AMD Radeon Pro 575X vs NVIDIA GeForce RTX 5090 Mobile Comparison

# NVIDIA GeForce RTX 5090 Mobile vs AMD Radeon Pro 575X

This comparison pairs NVIDIA's current-generation GeForce RTX 5090 Mobile, built on Blackwell 2.0 and released in 2025, against AMD's Radeon Pro 575X, an end-of-life GCN 4.0 part from 2019. The benchmark data shows a decisive generational gap: the RTX 5090 Mobile wins both shared head-to-head tests by margins exceeding 420%, and its average benchmark score of 45,152 sits roughly 15% above the Radeon Pro 575X's 39,116. However, each card occupies a distinct performance tier, and the data reveals where the older AMD part still holds relevance.

Where Each One Wins

The RTX 5090 Mobile is the clear winner in every shared benchmark category. In Geekbench OpenCL, it scores 201,834 versus the Radeon Pro 575X's 34,773, a 480.4% advantage. In Geekbench Vulkan, the NVIDIA card posts 198,405 against 37,919, leading by 423.2%. These are not marginal differences; they represent completely different performance classes. The RTX 5090 Mobile also benefits from a broader benchmark suite, with PassMark scores across DirectX 9, 10, 11, and 12, plus G2D, G3D, and compute tests, while the AMD card only has three recorded Geekbench results (OpenCL, Vulkan, and Metal).

The Radeon Pro 575X's sole exclusive benchmark is Geekbench Metal, where it scores 44,655. This is a niche advantage: the NVIDIA card has no Metal result listed, so the AMD part wins by default in that specific API environment. However, this does not translate into any overall victory, as the head-to-head results show winsA at 2 and winsB at 0. The RTX 5090 Mobile's percentile ranking among all GPUs sits at 84, slightly ahead of the Radeon Pro 575X's 82, reinforcing that the NVIDIA card is not just faster in direct comparisons but also sits higher in the broader GPU landscape.

Architecture Differences

The architectural gap between these two GPUs is generational and substantial. The RTX 5090 Mobile uses the GB203 chip on TSMC's 5 nm process, packing 45,600 million transistors into a 378 mm² die for a density of 120.6 million transistors per mm². The Radeon Pro 575X uses the Ellesmere chip on GlobalFoundries' 14 nm process, with 5,700 million transistors on a 232 mm² die, yielding just 24.6 million transistors per mm². This means the NVIDIA chip has 8x the transistor count on a die that is only 63% larger, a direct reflection of process node advantages.

The RTX 5090 Mobile features 10,496 shading units, 328 texture mapping units, 112 ROPs, 82 ray tracing cores, and 328 tensor cores. The Radeon Pro 575X has 2,048 shading units, 128 TMUs, and 32 ROPs, with no ray tracing or tensor cores listed. The NVIDIA card supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, while the AMD part supports DirectX 12 (12_0) and Vulkan 1.3. Both support OpenGL 4.6. The RTX 5090 Mobile also uses PCIe 5.0 x16, compared to the Radeon Pro 575X's PCIe 3.0 x16, doubling the potential bus bandwidth.

Memory configurations differ drastically. The RTX 5090 Mobile has 24 GB of GDDR7 on a 256-bit bus, delivering 896.0 GB/s bandwidth. The Radeon Pro 575X has 4 GB of GDDR5 on a 256-bit bus, with 217.0 GB/s bandwidth. Clock speeds tell a similar story: the NVIDIA card runs at 990 MHz base and 1515 MHz boost with 1750 MHz memory (28 Gbps effective), while the AMD part's core clocks are not listed, and its memory runs at 1695 MHz (6.8 Gbps effective). Power draw also differs, with the RTX 5090 Mobile rated at 95 W TDP versus the Radeon Pro 575X's 150 W TDP, meaning the newer card delivers far more performance while consuming 37% less power.

Head-to-Head Benchmarks

The two shared benchmarks paint a stark picture. In Geekbench OpenCL, the RTX 5090 Mobile scores 201,834 against the Radeon Pro 575X's 34,773, a delta of 480.4%. This is not a close race; the NVIDIA card outperforms by nearly five times. The OpenCL test stresses general-purpose compute, and the RTX 5090 Mobile's 31.80 TFLOPS FP32 throughput versus the Radeon Pro 575X's 4.489 TFLOPS explains the magnitude of the gap. The NVIDIA card's texture rate of 496.9 GTexel/s and pixel rate of 169.7 GPixel/s dwarf the AMD part's 140.3 GTexel/s and 35.07 GPixel/s.

In Geekbench Vulkan, the RTX 5090 Mobile scores 198,405 versus 37,919, a 423.2% advantage. Vulkan is a low-level API that often favors newer architectures, and the RTX 5090 Mobile's Blackwell 2.0 design with dedicated ray tracing and tensor cores clearly excels here. The Radeon Pro 575X, with its GCN 4.0 architecture and no dedicated RT or tensor hardware, cannot compete in modern API workloads. The delta percentages are consistent across both tests, suggesting the performance gap is architectural rather than test-specific.

