AMD Radeon R9 M390X vs NVIDIA Quadro RTX 5000 Comparison

The NVIDIA Quadro RTX 5000 and the AMD Radeon R9 M390X are separated by three years of architecture evolution and a massive gap in performance class. The data shows a decisive victory for the NVIDIA part, which outperforms the AMD mobile chip by 444.6% in the only shared benchmark, while the AMD card’s sole advantage lies in its dramatically lower power draw. This comparison is not a close contest; it is a demonstration of how far GPU technology advanced between 2015 and 2018.

FAQ

Q: Which GPU is faster in the head-to-head benchmark?

A: The NVIDIA Quadro RTX 5000 wins the only shared test, Geekbench Vulkan, with a score of 92,309 against the AMD Radeon R9 M390X’s 16,951. This represents a 444.6% performance advantage for the NVIDIA card.

Q: How do their average benchmark scores compare?

A: The NVIDIA Quadro RTX 5000 has an average benchmark score of 21,629, while the AMD Radeon R9 M390X scores 20,662. The NVIDIA card is about 4.7% higher on average, though this margin is smaller than the Vulkan delta suggests due to the different benchmark sets each card supports.

Q: What are the key architectural differences?

A: The NVIDIA Quadro RTX 5000 uses the Turing architecture on a 12 nm process with 13,600 million transistors, while the AMD Radeon R9 M390X uses GCN 3.0 on a 28 nm process with 5,000 million transistors. The NVIDIA chip also features 48 RT cores and 384 tensor cores, which the AMD chip lacks entirely.

Q: How does memory capacity and bandwidth differ?

A: The NVIDIA Quadro RTX 5000 offers 16 GB of GDDR6 memory with a 256-bit bus and 448.0 GB/s bandwidth. The AMD Radeon R9 M390X provides 4 GB of GDDR5 on a 256-bit bus with 160.0 GB/s bandwidth. This gives the NVIDIA card four times the capacity and 2.8 times the bandwidth.

Q: Which card is more power-efficient?

A: The AMD Radeon R9 M390X has a TDP of 75 W, compared to the NVIDIA Quadro RTX 5000’s 230 W. The AMD card draws roughly one-third the power of the NVIDIA part, making it far more suitable for power-constrained mobile systems.

Q: What is the performance percentile ranking for each?

A: The NVIDIA Quadro RTX 5000 sits in the 67th percentile of all GPUs, while the AMD Radeon R9 M390X sits in the 66th percentile. Despite the massive head-to-head gap, both cards rank similarly relative to the entire GPU landscape.

Architecture Differences

The NVIDIA Quadro RTX 5000 is built on the Turing architecture, fabricated by TSMC on a 12 nm process. Its TU104 chip packs 13,600 million transistors into a 545 mm² die, yielding a transistor density of 25.0M per mm². In contrast, the AMD Radeon R9 M390X uses the older GCN 3.0 architecture with the Amethyst chip, manufactured on a 28 nm process at TSMC. The AMD silicon contains 5,000 million transistors across a 366 mm² die, for a density of 13.7M per mm². The NVIDIA chip has 2.7 times more transistors and nearly 1.5 times the die area.

The compute resources differ substantially. The NVIDIA card features 3,072 shading units, 192 texture mapping units, and 64 ROPs. The AMD card has 2,048 shading units, 128 TMUs, and 32 ROPs. More importantly, the NVIDIA GPU includes 48 RT cores for ray tracing and 384 tensor cores for AI workloads, while the AMD Radeon R9 M390X has neither. This makes the Quadro RTX 5000 a fundamentally more capable processor for specialized workloads beyond traditional rasterization.

Memory architecture also diverges. The NVIDIA card uses 16 GB of GDDR6 with a 256-bit bus, achieving 448.0 GB/s bandwidth. The AMD card uses 4 GB of GDDR5 on the same 256-bit bus, but only reaches 160.0 GB/s. Clock speeds tell a similar story: the NVIDIA part runs at 1620 MHz base and 1815 MHz boost, while the AMD card’s base and boost clocks are not listed in the data. The memory clock for AMD is 1250 MHz (5 Gbps effective), versus NVIDIA’s 1750 MHz (14 Gbps effective).

API support gives the NVIDIA card another edge. It supports DirectX 12 Ultimate (12_2), while the AMD card supports DirectX 12 (12_0). Both support OpenGL 4.6, but NVIDIA’s Vulkan version is 1.4 compared to AMD’s 1.2.170. The NVIDIA card also has a much richer feature set for display output, offering 4x DisplayPort 1.4a and 1x USB Type-C, whereas the AMD card’s outputs are listed as portable device dependent.

Where Each One Wins

The NVIDIA Quadro RTX 5000 wins on nearly every measurable axis of raw performance. Its FP32 compute is 11.15 TFLOPS versus the AMD card’s 2.961 TFLOPS, a 3.8x advantage. Pixel rate is 116.2 GPixel/s versus 23.14 GPixel/s, a 5x lead. Texture rate is 348.5 GTexel/s versus 92.54 GTexel/s, a 3.8x margin. The NVIDIA card also supports FP16 at 22.30 TFLOPS (2:1 ratio), while the AMD card’s FP16 is identical to its FP32 at 2.961 TFLOPS (1:1 ratio).

