AMD Radeon 660M vs AMD Radeon Pro 5600M Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Pro 5600M posts an average benchmark score of 16,351, while the AMD Radeon 660M records 13,812. The Pro 5600M sits at the 59th percentile of all GPUs, versus the 660M's 55th percentile.

Q: How do the two GPUs compare in Geekbench OpenCL performance?

A: The Radeon Pro 5600M scores 47,783 in Geekbench OpenCL, which is 271.1% higher than the Radeon 660M's 12,876. This is the largest performance delta in the head-to-head dataset.

Q: Which GPU supports hardware ray tracing?

A: The AMD Radeon 660M includes 6 ray tracing cores, while the Radeon Pro 5600M has no ray tracing cores listed. The 660M also supports DirectX 12 Ultimate (12_2), whereas the Pro 5600M is limited to DirectX 12 (12_1).

Q: What is the difference in memory configuration?

A: The Radeon Pro 5600M has 8 GB of HBM2 memory on a 2048-bit bus with 394.2 GB/s of bandwidth. The Radeon 660M uses System Shared memory, with its bandwidth listed as System Dependent.

Q: Which GPU has a higher boost clock?

A: The Radeon 660M boosts to 1900 MHz, compared to the Radeon Pro 5600M's 1144 MHz boost. However, the Pro 5600M has far more shading units: 2560 versus 384.

Q: What are the production statuses of these GPUs?

A: Both are marked End-of-life. The Radeon Pro 5600M was released in June 2020, while the Radeon 660M followed in January 2022 and lists the Vega II IGP as its predecessor.

Architecture Differences

The AMD Radeon Pro 5600M is built on the Navi 12 chip using the RDNA 1.0 architecture, fabricated on a 7 nm TSMC process. The AMD Radeon 660M, in contrast, uses the Rembrandt chip with RDNA 2.0 on a 6 nm TSMC process. This node shrink is accompanied by a substantial difference in transistor density: the 660M's die measures 208 mm² and packs 13,100 million transistors for a density of 63.0M per mm², while the Pro 5600M's transistor count and die size are not recorded in the database.

The shader configuration differs dramatically. The Pro 5600M carries 2560 shading units, 160 texture mapping units, and 64 raster operation pipelines. The 660M has 384 shading units, 24 TMUs, and 16 ROPs. That is a 6.7x advantage in shading units for the Pro 5600M, and a 4x advantage in both TMUs and ROPs. Ray tracing capability belongs exclusively to the 660M, which has 6 ray tracing cores; the Pro 5600M has none listed.

Memory architecture is another major split. The Pro 5600M uses dedicated HBM2 memory: 8 GB on a 2048-bit bus, delivering 394.2 GB/s of bandwidth. The 660M relies on System Shared memory, with bus width and type also marked System Shared, and bandwidth listed as System Dependent. This means the 660M's memory performance is contingent on the host system, whereas the Pro 5600M has a fixed, dedicated pool.

Clock behavior favors the 660M on raw frequency. Its base clock is 1500 MHz and boost reaches 1900 MHz. The Pro 5600M runs at an 822 MHz base and 1144 MHz boost. Despite the lower clocks, the Pro 5600M achieves far higher throughput metrics: 73.22 GPixel/s pixel rate and 183.0 GTexel/s texture rate, versus 30.40 GPixel/s and 45.60 GTexel/s for the 660M. FP32 compute is 5.857 TFLOPS for the Pro 5600M versus 1,459.2 GFLOPS (1.459 TFLOPS) for the 660M. FP16 figures are 11.71 TFLOPS (2:1) and 2.918 TFLOPS (2:1), respectively.

The bus interface also differs: the Pro 5600M uses PCIe 4.0 x16, while the 660M uses PCIe 4.0 x8. API support leans toward the newer GPU: the 660M offers DirectX 12 Ultimate (12_2) and Vulkan 1.4, while the Pro 5600M is capped at DirectX 12 (12_1) and Vulkan 1.3. Both support OpenGL 4.6. Power draw is close, with the Pro 5600M rated at 50 W and the 660M at 40 W, both as IGP slot-width parts with no power connectors.

The Verdict

The data points to a clear performance hierarchy. The Radeon Pro 5600M wins both shared head-to-head benchmarks: Geekbench OpenCL by 271.1% and Geekbench Vulkan by 214.8%. Its average benchmark score of 16,351 places it in the 59th percentile, while the 660M's 13,812 sits at the 55th percentile. The Pro 5600M is also within 0.1% of the AMD Radeon RX 5700 XT's average score of 16,361, and within 0.9% of the NVIDIA GeForce RTX 5090 D V2's 16,504. The 660M, by comparison, trails the NVIDIA RTX A2000 Mobile by 0.1% (13,821 average, essentially identical) and sits 0.5% ahead of the AMD Radeon RX 7900 XT's 13,745.

