AMD Radeon HD 8790M vs NVIDIA Quadro K620M Comparison

The NVIDIA Quadro K620M and AMD Radeon HD 8790M are two end-of-life mobile workstation GPUs from different architectural generations, both built on TSMC’s 28 nm process. The data shows they occupy similar performance percentiles, but their design philosophies and benchmark results reveal distinct strengths. The Quadro K620M lands in the 34th percentile of all GPUs, while the Radeon HD 8790M sits just behind at the 33rd percentile, indicating they are closely matched overall but achieve their scores through very different means.

Head-to-Head Benchmarks

The only direct head-to-head result available is the Geekbench OpenCL test, and it is a decisive win for the NVIDIA Quadro K620M. NVIDIA scores 5,957 points against AMD’s 5,017 points, a delta of 18.7%. This is not a marginal victory; it is a substantial gap that places the Quadro nearly a fifth ahead of the Radeon in compute workloads that leverage OpenCL. The Radeon HD 8790M’s raw compute ceiling, measured at 691.2 GFLOPS FP32, is lower than the Quadro’s 863.2 GFLOPS, which directly explains the OpenCL outcome. Higher shading throughput translates into faster execution of parallel math, and the data confirms this.

However, the Radeon HD 8790M has its own benchmark card to play. It is the only GPU in this comparison with a Geekbench Vulkan score, hitting 6,365 points. This is notably higher than its own OpenCL score of 5,017, suggesting that AMD’s GCN architecture responds well to Vulkan’s explicit API model. The Quadro K620M has no Vulkan benchmark score in the data, so a direct comparison there is impossible, but the Radeon’s Vulkan result is 21% better than its OpenCL result, hinting at untapped potential in modern graphics APIs. For perspective, the Quadro’s nearest rivals in average score include the AMD Radeon HD 8730M at 5,955 (a 0% delta) and the Intel UHD Graphics 730 at 5,929 (a 0.5% delta for the Quadro). The Radeon HD 8790M’s nearest rival is the NVIDIA Quadro M500M at 5,604, a 1.6% delta in AMD’s favor, showing it holds its own against a newer Quadro part.

Architecture Differences

The architectural divide between these two GPUs is stark. NVIDIA uses the GM108S chip based on Maxwell architecture, while AMD employs the Mars chip built on GCN 1.0. Both are manufactured by TSMC on a 28 nm process and share the same 77 mm² die size, but NVIDIA packs more transistors into that space: 1,020 million versus AMD’s 950 million. This gives the Quadro a higher transistor density of 13.2M per mm² compared to AMD’s 12.3M per mm². The extra transistors help explain the Quadro’s higher clock speeds — 1,029 MHz base and 1,124 MHz boost versus AMD’s 850 MHz base and 900 MHz boost — which are critical to its FP32 performance lead.

Memory architecture is where AMD strikes back. The Radeon HD 8790M uses 2 GB of GDDR5 on a 128-bit bus, yielding 64.00 GB/s of bandwidth. The Quadro K620M also has 2 GB, but it is DDR3 on a 64-bit bus, delivering only 16.02 GB/s. That is a four-fold bandwidth advantage for AMD, which is significant for texture-heavy workloads and large data transfers. The Radeon also has more texture mapping units (24 TMUs versus 16), giving it a higher texture rate of 21.60 GTexel/s against the Quadro’s 17.98 GTexel/s. Both have 384 shading units and 8 ROPs, but the Quadro’s pixel rate is higher at 8.992 GPixel/s versus AMD’s 7.200 GPixel/s, thanks to its clock advantage. API support differs subtly: both support DirectX 12 and OpenGL 4.6, but the Quadro lists Vulkan 1.4 while the Radeon lists Vulkan 1.2.170; DirectX feature levels also differ, with NVIDIA at 11_0 and AMD at 11_1.

FAQ

Q: Which GPU has a higher clock speed?

A: The NVIDIA Quadro K620M runs at a base clock of 1,029 MHz and a boost clock of 1,124 MHz. The AMD Radeon HD 8790M is slower, with a base of 850 MHz and a boost of 900 MHz.

Q: Does the AMD Radeon HD 8790M have better memory bandwidth?

A: Yes. The Radeon uses GDDR5 on a 128-bit bus, achieving 64.00 GB/s. The Quadro’s DDR3 on a 64-bit bus delivers only 16.02 GB/s, a quarter of AMD’s throughput.

Q: What is the average benchmark score for each GPU?

