Intel UHD Graphics 730 vs NVIDIA GeForce GTX 670MX Comparison

Intel UHD Graphics 730 and NVIDIA GeForce GTX 670MX occupy the same 33rd percentile of all GPUs, yet they achieve parity through entirely different means. The Intel part is a modern integrated solution from Rocket Lake, while the GTX 670MX is a Kepler-based discrete mobile chip from 2012. Benchmark data shows a split decision: NVIDIA wins OpenCL by 2.2%, Intel wins Vulkan by 10.4%. This makes the choice dependent on workload and platform rather than raw performance dominance.

Where Each One Wins

The Intel UHD Graphics 730 is the clear winner in Vulkan-based workloads. Its Geekbench Vulkan score of 5870 surpasses the GTX 670MX’s 5316 by 10.4%. This is a substantial margin for a modern API, reflecting Intel’s Generation 12.1 architecture with Vulkan 1.4 support versus NVIDIA’s Kepler with Vulkan 1.2.175. For any application or game leveraging Vulkan, the Intel part holds a decisive edge.

The NVIDIA GeForce GTX 670MX takes the OpenCL crown. Its Geekbench OpenCL score of 6125 edges out Intel’s 5988 by 2.2%. While the margin is slim, it is consistent across the board: the GTX 670MX also posts a higher average benchmark score of 5721 compared to Intel’s 5929. Wait — that is incorrect. The Intel part has a higher average score (5929 vs 5721). But the GTX 670MX wins the specific OpenCL test. In practice, OpenCL compute tasks, such as certain rendering or physics simulations, will run slightly faster on the NVIDIA chip.

For raw throughput metrics, the GTX 670MX dominates. It delivers 1,153.9 GFLOPS FP32, more than double Intel’s 499.2 GFLOPS. Its texture rate of 48.08 GTexel/s is over three times Intel’s 15.60 GTexel/s. The pixel rate also favors NVIDIA at 12.02 GPixel/s versus 10.40 GPixel/s. These numbers suggest the discrete GPU has more raw shader power, even if software optimizations in Vulkan narrow the gap.

Architecture Differences

The architectural gap is generational. Intel’s UHD Graphics 730 uses the Rocket Lake chip built on Intel’s 14 nm+++ process. It is a Generation 12.1 design with 192 shading units, 12 texture mapping units, and 8 ROPs. The GTX 670MX uses the GK104 chip on TSMC’s 28 nm process, packing 960 shading units, 80 TMUs, and 24 ROPs. NVIDIA’s chip contains 3,540 million transistors on a 294 mm² die, while Intel’s transistor count is not specified in the data.

Memory configuration differs significantly. The Intel part uses System Shared memory — no dedicated VRAM, with bandwidth described as System Dependent. The GTX 670MX has 3 GB of GDDR5 on a 192-bit bus, delivering 67.20 GB/s of dedicated bandwidth. This is a critical advantage for the NVIDIA part in memory-intensive workloads, as it does not compete with the CPU for system RAM.

Clock behavior also diverges. Intel runs at a 300 MHz base with a 1300 MHz boost. NVIDIA runs flat at 601 MHz for both base and boost. The higher boost clock on Intel helps compensate for fewer shaders, but the sheer shader count difference (192 vs 960) is overwhelming. API support shows Intel ahead with DirectX 12 (12_1) versus NVIDIA’s DirectX 12 (11_0), and Vulkan 1.4 versus 1.2.175. Both support OpenGL 4.6.

FAQ

Q: Which GPU has better Vulkan performance?

A: The Intel UHD Graphics 730 wins the Geekbench Vulkan test with a score of 5870, beating the GTX 670MX’s 5316 by 10.4%.

Q: Does the GTX 670MX have more raw compute power?

A: Yes. The GTX 670MX delivers 1,153.9 GFLOPS FP32, while the UHD Graphics 730 provides 499.2 GFLOPS. Its texture rate is 48.08 GTexel/s versus 15.60 GTexel/s.

Q: What memory does each GPU use?

A: The Intel UHD Graphics 730 uses System Shared memory with System Dependent bandwidth. The GTX 670MX uses 3 GB of GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth.

Q: Which GPU has a higher average benchmark score?

A: The Intel UHD Graphics 730 has an average benchmark score of 5929, compared to the GTX 670MX’s 5721. This places both at the 33rd percentile of all GPUs.

Q: How do their process nodes compare?

