AMD Radeon HD 8690M vs NVIDIA GeForce GTX 980M Comparison

FAQ

Q: Which GPU has the higher average benchmark score in the database?

A: The NVIDIA GeForce GTX 980M has an average benchmark score of 5308, while the AMD Radeon HD 8690M has an average score of 6137. The AMD part actually holds the higher average, though this is based on a single OpenCL test for the AMD card versus a broader suite for the NVIDIA card.

Q: How do the two GPUs compare in the Geekbench OpenCL test?

A: The NVIDIA GeForce GTX 980M scores 23832 in Geekbench OpenCL, while the AMD Radeon HD 8690M scores 6137. The NVIDIA part leads by 74.2% in this specific test, making it the clear winner in compute performance.

Q: What is the transistor count difference between the two chips?

A: The AMD Radeon HD 8690M uses the Sun chip with 690 million transistors on a 56 mm² die. The NVIDIA GeForce GTX 980M uses the GM204 chip with 5,200 million transistors on a 398 mm² die, representing a roughly 7.5 times higher transistor count.

Q: Which GPU has more memory bandwidth?

A: The NVIDIA GeForce GTX 980M has a 256-bit memory bus with 160.4 GB/s bandwidth, while the AMD Radeon HD 8690M has a 64-bit bus with 32.00 GB/s. The NVIDIA part offers five times the memory bandwidth of the AMD part.

Q: What is the DirectX API support difference between the two?

A: The AMD Radeon HD 8690M supports DirectX 12 (11_1), while the NVIDIA GeForce GTX 980M supports DirectX 12 (12_1). The NVIDIA card also supports Vulkan 1.4, whereas the AMD card supports Vulkan 1.2.170.

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

A: The AMD Radeon HD 8690M sits at the 36th percentile, while the NVIDIA GeForce GTX 980M sits at the 31st percentile. Despite the NVIDIA part's much higher raw scores in most tests, its broader benchmark suite pulls its percentile down relative to the AMD part's single-test result.

Architecture Differences

The AMD Radeon HD 8690M is built on the GCN 1.0 architecture using the Sun chip, manufactured by TSMC on a 28 nm process. It packs 690 million transistors onto a 56 mm² die, yielding a transistor density of 12.3M per mm². The GPU features 320 shading units, 20 texture mapping units, and 8 ROPs, with a pixel rate of 7.800 GPixel/s and a texture rate of 19.50 GTexel/s. Its FP32 compute performance is rated at 624.0 GFLOPS.

The NVIDIA GeForce GTX 980M uses the Maxwell 2.0 architecture with the GM204 chip, also built by TSMC on a 28 nm process. This is a much larger silicon: 5,200 million transistors on a 398 mm² die, with a transistor density of 13.1M per². The GTX 980M packs 1536 shading units, 96 TMUs, and 64 ROPs. Its pixel rate is 72.13 GPixel/s, texture rate is 108.2 GTexel/s, and FP32 compute reaches 3.462 TFLOPS.

The architectural gap is substantial. The NVIDIA part has nearly 8 times the shading units, 4.8 times the TMUs, and 8 times the ROPs of the AMD part. Clock speeds also favor NVIDIA, with a base of 1038 MHz and boost of 1127 MHz versus the AMD base of 925 MHz and boost of 975 MHz. The memory subsystems differ drastically: the AMD card uses 2 GB GDDR5 on a 64-bit bus at 4 Gbps effective, while the NVIDIA card uses 8 GB GDDR5 on a 256-bit bus at 5 Gbps effective, resulting in 32.00 GB/s versus 160.4 GB/s bandwidth.

API support also differs. The AMD GPU lists DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The NVIDIA GPU lists DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The NVIDIA part also has a newer Vulkan feature level. Form factor differs too: the AMD card uses a PCIe 3.0 x8 interface, while the NVIDIA card uses MXM-B (3.0) and is described as an MXM Module with no power connectors and portable-device-dependent display outputs.

The Verdict

Based solely on recorded data, the NVIDIA GeForce GTX 980M is the stronger GPU in almost every measurable specification. It outperforms the AMD Radeon HD 8690M in shading units, texture units, ROPs, memory capacity, memory bandwidth, clock speeds, pixel rate, texture rate, and FP32 compute. The 74.2% delta in the Geekbench OpenCL test confirms this is not a close contest in raw performance.

The AMD Radeon HD 8690M, however, has one advantage: its average benchmark score is 6137 versus 5308 for the NVIDIA part. This is driven by the AMD card's single Geekbench OpenCL score of 6137, which is higher than the NVIDIA card's average across multiple tests. But that average includes weaker scores like 649 in 3DMark Steel Nomad DX12, 35 in Passmark DirectX 10, and 31 in Passmark DirectX 12, which drag the NVIDIA average down.

The choice depends on the workload. For users who need pure compute throughput, the GTX 980M is the obvious pick. For users who only look at the database's average score and the 36th percentile ranking of the AMD part, they might choose the AMD card. But the specification sheet and the head-to-head test both show the NVIDIA card is faster. The AMD card is better suited for legacy or low-power applications, while the NVIDIA card is a dedicated high-end mobile solution.

