NVIDIA Quadro K620 vs NVIDIA RTX A400 Comparison

The NVIDIA Quadro K620 and NVIDIA RTX A400 represent two distinct eras of workstation graphics. The K620 is an end-of-life entry from 2014, built on the Maxwell architecture, while the RTX A400 is a current-generation Ampere card released in 2024. Benchmark data shows a decisive generational shift, with the RTX A400 winning both head-to-head tests. The data indicates that the RTX A400 delivers substantially higher raw performance, but the K620 retains relevance in specific legacy or compatibility-focused scenarios.

Where Each One Wins

The benchmark results create a clear dichotomy. The RTX A400 is the outright performance winner in every measured GPU compute and graphics API test, making it the logical choice for modern workloads that demand raw throughput. Its wins in Geekbench OpenCL and Vulkan are not marginal; they are multi-fold improvements that position it as a capable card for contemporary compute tasks and API-level graphics rendering.

The Quadro K620, despite its losses, shows its strength in a different domain: compatibility and legacy support. Its generation is listed as "Quadro Kepler (Kx200)" with a predecessor of "Quadro Fermi," indicating a long lineage of driver maturity for older software stacks. For environments running legacy CAD or visualization applications that rely on fixed-function pipelines or older OpenGL paths, the K620’s architecture may be more predictable. However, this is a qualitative advantage, as the quantitative data shows it cannot compete on speed.

In terms of market positioning, the RTX A400’s average benchmark score of 6078 places it near the NVIDIA GeForce MX230 (6077) and Quadro P2000 (6049), while the K620’s average of 6282 places it near the AMD Radeon Pro WX 4100 (6330). This is counterintuitive given the K620’s lower raw scores, but it reflects the different benchmark suites used to calculate these averages. The RTX A400’s percentile rank of 35 versus the K620’s 36 shows they sit nearly identically in the overall GPU hierarchy, despite the A400’s superiority in the specific tests listed.

Architecture Differences

The architectural gap between these two cards is vast, spanning a decade of process and design evolution. The K620 uses the GM107 chip on a 28 nm process at TSMC, containing 1,870 million transistors on a 148 mm² die. The RTX A400 uses the GA107 chip fabricated by Samsung on an 8 nm process, with 8,700 million transistors on a 200 mm² die. This translates to a transistor density of 12.6M per mm² for the K620 versus 43.5M per mm² for the RTX A400, explaining the massive computational leap.

The compute configurations differ significantly. The K620 has 384 shading units, 24 texture mapping units (TMUs), and 16 render output units (ROPs). The RTX A400 doubles the shading units to 768 while keeping the same 24 TMUs and 16 ROPs. More critically, the RTX A400 introduces dedicated hardware that the K620 lacks entirely: 6 ray tracing cores and 24 tensor cores. This makes the RTX A400 compliant with DirectX 12 Ultimate (12_2), whereas the K620 is limited to DirectX 12 (11_0). Both cards support OpenGL 4.6 and Vulkan 1.4, but the A400’s feature set is fundamentally newer.

Memory architecture also tells a story of progress. The K620 uses 2 GB of DDR3 on a 128-bit bus, yielding 28.80 GB/s of bandwidth. The RTX A400 uses 4 GB of GDDR6 on a 64-bit bus, achieving 96.00 GB/s. The A400’s narrower bus is more than compensated by the faster memory type, tripling available bandwidth. Clock speeds are higher on the A400 as well, with a base of 1417 MHz and boost of 1762 MHz versus the K620’s 1058 MHz base and 1124 MHz boost. The memory clock on the A400 runs at 1500 MHz (12 Gbps effective) versus 900 MHz (1800 Mbps effective) on the K620.

Head-to-Head Benchmarks

The most striking result is in Geekbench OpenCL, where the RTX A400 scores 22844 against the K620’s 6693. This represents a deltaPct of -70.7% for the K620, meaning the A400 is roughly 3.4 times faster. This test is a general compute benchmark that stresses raw FP32 throughput, and the numbers reflect the architectural advantage. The K620’s FP32 performance is listed at 863.2 GFLOPS, while the RTX A400 delivers 2.706 TFLOPS, a 3.1x increase that aligns closely with the benchmark delta.

The Geekbench Vulkan test shows a similar gap. The RTX A400 scores 22237 versus the K620’s 5870, a deltaPct of -73.6%. Vulkan is a low-overhead graphics API that benefits from modern hardware features like asynchronous compute and better driver scheduling. The A400’s support for DirectX 12 Ultimate and its newer shader model likely contribute to this performance, as does its higher texture rate of 42.29 GTexel/s versus 26.98 GTexel/s on the K620. The pixel rate also favors the A400 at 28.19 GPixel/s versus 17.98 GPixel/s.

