AMD Radeon 680M vs NVIDIA GeForce RTX 3070 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive overall victory for the NVIDIA GeForce RTX 3070, which claims two of the three shared benchmark tests. The margins in those wins are substantial, while the single AMD Radeon 680M win is narrow, creating a clear picture of where each part excels.

In the 3DMark Steel Nomad DX12 test, the RTX 3070 scores 3162 against 378 for the Radeon 680M. That is a delta of 736.5%, meaning the NVIDIA part delivers more than eight times the performance in this modern DirectX 12 workload. This is the largest gap in the head-to-head set and reflects the fundamental difference in raw graphics throughput between a discrete desktop GPU and an integrated processor.

The Geekbench OpenCL test tells a similar story, though with a slightly smaller margin. The RTX 3070 records 112821 points, while the Radeon 680M manages 23468. The delta here is 380.7%, meaning the RTX 3070 is roughly 4.8 times faster in this compute-oriented benchmark. OpenCL tends to scale heavily with shading unit count and memory bandwidth, both of which heavily favor the discrete card.

The one bright spot for the AMD Radeon 680M is the Geekbench Vulkan test. Here, the AMD part scores 21965, slightly edging out the RTX 3070's 21022. The delta is -4.3% from the NVIDIA perspective, meaning the Radeon 680M leads by just over four percent. This is a notable result, as it shows the integrated GPU can keep pace with, and even surpass, a high-end discrete card in a specific API scenario. It likely reflects the RDNA 2.0 architecture's efficient Vulkan driver overhead and the 680M's relatively high boost clock of 2200 MHz.

Looking at the broader database averages, the RTX 3070 holds an average benchmark score of 17208 across all recorded tests, while the Radeon 680M averages 15270. That puts the NVIDIA card roughly 12.7% ahead in aggregate performance. The RTX 3070 sits at the 61st percentile among all GPUs, while the Radeon 680M is at the 57th percentile. Both are mid-pack performers in the grand scheme, but the RTX 3070's percentile ranking reflects its stronger overall standing.

It is worth remembering the RTX 3070 has a much larger benchmark footprint in the database. It has recorded scores across ten tests, including PassMark DirectX 9, 10, 11, and 12, plus G2D, G3D, and GPU Compute. The Radeon 680M only has three recorded benchmarks. This difference in test coverage means the average score for the RTX 3070 is drawn from a wider variety of workloads, which can affect how directly comparable the aggregate numbers are.

FAQ

Q: Which GPU wins in the 3DMark Steel Nomad DX12 test?

A: The NVIDIA GeForce RTX 3070 wins decisively with a score of 3162 against 378 for the AMD Radeon 680M, a delta of 736.5%.

Q: Is there any test where the AMD Radeon 680M beats the RTX 3070?

A: Yes. In the Geekbench Vulkan test, the Radeon 680M scores 21965, while the RTX 3070 scores 21022. That gives the AMD part a 4.3% lead.

Q: How do their average benchmark scores compare?

A: The RTX 3070 has an average benchmark score of 17208, while the Radeon 680M averages 15270. The RTX 3070 sits at the 61st percentile of all GPUs, and the Radeon 680M sits at the 57th percentile.

Q: What is the closest rival to the RTX 3070 in the database?

A: The AMD Radeon RX 7600 XT is the nearest rival, with an average score of 17083, which is 0.7% lower than the RTX 3070's average.

Q: What is the closest rival to the Radeon 680M in the database?

A: The NVIDIA GeForce RTX 3050 OEM is the nearest rival, with an average score of 15199, which is 0.5% higher than the Radeon 680M's average.

Q: How many shading units does each GPU have?

A: The RTX 3070 has 5888 shading units, while the Radeon 680M has 768. This is a 7.7x difference in favor of the NVIDIA card.

Architecture Differences

The two GPUs represent fundamentally different design philosophies and manufacturing approaches. The NVIDIA GeForce RTX 3070 is a discrete, add-in-board graphics card built on the GA104 chip using the Ampere architecture. It was released on August 31, 2020, and is now marked as end-of-life. The AMD Radeon 680M is an integrated graphics processor, part of the Rembrandt+ chip using the RDNA 2.0 architecture. It is categorized as a Navi II IGP for Rembrandt Mobile and remains in active production, with a release date of January 2, 2023.

The manufacturing process differs significantly. The RTX 3070 uses an 8 nm process from Samsung, while the Radeon 680M uses a 6 nm process from TSMC. This gives the AMD chip a higher transistor density: 63.0 million transistors per square millimeter versus 44.4 million for the NVIDIA chip. However, the raw transistor counts and die sizes tell a more complex story. The GA104 chip packs 17,400 million transistors on a 392 mm² die, while the Rembrandt+ chip contains 13,100 million transistors on a 208 mm² die. The NVIDIA chip is both larger in absolute terms and has more total transistors, but the AMD chip is more densely packed.

