GPU Comparison

NVIDIA’s GeForce 820A and AMD’s Radeon 550 are both end-of-life graphics solutions, but they occupy very different positions in the hardware landscape. The 820A is an integrated graphics processor (IGP) from 2014, built on an older architecture, while the Radeon 550 is a dual-slot discrete card from 2017. The benchmark data shows a decisive single win for the AMD part, yet the story is more nuanced when considering their respective roles and the specifications that define them.

Where Each One Wins

The benchmark results are unambiguous in terms of raw compute performance. The AMD Radeon 550 wins the only head-to-head benchmark available, the Geekbench OpenCL test, with a score of 10,339 against the NVIDIA GeForce 820A’s 2,983. This represents a massive 71.1% advantage for the AMD card, indicating that in any compute-heavy workload, such as OpenCL-accelerated rendering, video encoding, or general-purpose GPU tasks, the Radeon 550 is the clear choice. The data suggests the 820A is not competitive in this arena; its score is less than a third of the Radeon’s.

However, the NVIDIA 820A’s wins are not found in the head-to-head data but in its design context. As an IGP with a 15 W TDP and no power connectors, it is built for ultra-low-power, portable devices where the GPU is integrated into the system. Its "win" is in scenarios where a discrete card cannot even be installed. The Radeon 550, with its 50 W TDP and dual-slot footprint, requires a dedicated PCIe slot and a suggested 250 W power supply, making it inappropriate for the same class of hardware. Therefore, the 820A wins by default in the integrated, low-power segment, while the Radeon 550 wins in any scenario where performance is the primary metric.

Architecture Differences

The two GPUs are separated by three years of architectural evolution and a complete change in manufacturing philosophy. The NVIDIA GeForce 820A is based on the GF117 chip, using the Fermi 2.0 architecture, and is fabricated on a 28 nm process at TSMC. This chip packs 585 million transistors into a 116 mm² die, yielding a transistor density of 5.0 million per square millimeter. In contrast, the AMD Radeon 550 uses the Lexa chip, which employs the GCN 4.0 architecture, built on a 14 nm process at GlobalFoundries. The Lexa chip contains 2,200 million transistors on a smaller 103 mm² die, achieving a much higher density of 21.4 million per square millimeter. This newer process node and denser design are fundamental to the Radeon’s performance advantage.

The memory subsystems are vastly different. The 820A uses 1024 MB of DDR3 memory on a 64-bit bus, delivering a bandwidth of 16.02 GB/s. The Radeon 550 uses 2 GB of GDDR5 memory, also on a 64-bit bus, but achieves 56.00 GB/s of bandwidth, over 3.4 times higher. This bandwidth disparity alone explains a significant portion of the performance gap in memory-intensive tasks. The compute resources are also lopsided: the 820A has 96 shading units, 16 texture mapping units (TMUs), and 8 raster operation units (ROPs), while the Radeon 550 has 512 shading units, 32 TMUs, and 16 ROPs. The AMD card has more than 5 times the shading units and double the TMUs and ROPs, leading to a theoretical pixel rate of 18.93 GPixel/s and texture rate of 37.86 GTexel/s, versus the 820A’s 2.500 GPixel/s and 10.00 GTexel/s. In terms of raw floating-point throughput, the Radeon 550 delivers 1,211.4 GFLOPS in FP32, while the 820A manages only 240.0 GFLOPS, a 5x difference. The Radeon also supports FP16 at a 1:1 ratio, a feature the 820A lacks entirely.

Head-to-Head Benchmarks

The single head-to-head benchmark is Geekbench OpenCL, and the result is a landslide. The AMD Radeon 550 scores 10,339, while the NVIDIA GeForce 820A scores 2,983. The delta is -71.1% for the NVIDIA part, meaning it trails by over 71%. To put this in perspective, the 820A’s score of 2,983 places it in the 19th percentile of all GPUs, with its nearest rivals being the NVIDIA GeForce GTX 860M (scoring 2,967, a 0.5% difference), the GTX 750 Ti (2,953, 1% difference), and the Quadro P600 (2,923, 2.1% difference). This suggests the 820A is performing at a level comparable to those mid-range discrete cards from its era, which is surprising for an IGP, but the Radeon 550’s score of 10,339 places it in the 18th percentile, with its nearest rivals being the RTX 3060 Ti GDDR6X (2,795, 0.2% difference) and the GeForce 820M (2,784, 0.6% difference).

The contrast is stark. The Radeon 550’s OpenCL score is roughly 3.5 times higher than the 820A’s, and this is not a marginal improvement, it is a generational leap. The Radeon 550 also has a Vulkan score of 11,257, a feature the 820A does not support at all (its Vulkan field is null). In other compute and graphics tests, the Radeon 550 shows a mixed but generally positive profile: it scores 2,124 in Passmark G3D, 936 in Passmark GPU Compute, and 479 in Passmark G2D. Its DirectX scores are low (35 in DX9, 15 in DX11, 14 in DX12, 7 in DX10), which might seem odd, but these are likely relative scores on a specific benchmark scale. The 820A has no equivalent data in these tests, so the comparison is limited to OpenCL, where AMD wins decisively.

