GPU Comparison

The AMD Radeon 660M and AMD Radeon Pro 555 target very different corners of the mobile GPU market, separated by five years of architecture evolution. The data shows a clear generational shift: the Radeon 660M wins both head-to-head compute benchmarks decisively, yet the Radeon Pro 555 still holds its own in raw shading throughput and has the advantage of dedicated video memory. These two GPUs represent the trade-off between modern integrated graphics with shared system memory and older discrete-class designs with their own VRAM.

Head-to-Head Benchmarks

The head-to-head comparison consists of two cross-platform compute tests, and the Radeon 660M takes both. In Geekbench OpenCL, the 660M scores 12,876 against the Pro 555's 11,682, a 10.2% advantage. The gap widens substantially in Vulkan, where the 660M posts 14,748 versus 12,303, a 19.9% lead. These are not marginal wins; the Vulkan result in particular shows the newer architecture extracting far more performance from the same class of workload.

The average benchmark scores tell a slightly different story. The 660M averages 13,812 across its two tests, while the Pro 555 averages 13,407 across three tests. That puts the 660M 3.0% ahead on average, but the Pro 555's extra Metal benchmark (16,236) is its strongest single result and pulls its average up. Notably, the Pro 555's OpenCL score is its weakest result, while its Vulkan score sits 5.0% above its OpenCL number. The 660M, by contrast, scores 14.5% higher in Vulkan than in OpenCL, suggesting its driver stack and architecture favor the newer API more strongly.

Looking at the percentile rankings, the two GPUs are nearly identical in overall standing. The 660M sits at the 55th percentile of all GPUs, while the Pro 555 sits at the 54th. Their nearest rivals bracket them closely: the 660M's nearest neighbor is the NVIDIA RTX A2000 Mobile at 13,821 average (only 0.1% higher), and the Pro 555's nearest is the AMD FirePro M6100 at 13,354 (0.4% lower). The data shows both GPUs occupying a performance tier where single-digit percentage swings separate them from competitors.

Architecture Differences

The architectural divide here is fundamental. The Radeon 660M uses RDNA 2.0 on a 6 nm TSMC process, built on the Rembrandt chip with 13,100 million transistors on a 208 mm² die. The Radeon Pro 555 uses GCN 4.0 on a 14 nm GlobalFoundries process, built on Polaris 21 with only 3,000 million transistors on a 123 mm² die. That is a massive transistor count difference, the 660M packs over four times as many transistors into a die that is 69% larger by area. The transistor density reflects this: 63.0 million transistors per mm² on the 660M versus 24.4 million per mm² on the Pro 555.

The core configurations differ in ways that matter. The Pro 555 has 768 shading units and 48 texture mapping units, exactly double the 660M's 384 shading units and 24 TMUs. Both have 16 ROPs. The 660M compensates with higher clocks, 1500 MHz base and 1900 MHz boost, while the Pro 555 lists no base or boost clock at all, only a memory clock of 1275 MHz (5.1 Gbps effective). The 660M's pixel rate of 30.40 GPixel/s is 123% higher than the Pro 555's 13.60 GPixel/s, and its texture rate of 45.60 GTexel/s is 11.8% higher than 40.80 GTexel/s despite having half the TMUs.

Memory is where the Pro 555 fights back. It has 2 GB of dedicated GDDR5 on a 128-bit bus with 81.60 GB/s of bandwidth. The 660M uses system shared memory with a system-dependent bandwidth figure. This is the classic integrated-versus-discrete trade-off: the 660M never runs out of memory capacity (it can draw on whatever the system has), but it must share bandwidth with the CPU. The Pro 555 has a fixed, dedicated pool that is small by modern standards but always available.

The feature sets also diverge. The 660M supports DirectX 12 Ultimate (12_2), Vulkan 1.4, and has 6 ray tracing cores. The Pro 555 tops out at DirectX 12 (12_0), Vulkan 1.3, and has no ray tracing cores at all. Both support OpenGL 4.6. The 660M uses PCIe 4.0 x8, while the Pro 555 uses PCIe 3.0 x8. Power draw differs significantly: the 660M is rated at 40 W TDP, the Pro 555 at 75 W.

Where Each One Wins

The Radeon 660M wins in every benchmark where they were directly compared. Its OpenCL advantage of 10.2% and Vulkan advantage of 19.9% show that for general-purpose compute and modern API workloads, the newer architecture is simply faster. Its higher pixel rate (30.40 GPixel/s versus 13.60 GPixel/s) suggests it will handle fill-rate-bound scenarios better. The 660M also has the ray tracing cores, which the Pro 555 lacks entirely, any workload that leverages RT acceleration will only run on the 660M.

