AMD Radeon HD 8690M vs NVIDIA Quadro K620 Comparison

The NVIDIA Quadro K620 and AMD Radeon HD 8690M are both end-of-life mobile and professional graphics solutions from the 2013-2014 era, built on a 28 nm process at TSMC. The data shows a narrow overall victory for the Quadro K620, which wins the sole head-to-head benchmark by a meaningful margin, yet both cards occupy the same performance percentile, making their real-world positioning surprisingly close.

Where Each One Wins

The NVIDIA Quadro K620 wins the only benchmark where both cards appear in the FACT PACK: Geekbench OpenCL. That is a decisive result for compute workloads, as the K620 scores 6693 versus 6137 for the AMD Radeon HD 8690M, a 9.1% advantage. This makes the Quadro the stronger choice for OpenCL-based tasks such as general-purpose GPU compute, video encoding acceleration, and any professional application that leverages this API. The K620’s average benchmark score of 6282 also slightly exceeds the AMD’s 6137, reinforcing its overall compute lead.

The AMD Radeon HD 8690M has no benchmark wins in the head-to-head comparison. However, it is not without its own strengths. The data shows the Radeon carries a 64-bit memory bus with GDDR5 memory, which delivers 32.00 GB/s of bandwidth—higher than the Quadro’s 28.80 GB/s despite the latter’s wider 128-bit bus with slower DDR3 memory. This suggests the AMD part may have an edge in bandwidth-bound scenarios, such as texture-heavy workloads or higher-resolution frame buffers, even though the benchmark results do not capture that advantage. Additionally, the Radeon supports a newer DirectX feature level (12_1 versus 12_0 for the Quadro), which could benefit certain gaming or DirectCompute applications that use those features.

Architecture Differences

The two GPUs come from fundamentally different architectures. The NVIDIA Quadro K620 is built on the Maxwell architecture (chip GM107), while the AMD Radeon HD 8690M uses GCN 1.0 (chip Sun). This architectural split drives most of their performance and feature differences.

Transistor counts and die sizes diverge sharply. The Quadro packs 1,870 million transistors on a 148 mm² die, yielding a transistor density of 12.6M / mm². The Radeon is much smaller, with 690 million transistors on a 56 mm² die, giving a density of 12.3M / mm². The Quadro’s larger die allows for more execution resources, but the Radeon’s smaller footprint suggests it was designed for lower-power mobile integration.

Core configurations differ as well. The Quadro has 384 shading units, 24 texture mapping units, and 16 ROPs. The Radeon has fewer of each: 320 shading units, 20 TMUs, and only 8 ROPs. The Quadro also clocks higher, with a base of 1058 MHz and boost of 1124 MHz, versus the Radeon’s 925 MHz base and 975 MHz boost. This combination of more cores and higher clocks gives the NVIDIA part a significant raw compute advantage, reflected in its pixel rate of 17.98 GPixel/s and texture rate of 26.98 GTexel/s, compared to the Radeon’s 7.800 GPixel/s and 19.50 GTexel/s.

Memory subsystems also differ architecturally. The Quadro uses 2 GB of DDR3 on a 128-bit bus, while the Radeon uses 2 GB of GDDR5 on a 64-bit bus. Despite the narrower bus, the GDDR5’s higher effective speed of 4 Gbps (versus 1800 Mbps for the Quadro) yields higher bandwidth for the AMD part: 32.00 GB/s versus 28.80 GB/s. This is a notable inversion where the smaller-bus card wins on bandwidth.

API support shows further differences. The Quadro supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4. The Radeon supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The AMD card has a slightly higher DirectX feature level, while the NVIDIA card has a newer Vulkan version.

Head-to-Head Benchmarks

The only direct comparison available is Geekbench OpenCL, where the NVIDIA Quadro K620 scores 6693 against the AMD Radeon HD 8690M’s 6137. That is a 9.1% delta in favor of the Quadro. In practical terms, this means the Quadro completes OpenCL workloads roughly 9% faster, which can translate to noticeable reductions in render times or compute jobs.

Looking at the broader rivalry context, both cards sit near a cluster of similarly performing GPUs. The Quadro’s nearest rivals include the NVIDIA GeForce RTX 5070 Ti SUPER (avg score 6270, delta 0.2%), the NVIDIA GeForce RTX 4070 Ti SUPER AD102 (avg score 6270, delta 0.2%), the AMD Radeon R7 M350 (avg score 6327, delta -0.7%), and the AMD Radeon Pro WX 4100 (avg score 6330, delta -0.8%). The Quadro’s average score of 6282 places it essentially at parity with these parts, within a narrow 0.2% to 0.8% band.

