AMD Radeon R9 M390X vs NVIDIA GeForce RTX 3080 Comparison

Head-to-Head Benchmarks

The recorded database contains one shared benchmark between the NVIDIA GeForce RTX 3080 and the AMD Radeon R9 M390X: Geekbench Vulkan. In this test, the RTX 3080 scores 33,620 points against 16,951 points for the R9 M390X. This translates to a 98.3% advantage for the NVIDIA part, meaning the RTX 3080 delivers nearly double the Vulkan performance of the older AMD mobile GPU. The RTX 3080 wins the only head-to-head comparison, with a 1-0 record in the database.

Looking at the broader average benchmark scores, the RTX 3080 posts an average of 23,172 points across all recorded tests, while the R9 M390X averages 20,662 points. That is a gap of roughly 12% in favor of the RTX 3080, though the two cards sit in different performance tiers relative to all other GPUs. The RTX 3080 lands at the 68th percentile, while the R9 M390X sits at the 66th percentile, a modest separation that reflects how the average score is weighted by different test suites. The RTX 3080 has ten recorded benchmarks spanning DirectX 9 through DirectX 12, OpenCL, Vulkan, and compute workloads, whereas the R9 M390X only has two recorded tests: Geekbench Metal and Geekbench Vulkan. The Metal score for the R9 M390X is 24,373, which is not directly comparable to any RTX 3080 test in the database.

The nearest rivals for the RTX 3080 show how tightly clustered its average score is. The NVIDIA P106-100 averages 23,249 points, a delta of -0.3% relative to the RTX 3080. The AMD Radeon Pro Vega 16 also averages 23,250 points, again -0.3%. The AMD Radeon RX 6600M averages 23,273 points, -0.4%, and the AMD Radeon R9 M290X averages 23,276 points, -0.4%. The RTX 3080 is effectively within a fraction of a percent of these four cards in average performance, indicating that the database places it in a very tight competitive band despite its much higher peak scores in individual tests.

For the R9 M390X, the nearest rivals are similarly close. The AMD Radeon RX 5600 XT averages 20,713 points, a delta of -0.2% relative to the R9 M390X. The Intel Arc A750 averages 20,582 points, +0.4%. The Intel Arc B570 averages 20,556 points, +0.5%. The NVIDIA GeForce RTX 3070 Mobile averages 20,534 points, +0.6%. The R9 M390X sits just above these four cards in average score, though the differences are small enough that they could be considered statistical noise.

The Vulkan head-to-head result, however, is not close. The 98.3% delta means the RTX 3080 more than doubles the R9 M390X in that specific API workload. That is a decisive margin, far larger than the average score gap would suggest. The R9 M390X's Vulkan implementation, based on GCN 3.0, appears to be a significant bottleneck compared to the Ampere architecture's Vulkan path.

Architecture Differences

The architectural gap between these two GPUs is substantial. The RTX 3080 uses the GA102 chip built on Samsung's 8 nm process, while the R9 M390X uses the Amethyst chip on TSMC's 28 nm process. The manufacturing node difference alone explains much of the performance and efficiency disparity. The RTX 3080 packs 28,300 million transistors on a 628 mm² die, giving a transistor density of 45.1 million per square millimeter. The R9 M390X has 5,000 million transistors on a 366 mm² die, a density of 13.7 million per square millimeter. That is a 3.3x density advantage for the RTX 3080, which directly translates to higher compute throughput per unit area.

The RTX 3080 is built on the Ampere architecture, which introduces dedicated ray tracing cores and tensor cores. It has 68 RT cores and 272 tensor cores, neither of which exist on the R9 M390X. The AMD part uses GCN 3.0, a design that predates hardware ray tracing and tensor acceleration. This means the RTX 3080 can accelerate ray-traced workloads and AI inference tasks, while the R9 M390X must handle those workloads on general-purpose shaders.

Shader resources differ massively. The RTX 3080 has 8,704 shading units, 272 texture mapping units, and 96 render output units. The R9 M390X has 2,048 shading units, 128 TMUs, and 32 ROPs. The RTX 3080 has 4.25x the shading units, 2.125x the TMUs, and 3x the ROPs. Pixel rate for the RTX 3080 is 164.2 GPixel/s versus 23.14 GPixel/s for the R9 M390X, a 7.1x advantage. Texture rate is 465.1 GTexel/s versus 92.54 GTexel/s, a 5x advantage. FP32 compute is 29.77 TFLOPS versus 2.961 TFLOPS, a 10x advantage. The RTX 3080 also matches FP16 at 29.77 TFLOPS with a 1:1 ratio, while the R9 M390X also offers 2.961 TFLOPS FP16, again 1:1.

Memory subsystems are from different eras. The RTX 3080 has 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s of bandwidth. The R9 M390X has 4 GB of GDDR5 on a 256-bit bus, delivering 160.0 GB/s. The RTX 3080 has 4.75x the memory bandwidth and 2.5x the capacity. The memory clock rates reflect this: the RTX 3080 runs its memory at 1188 MHz with 19 Gbps effective, while the R9 M390X runs at 1250 MHz with 5 Gbps effective. The GDDR6X standard enables far higher data rates per pin.

The RTX 3080 supports PCIe 4.0 x16, while the R9 M390X uses PCIe 3.0 x16. This doubles the interconnect bandwidth for the newer card, which matters for data transfers in compute workloads and some game streaming scenarios. The RTX 3080 also supports DirectX 12 Ultimate (12_2), while the R9 M390X only reaches DirectX 12 (12_0). The RTX 3080 supports Vulkan 1.4, while the R9 M390X supports Vulkan 1.2.170. Both support OpenGL 4.6.

