AMD Radeon Vega 10 Mobile vs NVIDIA GeForce GTX 670MX Comparison

Where Each One Wins

The recorded benchmark data splits cleanly between these two mobile graphics solutions. The AMD Radeon Vega 10 Mobile takes the single OpenCL head-to-head contest, posting 6476 points against the GTX 670MX's 6125, a 5.7% advantage. That win places the Vega IGP at the 37th percentile among all GPUs in the database, while the GTX 670MX sits at the 33rd percentile.

The GTX 670MX does hold one measurable edge: it has a Vulkan score of 5316 in the database, while the Vega 10 Mobile has no recorded Vulkan result. For users running Vulkan-based workloads, the NVIDIA part has verifiable performance data, whereas the AMD part simply lacks a measured score in that API.

Looking at the rival clusters, the Vega 10 Mobile's average benchmark score of 6476 positions it within 0.1% of the NVIDIA Quadro M5000M (6481) and 0.3% of the GeForce GT 555M (6493). It trails the GeForce GTX 670M (6513) by 0.6% but leads the RTX PRO 5000 72 GB Blackwell (6407) by 1.1%. The GTX 670MX, with an average score of 5721, sits 0.2% behind the Intel Iris Pro Graphics P6300 (5712) and 0.2% ahead of the GeForce GTX 550 Ti (5731), while also leading the Radeon HD 8790M (5691) by 0.5% and the Quadro M500M (5604) by 2.1%.

The practical takeaway from the data: the Vega 10 Mobile wins the compute-oriented OpenCL workload, while the GTX 670MX offers the only Vulkan measurement in this pairing. Neither part dominates across all APIs, which makes the choice dependent on which workload matters more.

Architecture Differences

These two GPUs come from different manufacturing generations and design philosophies. The AMD Radeon Vega 10 Mobile uses the Raven-M chip built on GCN 5.0 architecture, fabricated by GlobalFoundries on a 14 nm process. It packs 4,940 million transistors into a 210 mm² die, yielding a transistor density of 23.5 million per mm². The NVIDIA GeForce GTX 670MX uses the GK104 chip on Kepler architecture, built by TSMC on a 28 nm process, with 3,540 million transistors spread across a larger 294 mm² die, resulting in 12.0 million transistors per mm².

The Vega 10 Mobile is an integrated graphics processor (IGP) with a system-shared memory configuration. Its base clock runs at 300 MHz with a boost up to 1301 MHz. The GTX 670MX is a discrete mobile part with its own 3 GB of GDDR5 memory on a 192-bit bus, delivering 67.20 GB/s of bandwidth. Its clocks are fixed at 601 MHz for both base and boost, with memory running at 700 MHz (2.8 Gbps effective).

Core configuration differences are substantial. The Vega 10 Mobile fields 640 shading units, 40 texture mapping units, and 8 ROPs. The GTX 670MX counters with 960 shading units, 80 TMUs, and 24 ROPs. Despite having fewer shaders, the Vega 10 Mobile achieves higher FP32 throughput at 1.665 TFLOPS compared to the GTX 670MX's 1,153.9 GFLOPS, thanks to its higher boost clock. The Vega part also supports FP16 at 3.331 TFLOPS (2:1 ratio), a capability the GTX 670MX lacks entirely.

Pixel and texture rates tell a mixed story. The GTX 670MX leads in pixel fill at 12.02 GPixel/s versus 10.41 GPixel/s, but the Vega 10 Mobile leads in texture fill at 52.04 GTexel/s versus 48.08 GTexel/s. Power consumption differs by a factor of 7.5: the Vega IGP is rated at 10 W TDP, while the GTX 670MX draws 75 W. The Vega part connects via IGP bus with no power connectors, while the GTX 670MX uses PCIe 3.0 x16.

API support also diverges. The Vega 10 Mobile supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The GTX 670MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The release dates are far apart: the Vega part launched January 7, 2019, while the GTX 670MX came out September 30, 2012. Both are end-of-life products, with the Vega succeeding GCN 3.0 IGP and preceding Navi II IGP, while the GTX 670MX succeeds GeForce 500M and precedes GeForce 700M.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The AMD Radeon Vega 10 Mobile scores 6476 in Geekbench OpenCL, which is 5.7% higher than the NVIDIA GeForce GTX 670MX's 6125.

Q: Does either GPU support Vulkan?

