AMD Radeon Pro 555 vs NVIDIA GeForce GTX 670 Comparison

AMD Radeon Pro 555 and NVIDIA GeForce GTX 670 are two end-of-life graphics cards from different eras, built for different purposes. The data shows a clear split: the AMD card dominates in one API, while the NVIDIA card wins in two others, with the overall average benchmark scores landing within 5% of each other. The Radeon Pro 555 sits at the 54th percentile of all GPUs with an average benchmark score of 13,407, while the GTX 670 sits at the 52nd percentile with 12,773, making them close competitors in aggregate but with radically different performance profiles depending on the workload.

Head-to-Head Benchmarks

The most striking result in the head-to-head comparison is the Geekbench Metal test, where the AMD Radeon Pro 555 delivers a devastating blow. The AMD card scores 16,236, while the NVIDIA GeForce GTX 670 manages only 7,424. That is a delta of 118.7% in favor of the Radeon Pro 555 — more than double the performance. This is a categorical victory, not a marginal one. The GTX 670, a card from 2012, clearly lacks the modern API optimizations and hardware features that make Metal workloads fly on the 2017-era Radeon Pro 555. For any application that relies heavily on Metal — think macOS ecosystem software and certain creative suites — the AMD card is not just better; it is in a different league entirely.

However, the story flips completely when looking at the other two benchmark results. In Geekbench OpenCL, the NVIDIA GeForce GTX 670 scores 15,341, beating the Radeon Pro 555's 11,682 by 23.9%. The margin here is substantial, though not as extreme as the Metal result. The GTX 670's advantage likely stems from its much larger raw compute configuration, which we will explore in the architecture section. Similarly, in Geekbench Vulkan, the GTX 670 wins again, scoring 15,553 versus 12,303 for the Radeon Pro 555 — a delta of 20.9%. The NVIDIA card's Vulkan support is version 1.2.175, while the AMD card supports Vulkan 1.3, but despite the older API version, the GTX 670's sheer hardware muscle prevails.

Looking at the overall tally, the NVIDIA card wins two out of three benchmarks. The wins are consistent in magnitude: roughly 21% to 24% ahead in OpenCL and Vulkan. The AMD card's single win, however, is of a much larger magnitude at 118.7%. This asymmetry is crucial. If your workload is Metal-centric, the AMD card is the obvious choice. If your workload spans OpenCL or Vulkan, the GTX 670 is clearly superior. The average benchmark scores reflect this tension: the Radeon Pro 555 averages 13,407, while the GTX 670 averages 12,773. The AMD card is 5% ahead on average, buoyed by its massive Metal advantage, but the NVIDIA card wins the two more widely used cross-platform APIs.

Architecture Differences

The architectural gap between these two cards is generational. The AMD Radeon Pro 555 uses the Polaris 21 chip, built on GCN 4.0 architecture, fabricated on a 14 nm process at GlobalFoundries. The NVIDIA GeForce GTX 670 uses the GK104 chip, built on the Kepler architecture, fabricated on a 28 nm process at TSMC. The process node difference is stark: 14 nm versus 28 nm means the AMD chip packs transistors far more densely, at 24.4 million per square millimeter versus 12.0 million for the NVIDIA chip.

The transistor counts tell an interesting story. The GTX 670 has 3,540 million transistors on a 294 mm² die, while the Radeon Pro 555 has 3,000 million transistors on a much smaller 123 mm² die. The AMD chip achieves nearly the same transistor count in less than half the die area. This density advantage is a direct result of the newer manufacturing process. The Radeon Pro 555 is a 75 W card with an IGP slot width and no power connectors, designed for mobile integration. The GTX 670 is a 170 W dual-slot card requiring 2x 6-pin power connectors and a 450 W suggested PSU, designed as a desktop add-in board.

The memory configurations differ significantly in bus width and bandwidth. Both cards have 2 GB of GDDR5 memory, but the GTX 670 uses a 256-bit bus delivering 192.3 GB/s of bandwidth, while the Radeon Pro 555 uses a 128-bit bus delivering only 81.60 GB/s. The NVIDIA card's memory clock is 1502 MHz (6 Gbps effective), while the AMD card's memory clock is 1275 MHz (5.1 Gbps effective). The GTX 670's much wider memory bus gives it a 2.36x bandwidth advantage, which likely contributes to its OpenCL and Vulkan wins.

Compute resources also favor the GTX 670 heavily. The NVIDIA card has 1,344 shading units, 112 texture mapping units, and 32 ROPs. The Radeon Pro 555 has 768 shading units, 48 TMUs, and only 16 ROPs. The GTX 670 more than doubles the ROP count and nearly doubles the TMU count. Pixel rate is 27.44 GPixel/s for the GTX 670 versus 13.60 GPixel/s for the Radeon Pro 555. Texture rate is 109.8 GTexel/s versus 40.80 GTexel/s. FP32 performance is 2.634 TFLOPS for the GTX 670 versus 1,305.6 GFLOPS for the Radeon Pro 555. The GTX 670 also has a base clock of 915 MHz and boost clock of 980 MHz, while the Radeon Pro 555 does not list base or boost clocks.

FAQ

Q: Which card has better Metal performance?

