AMD Radeon RX 6700M vs NVIDIA GeForce RTX 3090 Comparison

The NVIDIA GeForce RTX 3090 and AMD Radeon RX 6700M occupy opposite ends of the hardware spectrum: one is a desktop flagship with a triple-slot cooler and a 350 W power envelope, the other a mobile GPU soldered into laptops with a 135 W design. Benchmark data confirms the RTX 3090 dominates in nine of ten head-to-head tests, while the RX 6700M claims a single, albeit significant, victory. The average benchmark score for the RTX 3090 sits at 27,565, placing it in the 73rd percentile of all GPUs, whereas the RX 6700M averages 25,633 and lands in the 71st percentile. These figures show a narrower gap in overall standing than the raw deltaPct values suggest, largely because the RX 6700M’s nearest rivals—the AMD Radeon Pro W5700 and NVIDIA GeForce RTX 3080 Ti Mobile—score within 0.4% of its average, meaning it is highly competitive within its own mobile class.

Where Each One Wins

The RTX 3090 is the clear winner for raw compute throughput and legacy API performance. In PassMark GPU Compute, it scores 15,356 against the RX 6700M’s 5,191, a delta of 195.8%—the largest margin of any test. This makes it the obvious choice for workloads like rendering, scientific simulation, or any task that leverages FP32 or general-purpose compute. It also wins every DirectX API test: DirectX 9 (268 vs 155, +72.9%), DirectX 10 (182 vs 92, +97.8%), DirectX 11 (220 vs 129, +70.5%), and DirectX 12 (110 vs 65, +69.2%). For users running older games or applications that rely on legacy DirectX paths, the RTX 3090 will deliver substantially higher frame rates and smoother performance.

The RX 6700M’s sole win comes in Geekbench Vulkan, where it scores 90,816 against the RTX 3090’s 53,927, a delta of -40.6% in AMD’s favor. This result indicates that in Vulkan-based titles or compute workloads, the mobile AMD part can outperform the desktop NVIDIA flagship by a wide margin. The architecture’s efficiency in Vulkan, combined with its higher boost clock of 2400 MHz versus the RTX 3090’s 1695 MHz, suggests that the RX 6700M is better optimized for this modern API. However, this single win does not offset the RTX 3090’s advantages elsewhere, particularly in OpenCL where the NVIDIA card scores 172,758 versus 77,666 (+122.4%).

For 2D performance, the RTX 3090 wins PassMark G2D with 1,063 points against 555 points (+91.5%), indicating faster desktop rendering, image scaling, and UI responsiveness. The RTX 3090 also dominates in 3DMark Steel Nomad DX12, scoring 5,118 versus 1,845 (+177.4%), a test that stresses next-generation DirectX 12 features. This makes the RTX 3090 the better pick for high-refresh-rate 1440p or 4K gaming, while the RX 6700M’s Vulkan strength could favor users on Linux or Vulkan-centric game engines.

Architecture Differences

The RTX 3090 is built on NVIDIA’s Ampere architecture using the GA102 chip, fabricated on Samsung’s 8 nm process. It packs 28,300 million transistors into a 628 mm² die, yielding a transistor density of 45.1 million per mm². The RX 6700M uses AMD’s RDNA 2.0 architecture with the Navi 22 chip, produced on TSMC’s 7 nm node, with 17,200 million transistors across a 335 mm² die—a density of 51.3 million per mm². The smaller, denser AMD chip allows for a dramatically lower power draw: 135 W versus 350 W, a reduction of over 60%.

Core configurations differ starkly. The RTX 3090 features 10,496 shading units, 328 texture mapping units, 112 render output units, 82 ray tracing cores, and 328 tensor cores. The RX 6700M has 2,304 shading units, 144 TMUs, 64 ROPs, and 36 ray tracing cores, with no tensor cores listed. This means the RTX 3090 has over 4.5 times the shading units and 2.3 times the ROPs, explaining its massive lead in fill-rate-bound tests. The RTX 3090’s pixel rate is 189.8 GPixel/s versus 153.6 GPixel/s, and its texture rate is 556.0 GTexel/s versus 345.6 GTexel/s.

FP32 compute tells a similar story: the RTX 3090 delivers 35.58 TFLOPS, while the RX 6700M manages 11.06 TFLOPS—a 3.2x difference. However, the RX 6700M’s FP16 output of 22.12 TFLOPS (at a 2:1 ratio) exceeds its FP32 rate, whereas the RTX 3090 runs FP16 at a 1:1 ratio (35.58 TFLOPS). This makes the RX 6700M relatively stronger in mixed-precision workloads if software is optimized for FP16, though the RTX 3090’s absolute FP16 number remains higher.