A: Both list Vulkan support in their API specifications. The GTX 670MX has a recorded Geekbench Vulkan score of 5316, while the Vega 10 Mobile has no Vulkan benchmark entry in the database.

Q: How do the memory configurations differ?

A: The Vega 10 Mobile uses system-shared memory with system-dependent bandwidth, while the GTX 670MX has 3 GB of dedicated GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth.

Q: Which GPU has more shading units?

A: The GTX 670MX has 960 shading units, compared to 640 on the Vega 10 Mobile. Despite fewer units, the Vega part achieves higher FP32 throughput at 1.665 TFLOPS versus 1,153.9 GFLOPS.

Q: What are the power requirements?

A: The Vega 10 Mobile is rated at 10 W TDP with no power connectors, while the GTX 670MX is rated at 75 W TDP, also with no power connectors listed.

Q: How does the GTX 670MX compare to its nearest rivals?

A: Its average score of 5721 puts it 0.2% behind the Intel Iris Pro Graphics P6300, 0.2% ahead of the GeForce GTX 550 Ti, 0.5% ahead of the Radeon HD 8790M, and 2.1% ahead of the Quadro M500M.

Specification Differences

| Specification | AMD Radeon Vega 10 Mobile | NVIDIA GeForce GTX 670MX |

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

| Chip | Raven-M | GK104 |

| Architecture | GCN 5.0 | Kepler |

| Process Node | 14 nm | 28 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | 4,940 million | 3,540 million |

| Die Size | 210 mm² | 294 mm² |

| Transistor Density | 23.5M / mm² | 12.0M / mm² |

| Base Clock | 300 MHz | 601 MHz |

| Boost Clock | 1301 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 | 640 | 960 |

| TMUs | 40 | 80 |

| ROPs | 8 | 24 |

| Pixel Rate | 10.41 GPixel/s | 12.02 GPixel/s |

| Texture Rate | 52.04 GTexel/s | 48.08 GTexel/s |

| FP32 | 1.665 TFLOPS | 1,153.9 GFLOPS |

| FP16 | 3.331 TFLOPS (2:1) | null |

| TDP | 10 W | 75 W |

| Bus Interface | IGP | PCIe 3.0 x16 |

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

| Vulkan | 1.3 | 1.2.175 |

| Release Date | 2019-01-07 | 2012-09-30 |

| Predecessor | GCN 3.0 IGP | GeForce 500M |

| Successor | Navi II IGP | GeForce 700M |

The table shows two designs with opposite priorities. The Vega 10 Mobile uses a denser, newer process with a much smaller power envelope, while the GTX 670MX uses a larger, older die with more memory and wider bus. The Vega part leads in FP32 compute and texture rate; the GTX 670MX leads in pixel rate and memory bandwidth.

Head-to-Head Benchmarks

The database records one direct comparison between these GPUs: Geekbench OpenCL. The AMD Radeon Vega 10 Mobile wins with 6476 points against the NVIDIA GeForce GTX 670MX's 6125, a 5.7% margin. This result aligns with the FP32 throughput gap, where the Vega part delivers 1.665 TFLOPS versus 1,153.9 GFLOPS for the NVIDIA part.

The GTX 670MX's only other benchmark, Geekbench Vulkan at 5316, has no direct counterpart for the Vega 10 Mobile, so a head-to-head comparison in that API cannot be made from the recorded data.

Context from the nearest rival lists reinforces the separation. The Vega 10 Mobile's OpenCL score of 6476 sits within 1% of several NVIDIA parts: 0.1% below the Quadro M5000M, 0.3% below the GeForce GT 555M, and 0.6% below the GeForce GTX 670M, while running 1.1% above the RTX PRO 5000 72 GB Blackwell. The GTX 670MX's average of 5721 clusters around lower-tier parts, with a 2.1% lead over the Quadro M500M being its largest margin among nearest rivals.

The percentile standings summarize the overall position: the Vega 10 Mobile ranks at the 37th percentile across all GPUs, four points above the GTX 670MX at the 33rd percentile. The win count in head-to-head benchmarks favors AMD by 1 to 0, driven entirely by the OpenCL result. For anyone comparing these two on raw compute performance, the data points clearly to the Vega 10 Mobile. For Vulkan-specific workloads, the GTX 670MX at least offers a measured baseline, even if no direct comparison exists.