Specification Differences

| Feature | AMD Radeon HD 8690M | NVIDIA GeForce GTX 980M |

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

| Architecture | GCN 1.0 | Maxwell 2.0 |

| Generation | Solar System (HD 8600M) | GeForce 900M |

| Process Node | 28 nm | 28 nm (same) |

| Transistors | 690 million | 5,200 million |

| Die Size | 56 mm² | 398 mm² |

| Transistor Density | 12.3M / mm² | 13.1M / mm² |

| Base Clock | 925 MHz | 1038 MHz |

| Boost Clock | 975 MHz | 1127 MHz |

| Memory Clock | 1000 MHz / 4 Gbps effective | 1253 MHz / 5 Gbps effective |

| Memory Size | 2 GB | 8 GB |

| Memory Type | GDDR5 | GDDR5 (same) |

| Memory Bus | 64 bit | 256 bit |

| Memory Bandwidth | 32.00 GB/s | 160.4 GB/s |

| Shading Units | 320 | 1536 |

| TMUs | 20 | 96 |

| ROPs | 8 | 64 |

| Pixel Rate | 7.800 GPixel/s | 72.13 GPixel/s |

| Texture Rate | 19.50 GTexel/s | 108.2 GTexel/s |

| FP32 | 624.0 GFLOPS | 3.462 TFLOPS |

| Bus Interface | PCIe 3.0 x8 | MXM-B (3.0) |

| DirectX | 12 (11_1) | 12 (12_1) |

| Vulkan | 1.2.170 | 1.4 |

| Slot Width | Not specified | MXM Module |

| Power Connectors | Not specified | None |

| Display Outputs | Not specified | Portable Device Dependent |

| Release Date | 2013-02-28 | 2014-10-06 |

| Predecessor | London | GeForce 800M |

| Successor | Gem System | GeForce 10 Mobile |

Head-to-Head Benchmarks

The database contains one head-to-head test between these two GPUs. In the Geekbench OpenCL test, the NVIDIA GeForce GTX 980M scores 23832, while the AMD Radeon HD 8690M scores 6137. The delta is 74.2% in favor of the NVIDIA card. This is a massive margin, and it is consistent with the specification differences: the NVIDIA card has roughly 5.5 times the shading units, 5 times the memory bandwidth, and 5.5 times the FP32 throughput.

Looking at the NVIDIA card's full benchmark suite, it shows a wide range of scores: 649 in 3DMark Steel Nomad DX12, 17703 in Geekbench Vulkan, 125 in Passmark DirectX 9, 57 in Passmark DirectX 11, 35 in Passmark DirectX 10, 31 in Passmark DirectX 12, 490 in Passmark G2D, 7338 in Passmark G3D, and 2816 in Passmark GPU Compute. The AMD card has no corresponding entries for these tests, so no direct comparison is possible. But the Geekbench OpenCL result alone is enough to establish the NVIDIA card's dominance.

The AMD card's nearest rivals in the database are the Intel Iris Pro Graphics 6200 (average score 6117, delta 0.3%), the NVIDIA RTX A400 (6078, delta 1%), the NVIDIA GeForce MX230 (6077, delta 1%), and the AMD Radeon HD 6950 (6210, delta 1.2%). This places the AMD card in a cluster of low-to-midrange GPUs. The NVIDIA card's nearest rivals are the NVIDIA GeForce 930A (5317, delta 0.2%), the NVIDIA GeForce 840M (5322, delta 0.3%), the NVIDIA GeForce 940M (5284, delta 0.5%), and the AMD Radeon R7 M445 (5358, delta 0.9%). The GTX 980M sits slightly below these in average score, but with a much higher peak in OpenCL.

Where Each One Wins

AMD Radeon HD 8690M wins on: The database's average benchmark score. The AMD card's 6137 average is higher than the NVIDIA card's 5308, a difference of about 13%. It also holds a higher percentile position, sitting at the 36th percentile versus the 31st percentile. For users who weight the database's aggregate scoring, the AMD part appears better, but this is a single-test artifact. The AMD card is also built on a smaller die (56 mm² vs 398 mm²), which may imply lower thermal load, though the database does not list TDP for either card. Its PCIe 3.0 x8 interface is more conventional for older laptops than the MXM-B form factor of the NVIDIA card.

NVIDIA GeForce GTX 980M wins in every direct performance comparison. The Geekbench OpenCL score shows a 74.2% lead. The specification sheet shows it has more shading units (1536 vs 320), more TMUs (96 vs 20), more ROPs (64 vs 8), more memory (8 GB vs 2 GB), a wider memory bus (256 bit vs 64 bit), and higher bandwidth (160.4 GB/s vs 32.00 GB/s). Its pixel rate is 72.13 GPixel/s versus 7.800 GPixel/s, texture rate is 108.2 GTexel/s versus 19.50 GTexel/s, and FP32 is 3.462 TFLOPS versus 624.0 GFLOPS. The NVIDIA card also supports higher API versions: DirectX 12 (12_1) versus 12 (11_1), and Vulkan 1.4 versus 1.2.170. The NVIDIA card's Vulkan score of 17703 is also strong, though the AMD card has no Vulkan test.

In summary, the NVIDIA GeForce GTX 980M is the clear choice for any high-performance workload, especially compute-heavy tasks like OpenCL. The AMD Radeon HD 8690M is a lower-tier part that only wins in the average score metric, which is a statistical quirk rather than a real performance advantage. For users with the latest drivers and workloads it supports, the NVIDIA card is the only one to consider.