Power characteristics differ drastically. The RTX 3080 has a TDP of 320 W and requires a 700 W power supply, with a single 12-pin connector. The R9 M390X has a TDP of 75 W and uses no power connectors, running entirely off the MXM module slot. This makes the R9 M390X far more suitable for portable or low-power systems, but it also caps its performance ceiling. The RTX 3080 is a dual-slot card measuring 285 mm in length, 112 mm in height, and 40 mm in width. The R9 M390X has no recorded dimensions, as it is an MXM module whose physical size depends on the host device.

FAQ

Q: Which GPU wins in the only shared benchmark, Geekbench Vulkan?

A: The NVIDIA GeForce RTX 3080 wins decisively with a score of 33,620 against 16,951 for the AMD Radeon R9 M390X, a 98.3% advantage.

Q: How do their average benchmark scores compare?

A: The RTX 3080 averages 23,172 points across ten tests, while the R9 M390X averages 20,662 points across two tests. The RTX 3080 is about 12% higher on average.

Q: What are the key memory differences?

A: The RTX 3080 has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. The R9 M390X has 4 GB of GDDR5 on a 256-bit bus with 160.0 GB/s bandwidth.

Q: Does the R9 M390X support ray tracing or tensor cores?

A: No. The R9 M390X has no RT cores or tensor cores. The RTX 3080 has 68 RT cores and 272 tensor cores.

Q: What is the process node difference?

A: The RTX 3080 is built on Samsung's 8 nm process, while the R9 M390X is built on TSMC's 28 nm process. The RTX 3080 also has a transistor density of 45.1M per mm² versus 13.7M per mm² for the R9 M390X.

Q: Which GPU has a higher percentile ranking among all GPUs?

A: The RTX 3080 sits at the 68th percentile, while the R9 M390X sits at the 66th percentile.

The Verdict

The data clearly favors the NVIDIA GeForce RTX 3080 in raw performance. It wins the only head-to-head benchmark by 98.3%, and its average score is roughly 12% higher despite covering a much wider test suite. The architectural gap is enormous: the RTX 3080 has 10x the FP32 compute, 7.1x the pixel rate, 5x the texture rate, and 4.75x the memory bandwidth. It also adds hardware ray tracing and tensor cores, features the R9 M390X completely lacks.

Who should pick the RTX 3080? Any user who needs high-end gaming performance, ray-traced workloads, or AI acceleration. The 10 GB GDDR6X memory and 760.3 GB/s bandwidth make it suitable for high-resolution textures and compute-heavy applications. The 68 RT cores and 272 tensor cores provide dedicated hardware for next-generation rendering and inference tasks.

Who should pick the R9 M390X? Users constrained by power and physical space. The 75 W TDP with no external power connectors means it can fit in portable devices and MXM-based systems where a 320 W dual-slot card cannot. It also has a lower transistor count and smaller die, which may be relevant for thermal-constrained chassis. However, the data shows its performance is firmly in a different class, with a Vulkan score less than half of the RTX 3080.

The percentile ranking shows both cards are above average, but the RTX 3080 at the 68th percentile relative to all GPUs, versus the 66th percentile for the R9 M390X, reinforces that the RTX 3080 belongs to a higher performance tier. The nearest rivals for each card confirm this: the RTX 3080 competes with cards like the RX 6600M and R9 M290X, while the R9 M390X competes with the RX 5600 XT and Arc A750. These are different performance neighborhoods.

For a database-driven recommendation, the RTX 3080 is the clear choice for performance-oriented workloads. The R9 M390X has no benchmark advantage in any recorded test, and its only saving grace is its low power draw and mobile form factor. If the use case demands anything beyond basic rendering, the RTX 3080 dominates.

Specification Differences

| Specification | NVIDIA GeForce RTX 3080 | AMD Radeon R9 M390X |

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

| Architecture | Ampere | GCN 3.0 |

| Process Node | 8 nm | 28 nm |

| Foundry | Samsung | TSMC |

| Transistors | 28,300 million | 5,000 million |

| Die Size | 628 mm² | 366 mm² |

| Transistor Density | 45.1M / mm² | 13.7M / mm² |

| Memory Size | 10 GB | 4 GB |

| Memory Type | GDDR6X | GDDR5 |

| Memory Bus Width | 320 bit | 256 bit |

| Memory Bandwidth | 760.3 GB/s | 160.0 GB/s |

| Memory Clock | 1188 MHz, 19 Gbps effective | 1250 MHz, 5 Gbps effective |

| Shading Units | 8704 | 2048 |

| TMUs | 272 | 128 |

| ROPs | 96 | 32 |

| RT Cores | 68 | None |

| Tensor Cores | 272 | None |

| Pixel Rate | 164.2 GPixel/s | 23.14 GPixel/s |

| Texture Rate | 465.1 GTexel/s | 92.54 GTexel/s |

| FP32 Compute | 29.77 TFLOPS | 2.961 TFLOPS |

| FP16 Compute | 29.77 TFLOPS (1:1) | 2.961 TFLOPS (1:1) |

| TDP | 320 W | 75 W |

| Slot Width | Dual-slot | MXM Module |

| Power Connectors | 1x 12-pin | None |

| Suggested PSU | 700 W | None |

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

| Display Outputs | 1x HDMI 2.1, 3x DisplayPort 1.4a | Portable Device Dependent |

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

| Vulkan Support | 1.4 | 1.2.170 |

| OpenGL Support | 4.6 | 4.6 |

| Release Date | 2020-08-31 | 2015-05-04 |

| Predecessor | GeForce 20 | Solar System |

| Successor | GeForce 40 | Polaris Mobile |

| Launch MSRP | 699 USD | None |