AMD Radeon 680M vs AMD Radeon 780M Comparison

Where Each One Wins

The benchmark data splits these two integrated graphics parts into distinct personalities. The AMD Radeon 780M takes the overall win count, two out of three recorded tests, but the AMD Radeon 680M has a clear domain of its own.

The 780M dominates in synthetic DirectX 12 workloads. In the 3DMark Steel Nomad test, it scores 480 against the 680M's 378, a 27% advantage. That is the kind of margin that separates playable frame rates from slideshows in modern titles. The same part also wins decisively in the Vulkan API test, posting 33,683 versus 21,965, a commanding 53.3% lead. These are the tests that stress raw geometry throughput, shader execution, and API-level efficiency, areas where the newer architecture shows its strength.

The 680M, however, wins the OpenCL compute test outright. It scores 23,468 against the 780M's 18,602, a 20.7% advantage. OpenCL workloads often behave differently from gaming-oriented APIs, and this result suggests the 680M retains an edge in certain general-purpose compute tasks, possibly due to driver optimization maturity or memory access patterns that favor its higher base clock. For users running OpenCL-accelerated applications, the older part is the better choice.

The average benchmark score across all recorded tests also favors the 780M, which lands at 17,588 versus the 680M's 15,270. The 780M sits at the 61st percentile of all GPUs in the database, while the 680M sits at the 57th percentile. The 780M's nearest rivals in the database are the AMD Radeon Pro 560 (average score 17,551, a 0.2% gap) and the NVIDIA GeForce RTX 4060 (17,639, a 0.3% gap), meaning it trades blows with those discrete parts. The 680M, by contrast, sits alongside the NVIDIA GeForce GTX 580 (15,283, a 0.1% gap) and the RTX 2060 (15,290, also a 0.1% gap), placing it in a lower performance tier overall.

Architecture Differences

The two iGPUs come from different generations of AMD's graphics roadmap. The 780M is built on the RDNA 3.0 architecture, manufactured on a 4 nm process at TSMC, while the 680M uses RDNA 2.0 on a 6 nm process, also at TSMC. That process shrink is significant, allowing the 780M to pack 25,390 million transistors into a 178 mm² die, versus 13,100 million transistors on a 208 mm² die for the 680M. Transistor density tells the story plainly: the 780M achieves 142.6 million transistors per square millimeter, more than double the 680M's 63.0 million.

Both parts share the same core configuration on paper: 768 shading units, 48 texture mapping units, 32 raster output units, and 12 ray tracing cores. The architectural generation difference, however, changes how those resources are used. The 780M's RDNA 3.0 design processes FP32 and FP16 at a 1:1 ratio, delivering 8.909 TFLOPS for both precision formats. The 680M's RDNA 2.0 design uses a 2:1 ratio for FP16, meaning it delivers 3.379 TFLOPS for FP32 but 6.758 TFLOPS for FP16. The 780M's FP32 throughput is roughly 2.6 times higher, a massive gap for shader-heavy workloads.

Clock behavior also differs sharply. The 780M runs a base clock of 800 MHz and boosts to 2900 MHz, a wide dynamic range that suggests aggressive power management. The 680M runs a base of 2000 MHz and boosts to just 2200 MHz, a much narrower window. Despite the 780M's lower base, its boost clock is 700 MHz higher, which combined with the architectural improvements explains its rasterization advantage: 92.80 GPixel/s versus 70.40 GPixel/s, and 139.2 GTexel/s versus 105.6 GTexel/s.

The TDP ratings are counterintuitive. The 780M is rated at 15 W, while the 680M is rated at 50 W. This means the newer part delivers dramatically higher performance at one-third the power envelope, evidence of the efficiency gains from the 4 nm process. Both are integrated graphics with no power connectors, using system-shared memory with system-dependent bandwidth. The bus interface is PCIe 4.0 x8 for both. The 780M belongs to the Navi III IGP generation and the 680M to the Navi II IGP generation.

FAQ

Q: Which iGPU has the higher average benchmark score?

A: The AMD Radeon 780M, with an average score of 17,588 across all recorded tests, versus 15,270 for the AMD Radeon 680M.

Q: How does the 780M compare to the NVIDIA GeForce RTX 4060 in the database?

