AMD Radeon Pro Vega 20 vs AMD Radeon RX 6600M Comparison

# AMD Radeon Pro Vega 20 vs AMD Radeon RX 6600M

Head-to-Head Benchmarks

The database places the Radeon Pro Vega 20 against the Radeon RX 6600M across three cross-API workloads. The RX 6600M wins all three recorded head-to-head matchups, though the magnitude of the win shifts depending on the test.

In Geekbench Metal, the RX 6600M scores 92,237 against the Pro Vega 20's 30,427. That is a 67% advantage for the RX 6600M, a dominant margin. The OpenCL result tells a similar story: 67,765 for the RX 6600M versus 26,679 for the Pro Vega 20, a 60.6% lead. Vulkan narrows the gap slightly in percentage terms but not in outcome: 73,740 versus 26,410, a 64.2% win for the RX 6600M.

The average benchmark score in the database tells the same tale. The RX 6600M averages 23,273 across all recorded tests, while the Pro Vega 20 averages 27,839. That puts the Pro Vega 20 just 0.2% ahead of the RX 6600M in raw average, but the head-to-head tests show a different picture: the RX 6600M is the faster part in every single one of the recorded matchups.

Architecture Differences

These two AMD parts come from different design generations and different foundries. The Radeon Pro Vega 20 is built on the GCN 5.0 architecture, using the Vega 12 chip, and was produced at GlobalFoundries on a 14 nm node. It is a Radeon Pro Mac generation part, specifically the Radeon Pro Mac (Vega Series). The Radeon RX 6600M, in contrast, uses the Navi 23 chip on the RDNA 2.0 architecture, made by TSMC on a 7 nm process. That process difference is considerable, the RX 6600M being built on a much smaller node.

The Pro Vega 20 packs 4 GB of HBM2 memory across a 1024 bit bus, which gives it 189.4 GB/s of bandwidth. The RX 6600M instead uses 8 GB of GDDR6 on a 128 bit bus, yielding 224.0 GB/s. The newer chip also has a transistor count of 11,060 million on a 237 mm² die, for 46.7M transistors per square mm. The die size of the Vega 20 is not in the database.

Shader and pixel throughput also favor the RX 6600M. It has 1792 shading units and 64 ROPs, against 1280 shading units and 32 ROPs for the Pro Vega 20. The RX 6600M also carries 28 ray tracing cores, while the Pro Vega 20 has none. Texture and pixel fill rates: the RX 6600M reaches 270.6 GTexel/s and 154.6 GPixel/s, while the Pro Vega 20 is at 102.6 GTexel/s and 41.06 GPixel/s. Floating point compute shifts as well: 8.659 TFLOPS FP32 and 17.32 TFLOPS FP16 (2:1) for the RX 6600M, versus 3.284 TFLOPS FP32 and 6.569 TFLOPS FP16 for the Vega.

The Radeon Pro Vega 20 is a PCIe 3.0 x16 part with a 100 W TDP. The RX 6600M, by contrast, uses a PCIe 4.0 x8 interface and also carries a 100 W TDP.

Where Each One Wins

Given the head-to-head sweep, the RX 6600M is the clear performance pick in every recorded benchmark. Its wins range from 60.6% to 67% depending on the API, with the closest result being the Vulkan test at 64.2% ahead.

Where the Pro Vega 20 still holds some ground is not in speed but in scope. The Pro Vega 20 belongs to a different product generation (Radeon Pro Mac) and is the predecessor of the Navi Mobile line, which itself is end-of-life. The RX 6600M is from the current Navi Mobile generation. The database lists the Pro Vega 20 as having no direct successor, while the RX 6600M lists the Polaris Mobile as its predecessor. The Pro Vega 20 also runs a GCN 5.0 part, the RX 6600M is RDNA 2.0.

In terms of API support, the RX 6600M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The Pro Vega 20 also supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. Both parts are marked End-of-life in the database.

Specification Differences

| | AMD Radeon Pro Vega 20 | AMD Radeon RX 6600M |

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

| Architecture | GCN 5.0 | RDNA 2.0 |

| Chip | Vega 12 | Navi 23 |

| Process | 14 nm | 7 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | Not listed | 11,060 million |

| Die size | Not listed | 237 mm² |

| Memory | 4 GB HBM2 | 8 GB GDDR6 |

| Memory bus | 1024 bit | 128 bit |

| Bandwidth | 189.4 GB/s | 224.0 GB/s |

| Shading units | 1280 | 1792 |

| TMUs | 80 | 112 |

| ROPs | 32 | 64 |

| Ray cores | 0 | 28 |

| Pixel rate | 41.06 GPixel/s | 154.6 GPixel/s |

| Texture rate | 102.6 GTexel/s | 270.6 GTexel/s |

| FP32 | 3.284 TFLOPS | 8.659 TFLOPS |

| FP16 | 6.569 TFLOPS (2:1) | 17.32 TFLOPS (2:1) |

| Bus | PCIe 3.0 x16 | PCIe 4.0 x8 |

| Interface | 100 W | 100 W |

| API | DX 12 (12_1), OGL 4.6, VK 1.3 | DX 12 Ultimate (12_2), OGL 4.6, VK 1.4 |

FAQ

Q: Which part has higher memory bandwidth?

The RX 6600M has a higher memory bandwidth at 224.0 GB/s, versus 189.4 GB/s for the Pro Vega 20.

Q: Does one part have ray tracing?

Only the RX 6600M has 28 ray cores. The Pro Vega 20 has no ray cores.

Q: Which API levels do these parts support?

The RX 6600M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The Pro Vega 20 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3.

Q: Which part is older?

The Pro Vega 20 was released in November 2018. The RX 6600M was released in May 2021.

Q: What is the TDP of each?

Both are listed at 100 W.

The Verdict

The data is unambiguous. The AMD Radeon RX 6600M wins every recorded head-to-head benchmark, with leads of 60.6% to 67% depending on API, and it also wins on architecture: newer RDNA 2 process, more shading units, more ROPs, more memory, higher bandwidth, and the only one with ray cores. It also supports the newest DirectX and Vulkan versions.

The Radeon Pro Vega 20 remains a capable mobile-class part with a wide memory bus and HBM2 memory, but it belongs to an older architecture generation. For users whose only choice is between these two, the RX 6600M is the one the recorded data favors in every single test.