The AMD Radeon R9 M390X wins in power efficiency. Its 75 W TDP is less than one-third of the NVIDIA card’s 230 W. It also uses no external power connectors, while the NVIDIA card requires a 1x 6-pin plus 1x 8-pin configuration. The AMD card is an MXM module, meaning it is designed for laptop form factors where power and space are at a premium, whereas the NVIDIA card is a dual-slot desktop solution measuring 267 mm in length.

In benchmark breadth, the AMD card has one unique test: Geekbench Metal, where it scores 24,373. The NVIDIA card has no Metal benchmark listed, so it cannot be compared there. The NVIDIA card, however, has a wide range of Passmark tests (DirectX 9 through 12, G2D, G3D, and GPU compute) that the AMD card lacks entirely. This means the NVIDIA card’s performance profile is far more fully documented.

Specification Differences

| Specification | NVIDIA Quadro RTX 5000 | AMD Radeon R9 M390X |

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

| Architecture | Turing | GCN 3.0 |

| Process Node | 12 nm | 28 nm |

| Transistors | 13,600 million | 5,000 million |

| Die Size | 545 mm² | 366 mm² |

| Transistor Density | 25.0M / mm² | 13.7M / mm² |

| Shading Units | 3072 | 2048 |

| TMUs | 192 | 128 |

| ROPs | 64 | 32 |

| RT Cores | 48 | None |

| Tensor Cores | 384 | None |

| Memory Size | 16 GB | 4 GB |

| Memory Type | GDDR6 | GDDR5 |

| Memory Bus | 256 bit | 256 bit |

| Memory Bandwidth | 448.0 GB/s | 160.0 GB/s |

| Base Clock | 1620 MHz | Not listed |

| Boost Clock | 1815 MHz | Not listed |

| Memory Clock | 1750 MHz (14 Gbps) | 1250 MHz (5 Gbps) |

| FP32 Performance | 11.15 TFLOPS | 2.961 TFLOPS |

| FP16 Performance | 22.30 TFLOPS (2:1) | 2.961 TFLOPS (1:1) |

| Pixel Rate | 116.2 GPixel/s | 23.14 GPixel/s |

| Texture Rate | 348.5 GTexel/s | 92.54 GTexel/s |

| TDP | 230 W | 75 W |

| Slot Width | Dual-slot | MXM Module |

| Power Connectors | 1x 6-pin + 1x 8-pin | None |

| Suggested PSU | 550 W | Not listed |

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

| Vulkan Support | 1.4 | 1.2.170 |

| Release Date | 2018-08-12 | 2015-05-04 |

Head-to-Head Benchmarks

The only directly comparable benchmark between these two cards is Geekbench Vulkan. The NVIDIA Quadro RTX 5000 scores 92,309, while the AMD Radeon R9 M390X scores 16,951. This yields a delta of 444.6% in favor of the NVIDIA card. That is a staggering gap — the NVIDIA card is over five times faster in this API test. This result is consistent with the raw compute specifications, where the NVIDIA card holds a 3.8x lead in FP32 throughput, a 5x lead in pixel rate, and a 3.8x lead in texture rate.

The NVIDIA card’s Vulkan score also dwarfs its own average benchmark score of 21,629, indicating that Vulkan is a particularly strong workload for the Turing architecture. The AMD card’s Vulkan score of 16,951 falls below its average of 20,662, suggesting the GCN 3.0 architecture is less optimized for this API. For context, the NVIDIA card’s nearest rival by average score is the NVIDIA GeForce GTX 1060 6 GB at 21,856 (a -1% delta), while the AMD card’s nearest rival is the AMD Radeon RX 5600 XT at 20,713 (a -0.2% delta). Despite these similar average scores, the Vulkan head-to-head shows a completely different class of performance.

The AMD card does have a Geekbench Metal score of 24,373, which is its strongest listed result. However, without a corresponding NVIDIA Metal score, this cannot be used for direct comparison. The NVIDIA card also has multiple Passmark scores — including 15,616 in G3D and 6,525 in GPU compute — that have no AMD equivalent in the data, further limiting direct cross-referencing beyond the Vulkan test.

The Verdict

The verdict is unambiguous: the NVIDIA Quadro RTX 5000 is the superior GPU by a wide margin. In the only shared benchmark, it delivers 444.6% more performance than the AMD Radeon R9 M390X. It offers 4x the memory capacity, 2.8x the bandwidth, 3.8x the FP32 compute, and a full suite of modern features including RT cores and tensor cores that the AMD card lacks entirely. Its 67th percentile ranking versus the AMD card’s 66th percentile may look close, but that ranking is distorted by the different benchmark sets each card supports.

The AMD Radeon R9 M390X has exactly one compelling attribute: power consumption. At 75 W, it draws less than a third of the NVIDIA card’s 230 W, and its MXM form factor with no external power connectors makes it a drop-in solution for thin-and-light laptops. If the workload is basic 3D rendering or compute that fits within 4 GB of memory, the AMD card can get the job done without taxing a laptop’s thermal or power budget.

For any professional workstation task — large datasets, ray tracing, AI inference, or high-resolution texture work — the NVIDIA Quadro RTX 5000 is the only rational choice. Its 16 GB of GDDR6 memory, 48 RT cores, and 384 tensor cores make it a versatile tool for modern workloads, and its performance lead is so large that no power efficiency argument can overcome it. The AMD card is a legacy mobile part from 2015, while the NVIDIA card is a 2018 workstation-class solution. The data shows a clear generational and performance hierarchy: the Quadro RTX 5000 wins on every meaningful metric except power draw, and even there, the AMD card’s win is purely a matter of form factor efficiency rather than computational capability.