device Dependent," indicating a mobile-oriented design. The bus interface also differs: the K4000 uses PCIe 2.0 x16, while the GTX 670MX uses PCIe 3.0 x16.

The Verdict

The benchmark data presents a unambiguous conclusion for compute workloads: the Quadro K4000 is the stronger performer. It wins both head-to-head tests, with an 11.3% advantage in OpenCL and a decisive 31% lead in Vulkan. Its average benchmark score of 5982 also surpasses the GTX 670MX's 5721 by roughly 4.6%. For any user prioritizing raw compute results in these specific API tests, the K4000 is the clear choice.

However, the GTX 670MX is not without merit for specific use cases. Its GK104 chip provides more shading units and TMUs, which could theoretically benefit certain workloads that scale with those resources. The PCIe 3.0 x16 interface offers double the bandwidth of the K4000's PCIe 2.0 x16, which may matter in memory-transfer-heavy scenarios. Additionally, its lower 75 W TDP and lack of external power connectors make it a more flexible option for systems with tight power or physical constraints.

The data also suggests considering the K4000 in context of its rivals. Its nearest competitor, the NVIDIA Quadro K4000M, scores 5986, which is only 0.1% higher, indicating the K4000 sits at the top of its immediate performance cluster. On the GTX 670MX side, the closest rival is the NVIDIA GeForce GTX 550 Ti, scoring 5731, just 0.2% higher than the 670MX's 5721. These tight margins highlight that both cards are competitive within their respective performance tiers, but the K4000 holds the edge in direct comparison.

For professional users running OpenCL or Vulkan applications, the Quadro K4000's higher scores and better percentile ranking make it the recommended pick. For mobile or power-sensitive deployments where the GTX 670MX's form factor and connectorless design are advantages, the performance gap may be acceptable, but it remains a slower option in every measured test.

Specification Differences

The two cards differ across nearly all core specifications, with only a few shared values. Both use the Kepler architecture, TSMC's 28 nm process, 3 GB of GDDR5 memory, a 192-bit memory bus, 24 ROPs, and identical API support (DirectX 12 (11_0), OpenGL 4.6, Vulkan 1.2.175). Beyond that, the differences are substantial.

| Specification | NVIDIA Quadro K4000 | NVIDIA GeForce GTX 670MX |

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

| Chip | GK106 | GK104 |

| Generation | Quadro Kepler (Kx000) | GeForce 600M |

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

| Die Size | 221 mm² | 294 mm² |

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

| Base Clock | Not specified | 601 MHz |

| Boost Clock | Not specified | 601 MHz |

| Memory Clock | 1404 MHz (5.6 Gbps effective) | 700 MHz (2.8 Gbps effective) |

| Memory Bandwidth | 134.8 GB/s | 67.20 GB/s |

| Shading Units | 768 | 960 |

| TMUs | 64 | 80 |

| Pixel Rate | 12.96 GPixel/s | 12.02 GPixel/s |

| Texture Rate | 51.84 GTexel/s | 48.08 GTexel/s |

| FP32 Performance | 1,244.2 GFLOPS | 1,153.9 GFLOPS |

| TDP | 80 W | 75 W |

| Slot Width | Single-slot | Not specified |

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

| Suggested PSU | 250 W | Not specified |

| Bus Interface | PCIe 2.0 x16 | PCIe 3.0 x16 |

| Display Outputs | 1x DVI, 2x DisplayPort 1.2 | Portable Device Dependent |

| Dimensions | 241 mm (9.5 inches) length, 111 mm (4.4 inches) height | Not specified |

| Release Date | 2013-02-28 | 2012-09-30 |

| Predecessor | Quadro Fermi | GeForce 500M |

| Successor | Quadro Maxwell | GeForce 700M |

| Launch MSRP | 1,269 USD | Not specified |

Head-to-Head Benchmarks

The head-to-head data features only two shared benchmark tests, and the Quadro K4000 wins both. The first is Geekbench OpenCL, where the K4000 scores 6816 against the GTX 670MX's 6125. That is an 11.3% advantage, representing a solid but not overwhelming lead. This result is notable because the GTX 670MX has more shading units and TMUs, yet the K4000's higher memory bandwidth and clock rates allow it to pull ahead in this compute workload.

The second test, Geekbench Vulkan, produces a far more lopsided outcome. The K4000 achieves 6964, while the GTX 670MX manages only 5316. The 31% delta is the single largest performance gap between the two cards and suggests the K4000 has a significant architectural advantage in Vulkan-specific operations. This could be attributed to driver optimizations or the K4000's more efficient memory subsystem, but the data alone confirms a substantial margin.

Beyond the direct comparison, the nearest rival data adds context. The K4000's average score of 5982 places it just 0.1% behind the Quadro K4000M (5986) and 0.2% ahead of the AMD Radeon HD 8750M (5970). The GTX 670MX's 5721 average is 0.2% behind the GTX 550 Ti (5731) and 0.5% ahead of the AMD Radeon HD 8790M (5691). These figures indicate that while the K4000 wins the head-to-head, both cards are clustered tightly with their immediate competitors, meaning the performance differences between similar-tier GPUs are minimal.

The wins tally reflects the benchmark outcomes: the Quadro K4000 secures 2 wins, while the GTX 670MX records 0. This clean sweep in the available tests, combined with the K4000's higher average score and better percentile rank, makes the performance hierarchy unambiguous. The GTX 670MX's hardware advantages in shading units and TMUs do not translate into benchmark victories, underscoring that clock speed, memory bandwidth, and driver efficiency play decisive roles in these workloads.