The RTX A400 also has additional benchmark data from PassMark that the K620 lacks, including DirectX 9 (87), DirectX 10 (32), DirectX 11 (37), DirectX 12 (27), G2D (899), G3D (5983), and GPU compute (2557). These scores are not directly comparable to the K620 since no equivalent data is provided, but they illustrate the A400’s breadth of testing coverage. The G3D score of 5983 is notably higher than the average benchmark score of 6078, suggesting the card performs consistently across different workloads.

FAQ

Q: Which card has better raw compute performance?

A: The NVIDIA RTX A400 is decisively faster. In Geekbench OpenCL, it scores 22844 versus the Quadro K620’s 6693, a delta of -70.7% for the K620. Its FP32 throughput is 2.706 TFLOPS compared to 863.2 GFLOPS for the K620.

Q: Can the Quadro K620 handle modern graphics APIs?

A: It supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4. However, it lacks the hardware features for DirectX 12 Ultimate, which the RTX A400 supports. The K620’s Vulkan score of 5870 is far below the A400’s 22237.

Q: What is the memory capacity and bandwidth difference?

A: The K620 has 2 GB of DDR3 memory on a 128-bit bus with 28.80 GB/s bandwidth. The RTX A400 has 4 GB of GDDR6 on a 64-bit bus with 96.00 GB/s bandwidth, providing over three times the bandwidth.

Q: Does the RTX A400 support ray tracing?

A: Yes, it includes 6 dedicated ray tracing cores. The Quadro K620 has no ray tracing cores. This also enables the A400 to meet DirectX 12 Ultimate (12_2) requirements.

Q: How do their average benchmark scores compare?

A: The RTX A400 has an average benchmark score of 6078, while the Quadro K620 has an average of 6282. Despite the K620’s higher average, the A400 wins all direct head-to-head tests, indicating the averages are skewed by different test suites.

Q: What are the power requirements?

A: The K620 has a TDP of 45 W and a suggested PSU of 200 W. The RTX A400 has a TDP of 50 W and a suggested PSU of 250 W. Both are single-slot cards with no power connectors.

Specification Differences

The following table highlights only the fields where the two cards differ, omitting identical specifications like manufacturer (NVIDIA), TMUs (24), ROPs (16), slot width (single-slot), and power connectors (none).

| Specification | NVIDIA Quadro K620 | NVIDIA RTX A400 |

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

| Chip | GM107 | GA107 |

| Architecture | Maxwell | Ampere |

| Generation | Quadro Kepler (Kx200) | Workstation Ampere (Ax000) |

| Process Node | 28 nm (TSMC) | 8 nm (Samsung) |

| Transistors | 1,870 million | 8,700 million |

| Die Size | 148 mm² | 200 mm² |

| Transistor Density | 12.6M / mm² | 43.5M / mm² |

| Base Clock | 1058 MHz | 1417 MHz |

| Boost Clock | 1124 MHz | 1762 MHz |

| Memory Clock | 900 MHz (1800 Mbps effective) | 1500 MHz (12 Gbps effective) |

| Memory Size | 2 GB | 4 GB |

| Memory Type | DDR3 | GDDR6 |

| Memory Bus Width | 128 bit | 64 bit |

| Memory Bandwidth | 28.80 GB/s | 96.00 GB/s |

| Shading Units | 384 | 768 |

| RT Cores | None | 6 |

| Tensor Cores | None | 24 |

| Pixel Rate | 17.98 GPixel/s | 28.19 GPixel/s |

| Texture Rate | 26.98 GTexel/s | 42.29 GTexel/s |

| FP32 Performance | 863.2 GFLOPS | 2.706 TFLOPS |

| FP16 Performance | None listed | 2.706 TFLOPS (1:1) |

| TDP | 45 W | 50 W |

| Suggested PSU | 200 W | 250 W |

| Bus Interface | PCIe 2.0 x16 | PCIe 4.0 x8 |

| Display Outputs | 1x DVI, 1x DisplayPort 1.2 | 4x mini-DisplayPort 1.4a |

| DirectX Support | 12 (11_0) | 12 Ultimate (12_2) |

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

| Release Date | 2014-07-21 | 2024-04-15 |

| Predecessor | Quadro Fermi | Quadro Turing |

| Successor | Quadro Maxwell | Workstation Ada |

| Length | 160 mm (6.3 inches) | 163 mm (6.4 inches) |