Clock speeds are another differentiator. The RTX 3070 has a base clock of 1500 MHz and a boost clock of 1725 MHz. The Radeon 680M runs at a 2000 MHz base and 2200 MHz boost, which is considerably higher. This partially compensates for the AMD part's smaller execution resources, though not enough to close the performance gap in most tests.

The memory subsystems are fundamentally different. The RTX 3070 uses 8 GB of dedicated GDDR6 memory on a 256-bit bus, delivering 448.0 GB/s of bandwidth. The Radeon 680M uses system shared memory, with a bus width and bandwidth that are system dependent. This is a critical architectural distinction: the discrete card has its own high-speed memory pool, while the integrated GPU must share system memory, which is typically slower and subject to contention with the CPU.

The compute resources differ sharply. The RTX 3070 has 5888 shading units, 184 texture mapping units, and 96 raster output units. The Radeon 680M has 768 shading units, 48 TMUs, and 32 ROPs. The NVIDIA card also features 46 ray tracing cores and 184 tensor cores, while the AMD part has 12 ray tracing cores and no tensor cores. These differences manifest in the pixel and texture rates: the RTX 3070 achieves 165.6 GPixel/s and 317.4 GTexel/s, compared to 70.40 GPixel/s and 105.6 GTexel/s for the Radeon 680M.

FP32 and FP16 throughput also diverge. The RTX 3070 delivers 20.31 TFLOPS in both FP32 and FP16, with a 1:1 ratio. The Radeon 680M provides 3.379 TFLOPS in FP32 and 6.758 TFLOPS in FP16, using a 2:1 ratio. This means the NVIDIA card has roughly 6 times the FP32 throughput and 3 times the FP16 throughput of the AMD part.

The Verdict

The data points to a clear conclusion for users who need maximum graphics performance: the NVIDIA GeForce RTX 3070 is the superior choice. It wins the two most demanding benchmark tests by massive margins, holds a higher average score, and sits at a higher percentile ranking. The 736.5% lead in 3DMark Steel Nomad DX12 and the 380.7% lead in Geekbench OpenCL are not close calls. For gaming, content creation, or compute workloads that leverage DirectX 12 or OpenCL, the RTX 3070 is the obvious pick from this dataset.

The AMD Radeon 680M is not without merit, however. Its win in Geekbench Vulkan, albeit by a modest 4.3%, shows that the RDNA 2.0 architecture has strengths in certain API environments. Its higher clock speeds and more modern 6 nm process also suggest efficiency advantages, though the database does not provide direct power efficiency comparisons beyond the TDP figures. The Radeon 680M's 50 W TDP versus the RTX 3070's 220 W TDP indicates a significantly lower power draw, but the performance trade-off is steep.

For users constrained to a portable device or an integrated graphics solution, the Radeon 680M is the only option in this pairing. It is an active product with a 2023 release date, whereas the RTX 3070 is end-of-life. But for anyone building a desktop system with a dedicated graphics slot, the RTX 3070 offers a level of performance that the integrated AMD part cannot approach in the majority of recorded workloads.

The choice ultimately depends on the use case. If the priority is raw benchmark performance, the RTX 3070 wins decisively. If the priority is a low-power, integrated solution for a laptop or compact system, the Radeon 680M is the relevant product, and its Vulkan showing provides a point of interest for API-specific workloads.

Specification Differences

| Specification | NVIDIA GeForce RTX 3070 | AMD Radeon 680M |

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

| Architecture | Ampere | RDNA 2.0 |

| Process Node | 8 nm | 6 nm |

| Foundry | Samsung | TSMC |

| Transistors | 17,400 million | 13,100 million |

| Die Size | 392 mm² | 208 mm² |

| Transistor Density | 44.4M / mm² | 63.0M / mm² |

| Base Clock | 1500 MHz | 2000 MHz |

| Boost Clock | 1725 MHz | 2200 MHz |

| Memory Size | 8 GB | System Shared |

| Memory Type | GDDR6 | System Shared |

| Memory Bus Width | 256 bit | System Shared |

| Memory Bandwidth | 448.0 GB/s | System Dependent |

| Shading Units | 5888 | 768 |

| TMUs | 184 | 48 |

| ROPs | 96 | 32 |

| RT Cores | 46 | 12 |

| Tensor Cores | 184 | None |

| Pixel Rate | 165.6 GPixel/s | 70.40 GPixel/s |

| Texture Rate | 317.4 GTexel/s | 105.6 GTexel/s |

| FP32 Performance | 20.31 TFLOPS | 3.379 TFLOPS |

| FP16 Performance | 20.31 TFLOPS (1:1) | 6.758 TFLOPS (2:1) |

| TDP | 220 W | 50 W |

| Slot Width | Dual-slot | IGP |

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

| Suggested PSU | 550 W | Not specified |

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

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

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

| OpenGL Support | 4.6 | 4.6 |

| Vulkan Support | 1.4 | 1.4 |

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

| Release Date | 2020-08-31 | 2023-01-02 |

| Predecessor | GeForce 20 | Vega II IGP |

| Successor | GeForce 40 | Navi III IGP |

| Launch MSRP | 499 USD | Not available |