A: The Quadro K620M has an average benchmark score of 5,957, while the Radeon HD 8790M averages 5,691. The Quadro’s average is higher, but the Radeon has a Vulkan score of 6,365 that exceeds both averages.

Q: Which GPU has a higher transistor count?

A: The NVIDIA Quadro K620M contains 1,020 million transistors, versus 950 million in the AMD Radeon HD 8790M. Both are on a 77 mm² die, so the Quadro has higher density.

Q: Are these GPUs still in production?

A: No. Both are listed as end-of-life. The Quadro K620M was released on 2015-02-28, while the Radeon HD 8790M is older, released on 2013-03-31.

Q: What is the DirectX support difference?

A: The Quadro K620M supports DirectX 12 (11_0), while the Radeon HD 8790M supports DirectX 12 (11_1). The Radeon has a slightly higher feature level.

Specification Differences

| Specification | NVIDIA Quadro K620M | AMD Radeon HD 8790M |

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

| Chip | GM108S | Mars |

| Architecture | Maxwell | GCN 1.0 |

| Generation | Quadro Kepler-M (Kx200M) | Solar System (HD 8700M) |

| Transistors | 1,020 million | 950 million |

| Transistor Density | 13.2M / mm² | 12.3M / mm² |

| Base Clock | 1,029 MHz | 850 MHz |

| Boost Clock | 1,124 MHz | 900 MHz |

| Memory Clock | 1001 MHz (2 Gbps effective) | 1000 MHz (4 Gbps effective) |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus Width | 64 bit | 128 bit |

| Memory Bandwidth | 16.02 GB/s | 64.00 GB/s |

| TMUs | 16 | 24 |

| Pixel Rate | 8.992 GPixel/s | 7.200 GPixel/s |

| Texture Rate | 17.98 GTexel/s | 21.60 GTexel/s |

| FP32 | 863.2 GFLOPS | 691.2 GFLOPS |

| TDP | 30 W | Not specified |

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

| Vulkan Version | 1.4 | 1.2.170 |

| Release Date | 2015-02-28 | 2013-03-31 |

| Predecessor | Quadro Fermi-M | London |

| Successor | Quadro Maxwell-M | Gem System |

| Geekbench OpenCL Score | 5,957 | 5,017 |

| Geekbench Vulkan Score | Not listed | 6,365 |

| Avg Benchmark Score | 5,957 | 5,691 |

| Percentile vs All GPUs | 34 | 33 |

Where Each One Wins

The NVIDIA Quadro K620M wins in compute-heavy OpenCL workloads. Its 18.7% lead in the head-to-head Geekbench OpenCL test is its clearest victory, driven by higher clocks and more transistors. It also wins on pixel throughput at 8.992 GPixel/s, which benefits fill-rate-limited tasks like anti-aliasing and simple framebuffer operations. For users running CUDA-style parallel math, the Quadro’s 863.2 GFLOPS FP32 output is the stronger number.

The AMD Radeon HD 8790M wins on memory bandwidth by a factor of four (64.00 GB/s versus 16.02 GB/s), which is crucial for large textures, frame buffers, or any workload that saturates the memory bus. Its 24 TMUs give it a higher texture rate of 21.60 GTexel/s, favoring anisotropic filtering and complex texture sampling. The Radeon also shows a Vulkan score of 6,365, suggesting it is the better choice for modern Vulkan-based applications, even though its OpenCL score lags. The data implies AMD’s part is built for bandwidth-hungry graphics, while NVIDIA’s is tuned for raw arithmetic speed.

The Verdict

The choice between these two GPUs depends entirely on the workload. Benchmark results indicate the NVIDIA Quadro K620M is the superior compute card, with a decisive 18.7% OpenCL win and higher FP32 throughput. Its 34th percentile ranking also edges out the Radeon’s 33rd percentile, reinforcing its slight overall performance lead. For professional applications that rely on OpenCL acceleration, the Quadro is the data-backed pick.

The AMD Radeon HD 8790M, however, should not be dismissed. Its 64.00 GB/s bandwidth is unmatched in this pairing, and the Vulkan score of 6,365 proves it can outperform its own OpenCL result significantly. For gaming or graphics work that prioritizes texture throughput and memory bandwidth over compute shaders, the Radeon holds the advantage. The Radeon’s release date is earlier (2013-03-31 versus 2015-02-28), but its GCN architecture handles newer APIs like Vulkan 1.2.170 effectively. Ultimately, the Quadro K620M wins on compute and pixel rate, while the Radeon HD 8790M wins on bandwidth and texture rate. The data does not declare a single champion; it defines two specialists.