The Verdict

The data points to a clear verdict for performance: the AMD Radeon 550 is vastly superior in raw compute capability. Its OpenCL score is 10,339 versus 2,983, a 71.1% gap that is impossible to ignore. If the task is to run any GPU-accelerated workload, the Radeon 550 is the only rational choice between these two. Its 512 shading units, 2 GB of GDDR5, and 1,211.4 GFLOPS of FP32 performance dwarf the 820A’s 96 shading units, 1 GB of DDR3, and 240.0 GFLOPS. The Radeon 550 also supports modern APIs like Vulkan 1.3 and DirectX 12 (12_0), while the 820A is limited to DirectX 12 (11_0) and has no Vulkan support.

However, the NVIDIA GeForce 820A is not without merit in its specific niche. It is an IGP with a 15 W TDP, no power connectors, and a PCIe 2.0 x16 interface, designed for portable devices where space and power are at a premium. Its 116 mm² die and 585 million transistors are modest, but it succeeded in providing a baseline level of graphics capability in systems that could not accommodate a discrete card. For users with such a system, the 820A is the only option on this page. For anyone building or upgrading a desktop with a PCIe 3.0 x8 slot and a 250 W power supply, the Radeon 550 is the obvious pick, offering 3.4 times the memory bandwidth and 5 times the shading units. The verdict is straightforward: choose the Radeon 550 for performance, the 820A only if you are constrained to an integrated solution.

FAQ

Q: Which GPU is faster in the Geekbench OpenCL benchmark?

A: The AMD Radeon 550 is significantly faster, scoring 10,339 compared to the NVIDIA GeForce 820A’s 2,983, a delta of -71.1% for the NVIDIA part.

Q: What is the memory configuration difference between the two?

A: The NVIDIA 820A has 1024 MB of DDR3 memory with a 64-bit bus and 16.02 GB/s bandwidth. The AMD Radeon 550 has 2 GB of GDDR5 memory, also on a 64-bit bus, but with 56.00 GB/s bandwidth.

Q: Does the NVIDIA GeForce 820A support Vulkan?

A: No. The 820A’s Vulkan API field is null, while the AMD Radeon 550 supports Vulkan version 1.3.

Q: What are the TDP requirements for each card?

A: The NVIDIA GeForce 820A has a TDP of 15 W and uses no power connectors. The AMD Radeon 550 has a TDP of 50 W, also uses no power connectors, but has a suggested PSU rating of 250 W.

Q: How do the shading unit counts compare?

A: The AMD Radeon 550 has 512 shading units, while the NVIDIA GeForce 820A has 96. This is a 5.3x difference in favor of the AMD card.

Q: What is the process node and transistor count for each chip?

A: The NVIDIA 820A uses a 28 nm process at TSMC with 585 million transistors. The AMD Radeon 550 uses a 14 nm process at GlobalFoundries with 2,200 million transistors.

Specification Differences

| Specification | NVIDIA GeForce 820A | AMD Radeon 550 |

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

| Chip | GF117 | Lexa |

| Architecture | Fermi 2.0 | GCN 4.0 |

| Process Node | 28 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 585 million | 2,200 million |

| Die Size | 116 mm² | 103 mm² |

| Transistor Density | 5.0M / mm² | 21.4M / mm² |

| Base Clock | N/A | 1100 MHz |

| Boost Clock | N/A | 1183 MHz |

| Memory Clock | 1001 MHz (2 Gbps effective) | 1750 MHz (7 Gbps effective) |

| Memory Size | 1024 MB | 2 GB |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus Width | 64 bit | 64 bit |

| Memory Bandwidth | 16.02 GB/s | 56.00 GB/s |

| Shading Units | 96 | 512 |

| TMUs | 16 | 32 |

| ROPs | 8 | 16 |

| Pixel Rate | 2.500 GPixel/s | 18.93 GPixel/s |

| Texture Rate | 10.00 GTexel/s | 37.86 GTexel/s |

| FP32 Performance | 240.0 GFLOPS | 1,211.4 GFLOPS |

| FP16 Performance | N/A | 1,211.4 GFLOPS (1:1) |

| TDP | 15 W | 50 W |

| Slot Width | IGP | Dual-slot |

| Suggested PSU | N/A | 250 W |

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

| Display Outputs | Portable Device Dependent | 1x DVI, 1x HDMI 2.0b, 1x DisplayPort 1.4a |

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

| OpenGL Support | 4.6 | 4.6 |

| Vulkan Support | N/A | 1.3 |

| Release Date | 2014-03-16 | 2017-04-19 |

| Predecessor | GeForce 700A | Arctic Islands |

| Successor | GeForce 900A | Vega |

| Launch MSRP | N/A | 79 USD |