Looking at the broader benchmark context, the RTX 5090 Mobile's PassMark G3D score of 30,034 and GPU compute score of 13,401 indicate strong rasterization and compute performance. Its 3DMark Steel Nomad DX12 score of 5,871 further confirms DirectX 12 capability. The Radeon Pro 575X has no equivalent PassMark or 3DMark results, so direct comparison in those tests is impossible, but the Geekbench deltas are sufficient to establish the NVIDIA card's dominance.

The Verdict

The data unequivocally favors the RTX 5090 Mobile for any workload involving OpenCL or Vulkan. Its 480.4% lead in OpenCL and 423.2% lead in Vulkan mean that any application leveraging these APIs will see dramatically higher performance on the NVIDIA card. The RTX 5090 Mobile also has a 24 GB memory buffer versus 4 GB, which matters for large datasets or high-resolution textures. Its newer architecture supports DirectX 12 Ultimate and Vulkan 1.4, ensuring compatibility with modern games and applications.

The Radeon Pro 575X is only relevant in one specific scenario: Metal API workloads. Its Geekbench Metal score of 44,655 is the only benchmark where it has a recorded result that the NVIDIA card lacks. This makes it a potential option for macOS environments where Metal is the primary graphics API. However, the Radeon Pro 575X is end-of-life, has no ray tracing or tensor cores, supports only DirectX 12 (12_0), and consumes 150 W TDP versus the RTX 5090 Mobile's 95 W.

For general-purpose computing, gaming, or any modern workload, the RTX 5090 Mobile is the clear choice. Its 84th percentile ranking versus the Radeon Pro 575X's 82nd percentile, combined with the massive benchmark deltas, leaves no ambiguity. The Radeon Pro 575X's nearest rivals (Radeon Pro 580X at 38,706, GeForce MX570 A at 38,691) all score within 1.1% of its 39,116 average, placing it in a performance tier that the RTX 5090 Mobile exceeds by roughly 15% on average, and by much more in individual tests.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce RTX 5090 Mobile has an average benchmark score of 45,152, which is 15% higher than the AMD Radeon Pro 575X's average of 39,116.

Q: What is the largest performance gap in the head-to-head tests?

A: The largest gap is in Geekbench OpenCL, where the RTX 5090 Mobile leads by 480.4%, scoring 201,834 versus the Radeon Pro 575X's 34,773.

Q: Does the AMD Radeon Pro 575X win any shared benchmark?

A: No. In the two shared head-to-head benchmarks (Geekbench OpenCL and Geekbench Vulkan), the RTX 5090 Mobile wins both, giving NVIDIA 2 wins and AMD 0 wins.

Q: What is the memory capacity difference?

A: The RTX 5090 Mobile has 24 GB of GDDR7 memory, while the Radeon Pro 575X has 4 GB of GDDR5 memory. Both use a 256-bit bus width.

Q: Which GPU has a higher transistor count and density?

A: The RTX 5090 Mobile has 45,600 million transistors on a 378 mm² die (120.6M / mm²), while the Radeon Pro 575X has 5,700 million transistors on a 232 mm² die (24.6M / mm²).

Q: How do the power requirements compare?

A: The RTX 5090 Mobile is rated at 95 W TDP, while the Radeon Pro 575X is rated at 150 W TDP, meaning the newer NVIDIA card consumes less power despite higher performance.

Specification Differences

| Specification | NVIDIA GeForce RTX 5090 Mobile | AMD Radeon Pro 575X |

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

| Architecture | Blackwell 2.0 | GCN 4.0 |

| Process Node | 5 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 45,600 million | 5,700 million |

| Die Size | 378 mm² | 232 mm² |

| Transistor Density | 120.6M / mm² | 24.6M / mm² |

| Memory Size | 24 GB | 4 GB |

| Memory Type | GDDR7 | GDDR5 |

| Memory Clock | 1750 MHz (28 Gbps effective) | 1695 MHz (6.8 Gbps effective) |

| Memory Bandwidth | 896.0 GB/s | 217.0 GB/s |

| Shading Units | 10,496 | 2,048 |

| TMUs | 328 | 128 |

| ROPs | 112 | 32 |

| RT Cores | 82 | None |

| Tensor Cores | 328 | None |

| Pixel Rate | 169.7 GPixel/s | 35.07 GPixel/s |

| Texture Rate | 496.9 GTexel/s | 140.3 GTexel/s |

| FP32 Performance | 31.80 TFLOPS | 4.489 TFLOPS |

| TDP | 95 W | 150 W |

| Bus Interface | PCIe 5.0 x16 | PCIe 3.0 x16 |

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

| Vulkan Support | 1.4 | 1.3 |

| Production Status | Active | End-of-life |

| Release Date | 2025-03-26 | 2019-03-17 |