AMD Radeon R9 M390X vs NVIDIA RTX 2000 Ada Generation Comparison

The AMD Radeon R9 M390X and NVIDIA RTX 2000 Ada Generation are separated by nearly a decade of GPU architecture, and the benchmark data reflects that gulf. The R9 M390X is an end-of-life mobile part from 2015, while the RTX 2000 Ada is an active workstation card from 2024. Their average benchmark scores are surprisingly close — 20,662 for the AMD versus 18,954 for the NVIDIA — but that single number hides a massive performance gap in modern workloads, where the Ada card is essentially in a different class.

Where Each One Wins

The R9 M390X wins in legacy API scenarios and raw rasterization throughput relative to its age, but the data shows it is largely outclassed. In the only direct head-to-head benchmark available, Geekbench Vulkan, the AMD scores 16,951 against the NVIDIA’s 83,360 — an 79.7% deficit. That is not a close contest; it is a generation gap made visible.

The RTX 2000 Ada Generation wins decisively in every modern compute and graphics metric. Its Geekbench OpenCL score of 78,074 and Vulkan score of 83,360 dwarf the AMD’s numbers. The NVIDIA also shows strength in DirectX 12, with a Passmark score of 71, and in GPU compute, where it scores 7,834. For any workload that leverages Vulkan, OpenCL, or DirectX 12, the RTX 2000 Ada is the clear choice.

The AMD’s only comparative advantage is in older DirectX 9 and DirectX 10 paths, where the NVIDIA’s Passmark scores are 216 and 82, respectively. However, the AMD does not have those exact scores listed, so we can only infer from the NVIDIA’s performance that the older card may hold its own in legacy titles. In practical terms, the R9 M390X is a card for running older games at lower settings, while the RTX 2000 Ada is built for professional workloads and modern APIs.

Architecture Differences

The two GPUs are built on completely different foundations. The AMD Radeon R9 M390X uses the Amethyst chip with a GCN 3.0 architecture, manufactured on a 28 nm process at TSMC. It packs 5,000 million transistors into a 366 mm² die, yielding a transistor density of 13.7 million per square millimeter. This is a large, hot chip by modern standards, though its TDP is rated at just 75 W.

The NVIDIA RTX 2000 Ada Generation uses the AD107 chip with the Ada Lovelace architecture, built on a 5 nm process, also at TSMC. It contains 18,900 million transistors in a much smaller 159 mm² die, giving it a transistor density of 118.9 million per square millimeter — roughly 8.7 times denser than the AMD. Despite having nearly four times the transistors, the NVIDIA card consumes only 70 W TDP, slightly less than the AMD.

Memory configurations differ significantly. The AMD has 4 GB of GDDR5 on a 256-bit bus, delivering 160.0 GB/s of bandwidth. The NVIDIA has 16 GB of GDDR6 on a 128-bit bus, delivering 256.0 GB/s. The NVIDIA’s smaller bus width is compensated by faster memory, giving it 60% more bandwidth than the AMD.

Feature support is where the NVIDIA pulls far ahead. The RTX 2000 Ada includes 22 ray tracing cores and 88 tensor cores, neither of which exist on the R9 M390X. The NVIDIA also supports DirectX 12 Ultimate (12_2), while the AMD only reaches DirectX 12 (12_0). Vulkan support is newer on the NVIDIA (1.4 versus 1.2.170), and OpenGL is identical at 4.6.

Head-to-Head Benchmarks

The only directly comparable benchmark between these two cards is Geekbench Vulkan. The AMD R9 M390X scores 16,951, while the NVIDIA RTX 2000 Ada Generation scores 83,360. That is a delta of -79.7% for the AMD, meaning the NVIDIA is roughly 4.9 times faster in this test. This is the single most telling statistic in the entire comparison.

Looking at the broader benchmark suites, the NVIDIA’s dominance continues. In Geekbench OpenCL, the RTX 2000 Ada scores 78,074, which is 3.8 times higher than the AMD’s Vulkan score and far beyond anything the AMD could produce in a compute workload. The NVIDIA’s Passmark G3D score of 16,927 is nearly identical to its average benchmark score of 18,954, suggesting consistent performance across different test suites.

The AMD’s average benchmark score of 20,662 is actually higher than the NVIDIA’s 18,954, but this is misleading. The AMD’s scores come from only two tests — Geekbench Metal (24,373) and Geekbench Vulkan (16,951) — while the NVIDIA’s average pulls from ten different benchmarks, including several Passmark tests where it scores lower. The NVIDIA’s Passmark DirectX 9 score of 216 and DirectX 10 score of 82 are weak points, and its Passmark DirectX 11 score of 138 is modest. But in modern API workloads, the NVIDIA is untouchable.