A: The AMD Radeon Pro 555 wins Geekbench Metal overwhelmingly, scoring 16,236 versus 7,424 for the GTX 670, a 118.7% advantage. This is the single largest performance gap in any benchmark between these two cards.

Q: Which card is faster in OpenCL?

A: The NVIDIA GeForce GTX 670 wins Geekbench OpenCL with a score of 15,341, beating the Radeon Pro 555's 11,682 by 23.9%. The NVIDIA card's higher shading unit count and memory bandwidth likely drive this result.

Q: How do the cards compare in Vulkan?

A: The NVIDIA GeForce GTX 670 leads in Geekbench Vulkan, scoring 15,553 versus 12,303 for the Radeon Pro 555, a 20.9% delta. Despite the AMD card supporting Vulkan 1.3 versus NVIDIA's 1.2.175, the GTX 670's raw compute power wins out.

Q: What are the power requirements for each card?

A: The Radeon Pro 555 is a 75 W card with an IGP slot width and no power connectors. The GTX 670 is a 170 W dual-slot card requiring 2x 6-pin power connectors and a 450 W suggested PSU.

Q: Why is the average benchmark score so close if the per-test results are so lopsided?

A: The Radeon Pro 555 averages 13,407, while the GTX 670 averages 12,773 — a 5% difference. The AMD card's massive 118.7% win in Metal offsets the NVIDIA card's smaller 23.9% and 20.9% wins in OpenCL and Vulkan, bringing the averages close together.

Q: Which card has more memory bandwidth?

A: The GTX 670 has 192.3 GB/s of bandwidth on a 256-bit bus, compared to the Radeon Pro 555's 81.60 GB/s on a 128-bit bus. This is a 2.36x advantage for the NVIDIA card.

The Verdict

The data paints a clear picture for different use cases. If your primary workloads are Metal-based — which is typical for macOS applications, video editing, and certain creative software — the AMD Radeon Pro 555 is the definitive choice. Its 118.7% advantage in Geekbench Metal is not a marginal edge; it is a complete rout. The card's modern GCN 4.0 architecture and 14 nm process deliver efficiency that the older Kepler design cannot match in this API.

Conversely, the NVIDIA GeForce GTX 670 is the better pick for OpenCL and Vulkan workloads. With a 23.9% lead in OpenCL and a 20.9% lead in Vulkan, the GTX 670's superior shading unit count (1,344 versus 768), wider memory bus (256-bit versus 128-bit), and higher memory bandwidth (192.3 GB/s versus 81.60 GB/s) translate directly into performance in these cross-platform APIs. The GTX 670's 2.634 TFLOPS FP32 performance is nearly double the Radeon Pro 555's 1,305.6 GFLOPS, which explains its dominance in compute-heavy tasks.

The form factor distinction is also decisive. The Radeon Pro 555 is an IGP solution with no power connectors, designed for laptops and integrated systems. The GTX 670 is a dual-slot desktop card measuring 241 mm in length, requiring external power. If you need a mobile or low-power solution, the Radeon Pro 555 is the only option. If you have a desktop with a 450 W PSU and space for a full-length card, the GTX 670 offers better raw performance in most APIs.

In terms of API support, the Radeon Pro 555 supports DirectX 12 (12_0) and Vulkan 1.3, while the GTX 670 supports DirectX 12 (11_0) and Vulkan 1.2.175. The AMD card's more modern feature set may offer better future-proofing for newer titles, but the GTX 670's existing performance lead in OpenCL and Vulkan is substantial. Ultimately, the choice hinges on your software ecosystem: Metal users should choose AMD, while OpenCL/Vulkan users should choose NVIDIA.

Specification Differences

| Specification | AMD Radeon Pro 555 | NVIDIA GeForce GTX 670 |

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

| Architecture | GCN 4.0 | Kepler |

| Process Node | 14 nm | 28 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | 3,000 million | 3,540 million |

| Die Size | 123 mm² | 294 mm² |

| Transistor Density | 24.4M / mm² | 12.0M / mm² |

| Base Clock | Not listed | 915 MHz |

| Boost Clock | Not listed | 980 MHz |

| Memory Clock | 1275 MHz (5.1 Gbps effective) | 1502 MHz (6 Gbps effective) |

| Memory Bus Width | 128 bit | 256 bit |

| Memory Bandwidth | 81.60 GB/s | 192.3 GB/s |

| Shading Units | 768 | 1344 |

| TMUs | 48 | 112 |

| ROPs | 16 | 32 |

| Pixel Rate | 13.60 GPixel/s | 27.44 GPixel/s |

| Texture Rate | 40.80 GTexel/s | 109.8 GTexel/s |

| FP32 Performance | 1,305.6 GFLOPS | 2.634 TFLOPS |

| TDP | 75 W | 170 W |

| Slot Width | IGP | Dual-slot |

| Power Connectors | None | 2x 6-pin |

| Suggested PSU | Not listed | 450 W |

| Bus Interface | PCIe 3.0 x8 | PCIe 3.0 x16 |

| Display Outputs | Portable Device Dependent | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 |

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

| Vulkan Support | 1.3 | 1.2.175 |

| Dimensions | Not listed | 241 mm x 111 mm x 38 mm |

| Release Date | 2017-06-04 | 2012-05-09 |