Memory subsystems are also divergent. The RTX 3090 uses 24 GB of GDDR6X on a 384-bit bus, achieving 936.2 GB/s bandwidth. The RX 6700M has 10 GB of GDDR6 on a 160-bit bus, providing 320.0 GB/s. The RTX 3090’s bandwidth advantage is 2.9x, which is critical for 4K textures and large datasets.

Head-to-Head Benchmarks

The most decisive victory for the RTX 3090 is in PassMark GPU Compute, where its score of 15,356 more than triples the RX 6700M’s 5,191 (+195.8%). This test typically measures raw arithmetic throughput, and the RTX 3090’s 10,496 shading units and 328 tensor cores provide overwhelming parallelism. In 3DMark Steel Nomad DX12, the RTX 3090 scores 5,118 versus 1,845 (+177.4%), highlighting its superiority in future-proofed DirectX 12 workloads. The gap narrows slightly in OpenCL (172,758 vs 77,666, +122.4%), but the RTX 3090 still doubles the mobile part.

In legacy DirectX tests, the RTX 3090 leads by smaller margins: DirectX 10 at 182 vs 92 (+97.8%), DirectX 9 at 268 vs 155 (+72.9%), and DirectX 11 at 220 vs 129 (+70.5%). These results suggest that while the RTX 3090 is faster in older APIs, the relative advantage is less extreme than in compute-heavy tests, likely because those workloads are more latency-bound than throughput-bound. The DirectX 12 test shows a 69.2% lead (110 vs 65), indicating that even modern games will favor the desktop card, but not by the 3x margins seen in compute.

The RX 6700M’s win in Geekbench Vulkan is noteworthy not just for the delta (-40.6%) but for the absolute scores: 90,816 versus 53,927. This is the only test where the mobile GPU outperforms the desktop flagship, and the margin is substantial. The PassMark G3D test, which aggregates gaming performance, gives the RTX 3090 a 96.8% lead (26,645 vs 13,536), and the PassMark G2D test shows a 91.5% advantage (1,063 vs 555). Overall, the RTX 3090 wins nine tests, with an average delta of approximately 108% across all benchmarks, while the RX 6700M’s single win represents a significant but isolated strength.

Specification Differences

| Specification | NVIDIA GeForce RTX 3090 | AMD Radeon RX 6700M |

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

| Architecture | Ampere | RDNA 2.0 |

| Process Node | 8 nm | 7 nm |

| Foundry | Samsung | TSMC |

| Transistors | 28,300 million | 17,200 million |

| Die Size | 628 mm² | 335 mm² |

| Base Clock | 1395 MHz | 1489 MHz |

| Boost Clock | 1695 MHz | 2400 MHz |

| Memory Size | 24 GB | 10 GB |

| Memory Type | GDDR6X | GDDR6 |

| Memory Bus | 384 bit | 160 bit |

| Memory Bandwidth | 936.2 GB/s | 320.0 GB/s |

| Shading Units | 10496 | 2304 |

| TMUs | 328 | 144 |

| ROPs | 112 | 64 |

| RT Cores | 82 | 36 |

| Tensor Cores | 328 | None |

| FP32 Compute | 35.58 TFLOPS | 11.06 TFLOPS |

| FP16 Compute | 35.58 TFLOPS | 22.12 TFLOPS |

| TDP | 350 W | 135 W |

| Slot Width | Triple-slot | IGP |

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

| Suggested PSU | 750 W | None |

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

The RTX 3090’s launch MSRP is 1,499 USD. The RX 6700M has no listed launch MSRP. The RTX 3090 also has a longer physical footprint at 336 mm, while the RX 6700M has no listed dimensions, reflecting its mobile integration.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The RTX 3090 averages 27,565, while the RX 6700M averages 25,633, a difference of 1,932 points.

Q: Does the RX 6700M win any benchmark against the RTX 3090?

A: Yes, in Geekbench Vulkan, the RX 6700M scores 90,816 versus the RTX 3090’s 53,927, a delta of -40.6%.

Q: How much faster is the RTX 3090 in compute workloads?

A: In PassMark GPU Compute, the RTX 3090 scores 15,356 versus 5,191, a 195.8% lead, which is the largest margin of any test.

Q: What is the memory bandwidth difference?

A: The RTX 3090 has 936.2 GB/s, while the RX 6700M has 320.0 GB/s, making the RTX 3090 2.9x faster.

Q: How do the power requirements compare?

A: The RTX 3090 has a TDP of 350 W and requires a 750 W power supply, while the RX 6700M has a TDP of 135 W and no suggested PSU.

Q: Which GPU has more shading units?

A: The RTX 3090 has 10,496 shading units, versus 2,304 on the RX 6700M, a 4.5x difference.