A: The 780M is effectively tied with the RTX 4060. The RTX 4060 has an average score of 17,639, which is 0.3% higher, a negligible difference in real-world terms.

Q: Does the 680M win any benchmark against the 780M?

A: Yes. The 680M wins the Geekbench OpenCL test, scoring 23,468 versus the 780M's 18,602, a 20.7% advantage for the older part.

Q: What is the transistor density difference between the two?

A: The 780M has a transistor density of 142.6 million per square millimeter, while the 680M has 63.0 million per square millimeter. The 780M's density is more than double.

Q: Are both GPUs capable of ray tracing?

A: Yes. Both the 780M and the 680M feature 12 ray tracing cores and support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: Which part has the higher boost clock?

A: The 780M boosts to 2900 MHz, while the 680M boosts to 2200 MHz. The 780M's boost clock is 700 MHz higher.

Specification Differences

| Specification | AMD Radeon 780M | AMD Radeon 680M |

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

| Architecture | RDNA 3.0 | RDNA 2.0 |

| Process Node | 4 nm | 6 nm |

| Transistors | 25,390 million | 13,100 million |

| Die Size | 178 mm² | 208 mm² |

| Transistor Density | 142.6M / mm² | 63.0M / mm² |

| Base Clock | 800 MHz | 2000 MHz |

| Boost Clock | 2900 MHz | 2200 MHz |

| Pixel Rate | 92.80 GPixel/s | 70.40 GPixel/s |

| Texture Rate | 139.2 GTexel/s | 105.6 GTexel/s |

| FP32 Performance | 8.909 TFLOPS | 3.379 TFLOPS |

| FP16 Performance | 8.909 TFLOPS (1:1) | 6.758 TFLOPS (2:1) |

| TDP | 15 W | 50 W |

| Generation | Navi III IGP (Phoenix) | Navi II IGP (Rembrandt Mobile) |

| Release Date | 2024-01-30 | 2023-01-02 |

Head-to-Head Benchmarks

The recorded data includes three direct comparisons, and the margins vary wildly depending on the test.

3DMark Steel Nomad (DX12): The 780M wins decisively with a score of 480 against 378. That is a 27% improvement. This test is a strong indicator of modern DirectX 12 gaming performance, and the gap here is large enough to classify the 780M as a meaningfully faster gaming part. The 680M's score places it closer to the NVIDIA GeForce GTX 580 (15,283 average, a 0.1% gap) in the database's overall rankings, while the 780M's score aligns it with the RTX 4060 tier.

Geekbench OpenCL: The 680M takes this one, and it is not close. A score of 23,468 versus 18,602 gives the 680M a 20.7% lead. This is the only recorded test where the older architecture wins. OpenCL is often used for compute tasks like video encoding, physics simulation, and some productivity applications. The result suggests that the 680M's higher base clock (2000 MHz versus 800 MHz) and potentially different memory controller behavior give it an edge in certain compute patterns that do not scale with the 780M's wider execution width.

Geekbench Vulkan: The 780M reclaims the lead with a massive 53.3% margin, scoring 33,683 against 21,965. This is the largest delta in the entire head-to-head set. Vulkan is the other major modern graphics API, and this result confirms that the 780M's architectural improvements extend beyond DirectX 12. The fact that the 780M nearly doubles the 680M's Vulkan score while using one-third the TDP (15 W versus 50 W) highlights the efficiency of the 4 nm RDNA 3.0 design.

Across all three tests, the 780M's wins are larger in aggregate than the 680M's single win. The 27% and 53.3% victories for the 780M outweigh the 20.7% win for the 680M, and the average benchmark score reflects that: 17,588 versus 15,270, a 15.2% overall advantage for the 780M. The percentile ranking confirms the separation, with the 780M at the 61st percentile and the 680M at the 57th.

For gaming and modern API workloads, the data points clearly to the 780M. For OpenCL compute specifically, the 680M remains a viable alternative, and its lower average score is not the whole story. The 680M's nearest rival, the AMD Radeon RX 7600, sits at 15,171 with a 0.7% gap, meaning the 680M is competitive with that discrete card in the database's overall metric. The 780M, however, reaches into the territory of the RTX 4060, a far more powerful discrete option. The architectural generational leap is the deciding factor, and the numbers show it across every gaming-oriented benchmark.