The Radeon Pro 555 wins in the areas not covered by direct comparison. Its dedicated 2 GB GDDR5 memory with 81.60 GB/s bandwidth provides consistent, predictable memory performance that system-shared memory cannot guarantee. Its 768 shading units and 48 TMUs give it raw compute and texture-processing capacity that the 660M's 384 shaders and 24 TMUs cannot match, even if the 660M's higher clocks close the gap in practice. The Pro 555's Metal score of 16,236 is its best result and indicates strong performance in Apple's ecosystem, which is relevant given its "Radeon Pro Mac" generation label.

The average benchmark data shows the 660M ahead overall (13,812 versus 13,407), but the Pro 555's single strongest benchmark (Metal at 16,236) exceeds any single score the 660M achieved. This suggests the Pro 555 retains an edge in specific API environments, even though it loses in the two APIs where both were tested.

FAQ

Q: Which GPU is faster in compute workloads?

A: The Radeon 660M wins both head-to-head compute tests: 10.2% faster in Geekbench OpenCL (12,876 versus 11,682) and 19.9% faster in Geekbench Vulkan (14,748 versus 12,303).

Q: Does the Radeon Pro 555 have any performance advantage?

A: Yes. It has double the shading units (768 versus 384) and TMUs (48 versus 24), plus dedicated 2 GB GDDR5 memory with 81.60 GB/s bandwidth. Its Geekbench Metal score of 16,236 is higher than any score the 660M achieved.

Q: How do their average benchmark scores compare?

A: The 660M averages 13,812 across two tests, while the Pro 555 averages 13,407 across three tests. The 660M is 3.0% higher on average.

Q: What is the architectural generation gap?

A: The 660M uses RDNA 2.0 on a 6 nm TSMC process (13,100 million transistors, 208 mm² die). The Pro 555 uses GCN 4.0 on a 14 nm GlobalFoundries process (3,000 million transistors, 123 mm² die).

Q: Does the 660M support ray tracing?

A: Yes, it has 6 ray tracing cores and supports DirectX 12 Ultimate (12_2). The Pro 555 has no ray tracing cores and only supports DirectX 12 (12_0).

Q: What are the power requirements?

A: The 660M has a 40 W TDP, while the Pro 555 has a 75 W TDP. Both are integrated form factors with no power connectors.

The Verdict

The data points to a clear recommendation for most users: the Radeon 660M is the better GPU for compute and modern API workloads. It wins both head-to-head benchmarks, has a higher average score (13,812 versus 13,407), draws less power (40 W versus 75 W), and supports newer features like ray tracing and DirectX 12 Ultimate. Its 19.9% Vulkan lead is particularly significant, as Vulkan is increasingly the standard for cross-platform graphics.

The Radeon Pro 555 is the choice only for specific use cases. Its 2 GB of dedicated GDDR5 memory provides bandwidth that system-shared memory cannot guarantee, which matters for workloads sensitive to memory latency or contention. Its 768 shading units and 48 TMUs give it theoretical peak throughput that the 660M's smaller core count cannot match, even if the 660M's clocks and architecture achieve better real-world results. And its Metal score of 16,236 suggests it remains competitive in Apple-centric environments, which aligns with its "Radeon Pro Mac" designation.

For general-purpose mobile computing, gaming, or content creation, the 660M is the stronger choice based on this data. For legacy software, Metal-specific workflows, or scenarios where dedicated VRAM is non-negotiable, the Pro 555 remains relevant despite its age. The 660M's end-of-life status and the Pro 555's end-of-life status mean neither is a future-proof investment, but the 660M's architecture is five years newer and the benchmark data reflects that.

Specification Differences

| Specification | AMD Radeon 660M | AMD Radeon Pro 555 |

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

| Architecture | RDNA 2.0 | GCN 4.0 |

| Process Node | 6 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 13,100 million | 3,000 million |

| Die Size | 208 mm² | 123 mm² |

| Transistor Density | 63.0M / mm² | 24.4M / mm² |

| Base Clock | 1500 MHz | Not specified |

| Boost Clock | 1900 MHz | Not specified |

| Memory Size | System Shared | 2 GB |

| Memory Type | System Shared | GDDR5 |

| Memory Bus Width | System Shared | 128 bit |

| Memory Bandwidth | System Dependent | 81.60 GB/s |

| Shading Units | 384 | 768 |

| TMUs | 24 | 48 |

| ROPs | 16 | 16 |

| RT Cores | 6 | None |

| Pixel Rate | 30.40 GPixel/s | 13.60 GPixel/s |

| Texture Rate | 45.60 GTexel/s | 40.80 GTexel/s |

| FP32 Performance | 1,459.2 GFLOPS | 1,305.6 GFLOPS |

| FP16 Performance | 2.918 TFLOPS (2:1) | 1,305.6 GFLOPS (1:1) |

| TDP | 40 W | 75 W |

| Bus Interface | PCIe 4.0 x8 | PCIe 3.0 x8 |

| DirectX Support | 12 Ultimate (12_2) | 12 (12_0) |

| Vulkan Support | 1.4 | 1.3 |

| Release Date | 2022-01-03 | 2017-06-04 |