The Radeon HD 8690M’s nearest rivals show a similar pattern. The Intel Iris Pro Graphics 6200 (avg score 6117, delta 0.3%), the NVIDIA RTX A400 (avg score 6078, delta 1%), and the NVIDIA GeForce MX230 (avg score 6077, delta 1%) all sit close to the Radeon’s average of 6137. The AMD Radeon HD 6950 is slightly faster at 6210 (delta -1.2%). This indicates the Radeon is competitive with a range of integrated and entry-level discrete GPUs, though it trails the Quadro in the direct comparison.

The single benchmark win for the Quadro is consistent with its higher core count, clock speeds, and fill rates. The Radeon’s higher memory bandwidth does not translate into a benchmark victory, at least in the OpenCL test available. This suggests that compute performance, as measured by Geekbench, is more sensitive to shader throughput and clock speed than to memory bandwidth, at least within this performance class.

Specification Differences

The two cards differ across almost every major specification category. The table below highlights only the fields where they do not match.

| Specification | NVIDIA Quadro K620 | AMD Radeon HD 8690M |

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

| Architecture | Maxwell | GCN 1.0 |

| Generation | Quadro Kepler (Kx200) | Solar System (HD 8600M) |

| Transistors | 1,870 million | 690 million |

| Die Size | 148 mm² | 56 mm² |

| Transistor Density | 12.6M / mm² | 12.3M / mm² |

| Base Clock | 1058 MHz | 925 MHz |

| Boost Clock | 1124 MHz | 975 MHz |

| Memory Clock | 900 MHz / 1800 Mbps effective | 1000 MHz / 4 Gbps effective |

| Memory Type | DDR3 | GDDR5 |

| Bus Width | 128 bit | 64 bit |

| Memory Bandwidth | 28.80 GB/s | 32.00 GB/s |

| Shading Units | 384 | 320 |

| TMUs | 24 | 20 |

| ROPs | 16 | 8 |

| Pixel Rate | 17.98 GPixel/s | 7.800 GPixel/s |

| Texture Rate | 26.98 GTexel/s | 19.50 GTexel/s |

| FP32 Performance | 863.2 GFLOPS | 624.0 GFLOPS |

| TDP | 45 W | N/A |

| Slot Width | Single-slot | N/A |

| Power Connectors | None | N/A |

| Suggested PSU | 200 W | N/A |

| Bus Interface | PCIe 2.0 x16 | PCIe 3.0 x8 |

| Display Outputs | 1x DVI, 1x DisplayPort 1.2 | N/A |

| DirectX Support | 12 (11_0) | 12 (11_1) |

| Vulkan Support | 1.4 | 1.2.170 |

| Dimensions | 160 mm (6.3 inches) long, 69 mm (2.7 inches) high | N/A |

| Release Date | 2014-07-21 | 2013-02-28 |

| Predecessor | Quadro Fermi | London |

| Successor | Quadro Maxwell | Gem System |

| Geekbench OpenCL Score | 6693 | 6137 |

| Average Benchmark Score | 6282 | 6137 |

Notable absences in the AMD data include TDP, slot width, power connectors, suggested PSU, display outputs, and physical dimensions, which are not listed in the FACT PACK. The Radeon also lacks a Geekbench Vulkan score, while the Quadro records a score of 5870 in that test.

FAQ

Q: Which GPU is faster in OpenCL compute?

A: The NVIDIA Quadro K620 is faster, scoring 6693 versus 6137 for the AMD Radeon HD 8690M, a 9.1% advantage.

Q: Do these GPUs have the same memory bandwidth?

A: No. The AMD Radeon HD 8690M has higher memory bandwidth at 32.00 GB/s, despite a narrower 64-bit bus, because it uses GDDR5 memory. The NVIDIA Quadro K620 has 28.80 GB/s from DDR3 on a 128-bit bus.

Q: Which GPU has more shading units?

A: The NVIDIA Quadro K620 has 384 shading units, while the AMD Radeon HD 8690M has 320.

Q: What is the transistor count difference?

A: The NVIDIA Quadro K620 contains 1,870 million transistors, compared to 690 million for the AMD Radeon HD 8690M.

Q: Do both GPUs support the same DirectX version?

A: Both support DirectX 12, but at different feature levels. The NVIDIA Quadro K620 supports 12 (11_0), while the AMD Radeon HD 8690M supports 12 (11_1), which is a slightly higher feature level.

Q: Which GPU has a higher pixel fill rate?

A: The NVIDIA Quadro K620 has a significantly higher pixel rate of 17.98 GPixel/s, compared to 7.800 GPixel/s for the AMD Radeon HD 8690M.

Q: Are these GPUs still in production?

A: No. Both are marked as end-of-life. The NVIDIA Quadro K620 was released on 2014-07-21, while the AMD Radeon HD 8690M was released earlier on 2013-02-28.