The Verdict

The data is unambiguous: the NVIDIA RTX 2000 Ada Generation is the superior card for any modern workload. Its Vulkan performance is nearly five times better than the AMD R9 M390X, and its OpenCL compute score of 78,074 puts it in a different tier entirely. The NVIDIA also has 16 GB of VRAM versus the AMD’s 4 GB, which is critical for professional applications and modern games at high resolutions.

The AMD R9 M390X should only be considered for legacy use cases. Its 2.961 TFLOPS of FP32 performance and 92.54 GTexel/s texture rate are respectable for 2015, but they are dwarfed by the NVIDIA’s 12.00 TFLOPS and 187.4 GTexel/s. The AMD’s 23.14 GPixel/s pixel rate is less than a quarter of the NVIDIA’s 102.2 GPixel/s.

For a professional workstation, the RTX 2000 Ada is the obvious pick. Its 12.00 TFLOPS FP16 and FP32 performance, 22 ray tracing cores, and 88 tensor cores make it suitable for AI inference, rendering, and compute tasks. The AMD has no such acceleration features. For gaming in 2025, the NVIDIA’s DirectX 12 Ultimate support and Vulkan 1.4 compatibility ensure it will run modern titles, while the AMD’s DirectX 12 (12_0) and Vulkan 1.2.170 support is dated.

The RTX 2000 Ada also wins on efficiency. It delivers 12.00 TFLOPS at 70 W, while the AMD delivers 2.961 TFLOPS at 75 W. That is 171.4 GFLOPS per watt for the NVIDIA versus 39.5 GFLOPS per watt for the AMD — a 4.3 times efficiency advantage.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon R9 M390X has an average benchmark score of 20,662, compared to the NVIDIA RTX 2000 Ada Generation’s 18,954. However, the NVIDIA’s score is based on ten benchmarks, while the AMD’s is based on only two.

Q: How much faster is the NVIDIA in Vulkan performance?

A: In the Geekbench Vulkan test, the NVIDIA scores 83,360 against the AMD’s 16,951, making the NVIDIA approximately 79.7% faster.

Q: What is the memory capacity difference?

A: The NVIDIA RTX 2000 Ada has 16 GB of GDDR6 memory, while the AMD R9 M390X has 4 GB of GDDR5. The NVIDIA also has higher bandwidth at 256.0 GB/s versus 160.0 GB/s.

Q: Does the NVIDIA card support ray tracing?

A: Yes, the RTX 2000 Ada Generation has 22 ray tracing cores and 88 tensor cores. The AMD R9 M390X has no ray tracing or tensor core support.

Q: Which card has a smaller physical footprint?

A: The NVIDIA RTX 2000 Ada Generation is 168 mm long and 69 mm tall, making it a compact dual-slot card. The AMD R9 M390X uses an MXM module form factor, which is portable device dependent.

Q: What is the process technology difference?

A: The AMD uses a 28 nm process at TSMC, while the NVIDIA uses a 5 nm process, also at TSMC. The NVIDIA’s die is 159 mm² versus the AMD’s 366 mm².

Specification Differences

| Specification | AMD Radeon R9 M390X | NVIDIA RTX 2000 Ada Generation |

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

| Process Node | 28 nm | 5 nm |

| Transistors | 5,000 million | 18,900 million |

| Die Size | 366 mm² | 159 mm² |

| Memory | 4 GB GDDR5 | 16 GB GDDR6 |

| Memory Bus | 256 bit | 128 bit |

| Memory Bandwidth | 160.0 GB/s | 256.0 GB/s |

| Shading Units | 2048 | 2816 |

| TMUs | 128 | 88 |

| ROPs | 32 | 48 |

| Ray Tracing Cores | None | 22 |

| Tensor Cores | None | 88 |

| FP32 Performance | 2.961 TFLOPS | 12.00 TFLOPS |

| FP16 Performance | 2.961 TFLOPS | 12.00 TFLOPS |

| Pixel Rate | 23.14 GPixel/s | 102.2 GPixel/s |

| Texture Rate | 92.54 GTexel/s | 187.4 GTexel/s |

| TDP | 75 W | 70 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

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

| Vulkan Support | 1.2.170 | 1.4 |

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

| Release Date | 2015-05-04 | 2024-02-11 |