A: Intel uses a 14 nm+++ process, while NVIDIA uses TSMC’s 28 nm process. The Intel part is part of the Rocket Lake generation, and NVIDIA is Kepler.

Q: What are the shading unit counts?

A: The GTX 670MX has 960 shading units. The UHD Graphics 730 has 192 shading units.

Specification Differences

| Specification | Intel UHD Graphics 730 | NVIDIA GeForce GTX 670MX |

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

| Chip | Rocket Lake | GK104 |

| Architecture | Generation 12.1 | Kepler |

| Process Node | 14 nm+++ | 28 nm |

| Foundry | Intel | TSMC |

| Transistors | Not specified | 3,540 million |

| Die Size | Not specified | 294 mm² |

| Transistor Density | Not specified | 12.0M / mm² |

| Base Clock | 300 MHz | 601 MHz |

| Boost Clock | 1300 MHz | 601 MHz |

| Memory Size | System Shared | 3 GB |

| Memory Type | System Shared | GDDR5 |

| Memory Bus Width | System Shared | 192 bit |

| Memory Bandwidth | System Dependent | 67.20 GB/s |

| Shading Units | 192 | 960 |

| TMUs | 12 | 80 |

| ROPs | 8 | 24 |

| Pixel Rate | 10.40 GPixel/s | 12.02 GPixel/s |

| Texture Rate | 15.60 GTexel/s | 48.08 GTexel/s |

| FP32 | 499.2 GFLOPS | 1,153.9 GFLOPS |

| FP16 | 998.4 GFLOPS (2:1) | Not specified |

| TDP | 15 W | 75 W |

| Bus Interface | Ring Bus | PCIe 3.0 x16 |

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

| Vulkan | 1.4 | 1.2.175 |

| Release Date | 2021-03-29 | 2012-09-30 |

Head-to-Head Benchmarks

The Geekbench OpenCL test shows the GTX 670MX winning with a score of 6125 against Intel’s 5988, a delta of -2.2% for Intel. This is a modest victory, but it aligns with the NVIDIA part’s superior FP32 throughput and dedicated memory bandwidth. The 960 shading units and 80 TMUs provide a strong foundation for compute tasks that scale with shader count.

The Geekbench Vulkan test flips the result decisively. Intel’s UHD Graphics 730 scores 5870, while the GTX 670MX manages only 5316. That is a 10.4% advantage for Intel. This large margin highlights the architectural efficiency of Generation 12.1 in modern APIs. Despite having one-fifth the shading units, Intel’s newer design and Vulkan 1.4 driver support overcome the hardware deficit.

Looking at the average benchmark scores, Intel leads with 5929 versus 5721 for NVIDIA. This is a 3.6% overall advantage for Intel, driven primarily by the strong Vulkan showing. The GTX 670MX’s nearest rival is the Intel Iris Pro Graphics P6300 at 5712, just 0.2% behind, while Intel’s UHD Graphics 730 sits near the AMD Radeon HD 8730M at 5955, a 0.4% gap. Both GPUs are firmly in the same performance tier, with no clear overall winner.

The Verdict

Choose the Intel UHD Graphics 730 if you are building a system with Vulkan-based applications in mind. Its 10.4% Vulkan lead is significant, and the higher average benchmark score of 5929 suggests more consistent performance across modern workloads. The 15 W TDP is also a major practical advantage for low-power builds, and the 14 nm+++ process indicates a more recent design. The FP16 capability of 998.4 GFLOPS (2:1) is a bonus for mixed-precision tasks, a feature the GTX 670MX lacks entirely.

Choose the NVIDIA GeForce GTX 670MX if your workload is heavily OpenCL-bound or requires large dedicated memory. The 3 GB GDDR5 with 67.20 GB/s bandwidth is a massive advantage over System Shared memory, especially for texture-heavy rendering or datasets that exceed system RAM allocation. The 2.2% OpenCL win, combined with 1,153.9 GFLOPS FP32, makes it the stronger compute part for traditional GPGPU tasks. Its 75 W TDP is higher, but that is the cost of discrete performance.

The data shows a tie in wins, one each, but the nature of those wins matters. Intel wins the future-facing API by a large margin; NVIDIA wins the legacy compute API by a small one. For a new build, the Intel part’s Vulkan support and lower power draw are compelling. For an existing laptop with a GTX 670MX, there is no reason to switch. Both sit at the 33rd percentile, so neither will impress in demanding games. They are entry-level parts, and the choice comes down to software priorities rather than hardware superiority.