Who should choose which GPU? For workloads that favor raw compute, high memory bandwidth, and a wide 2048-bit memory bus, the Radeon Pro 5600M is the stronger part. Its 5.857 TFLOPS FP32 throughput and 394.2 GB/s of dedicated bandwidth give it a decisive edge in memory-bound and shader-heavy tasks. The 660M, on the other hand, is the only one of the two with ray tracing cores, and its RDNA 2.0 architecture with DirectX 12 Ultimate support makes it the more feature-forward option for applications that leverage newer graphics APIs. The 660M also runs at a lower 40 W TDP, making it the more power-efficient choice on paper, though both are integrated parts with no external power connectors.

Specification Differences

| Specification | AMD Radeon Pro 5600M | AMD Radeon 660M |

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

| Chip | Navi 12 | Rembrandt |

| Architecture | RDNA 1.0 | RDNA 2.0 |

| Process Node | 7 nm | 6 nm |

| Transistors | Not listed | 13,100 million |

| Die Size | Not listed | 208 mm² |

| Transistor Density | Not listed | 63.0M / mm² |

| Base Clock | 822 MHz | 1500 MHz |

| Boost Clock | 1144 MHz | 1900 MHz |

| Memory Size | 8 GB | System Shared |

| Memory Type | HBM2 | System Shared |

| Memory Bus Width | 2048 bit | System Shared |

| Memory Bandwidth | 394.2 GB/s | System Dependent |

| Shading Units | 2560 | 384 |

| TMUs | 160 | 24 |

| ROPs | 64 | 16 |

| Ray Tracing Cores | None | 6 |

| Pixel Rate | 73.22 GPixel/s | 30.40 GPixel/s |

| Texture Rate | 183.0 GTexel/s | 45.60 GTexel/s |

| FP32 | 5.857 TFLOPS | 1,459.2 GFLOPS |

| FP16 | 11.71 TFLOPS (2:1) | 2.918 TFLOPS (2:1) |

| TDP | 50 W | 40 W |

| Bus Interface | PCIe 4.0 x16 | PCIe 4.0 x8 |

| DirectX | 12 (12_1) | 12 Ultimate (12_2) |

| Vulkan | 1.3 | 1.4 |

| Release Date | 2020-06-14 | 2022-01-03 |

| Predecessor | Not listed | Vega II IGP |

| Successor | Not listed | Navi III IGP |

Head-to-Head Benchmarks

The database includes two direct comparison benchmarks between these GPUs, and the Radeon Pro 5600M wins both by a wide margin. In Geekbench OpenCL, the Pro 5600M scores 47,783 against the 660M's 12,876, a delta of 271.1%. This is the largest recorded gap in the head-to-head set. The Vulkan test is similarly one-sided: 46,425 for the Pro 5600M versus 14,748 for the 660M, a 214.8% advantage.

These results align with the broader benchmark picture. The Pro 5600M's average score of 16,351 is 18.4% higher than the 660M's 13,812, and its percentile ranking (59th versus 55th) reflects a meaningful gap in the overall distribution. The Pro 5600M's nearest rivals, the RX 5700 XT and RTX PRO 6000 Blackwell, sit just 0.1% and 0.3% above it, respectively, indicating that the Pro 5600M is positioned in a competitive mid-to-high tier. The 660M's closest rival, the RTX A2000 Mobile, is only 0.1% ahead, so the 660M is effectively tied with that mobile workstation part.

Looking at the Pro 5600M's full benchmark suite, its strongest results come in compute-oriented tests: Passmark GPU Compute at 4,031, and Geekbench Metal at 55,030. Its Passmark G3D score is 9,279, while legacy DirectX tests are lower (Passmark DirectX 9: 118, DirectX 10: 62, DirectX 11: 59, DirectX 12: 39). The 660M has only two recorded benchmarks, both in Geekbench, and neither approaches the Pro 5600M's numbers. The Vulkan gap of 214.8% is particularly notable because Vulkan is a cross-platform, low-overhead API where both GPUs should be able to express their respective architectures; the sheer scale of the difference underscores the Pro 5600M's compute and memory bandwidth advantages.

The wins tally is 2-0 in favor of the Radeon Pro 5600M. On paper, the 660M's higher clocks and newer architecture do not translate into benchmark victories. The Pro 5600M's 2048-bit HBM2 memory subsystem and 2560 shading units are simply in a different class. For users prioritizing raw compute throughput and memory bandwidth, the Pro 5600M is the clear choice. For those who need ray tracing, the 660M is the only option between the two, and its lower TDP and newer API support may matter in thin-and-light systems. But based strictly on the recorded benchmark data, the Radeon Pro 5600M is the faster GPU in every direct comparison available.