AMD Radeon RX 550 vs NVIDIA P106-090 Comparison

The Verdict

The recorded data separates these two legacy GPUs cleanly. The NVIDIA P106-090 is the stronger compute part across every benchmark in the database, winning all three head-to-head tests with decisive margins. The AMD Radeon RX 550, by contrast, posts a lower average benchmark score and sits in a lower percentile, but it brings something the NVIDIA card lacks entirely: display outputs. The P106-090 is a mining-oriented card with no video outputs, so it cannot drive a monitor. The RX 550 can.

For anyone building a system that must output to a screen, the RX 550 is the only viable option despite its lower raw scores. For compute workloads, offscreen rendering, or headless processing tasks, the P106-090 is the clear winner, delivering roughly 21.6% higher average benchmark performance (13470 vs 11075) and a 4-point higher percentile ranking (54th vs 50th). The data does not support any scenario where the RX 550 outperforms the P106-090 in raw throughput, but the NVIDIA card's lack of display outputs makes it unsuitable for conventional desktop use.

Architecture Differences

The two GPUs come from different architectural generations and foundries. The NVIDIA P106-090 uses the GP106 chip built on the Pascal architecture, fabricated by TSMC on a 16 nm process. The die measures 200 mm² and contains 4,400 million transistors, yielding a transistor density of 22.0M per mm². The AMD Radeon RX 550 uses the Lexa chip based on GCN 4.0, manufactured by GlobalFoundries on a 14 nm node. Its die is substantially smaller at 103 mm², with 2,200 million transistors and a density of 21.4M per mm².

The compute resources differ sharply. The P106-090 packs 768 shading units, 48 texture mapping units, and 48 raster operation units. The RX 550 has 512 shading units, 32 TMUs, and only 16 ROPs. That 3:1 ROP ratio explains a significant portion of the rasterization gap. Pixel rate for the NVIDIA card is 73.49 GPixel/s versus 18.93 GPixel/s for AMD, a 3.9x difference. Texture rate tells a similar story: 73.49 GTexel/s vs 37.86 GTexel/s.

FP32 throughput heavily favors NVIDIA: 2.352 TFLOPS versus 1,211.4 GFLOPS. FP16 performance is where the architectures diverge philosophically. The P106-090 runs FP16 at a heavily reduced 36.74 GFLOPS, a 1:64 ratio, while the RX 550 delivers full-rate FP16 at 1,211.4 GFLOPS, a 1:1 ratio. That means AMD's card computes FP16 at the same speed as FP32, while NVIDIA's Pascal implementation treats FP16 as a low-priority path.

Memory subsystems also differ. The P106-090 has 3 GB of GDDR5 on a 192-bit bus, yielding 192.2 GB/s of bandwidth. The RX 550 has 2 GB of GDDR5 on a 128-bit bus, delivering 112.0 GB/s. Clock speeds favor NVIDIA as well: 1354 MHz base and 1531 MHz boost versus 1100 MHz base and 1183 MHz boost for AMD. Memory clocks run at 2002 MHz (8 Gbps effective) for NVIDIA and 1750 MHz (7 Gbps effective) for AMD.

API support is close but not identical. Both support DirectX 12, with NVIDIA at 12_1 and AMD at 12_0. OpenGL 4.6 is common to both. Vulkan support differs slightly: NVIDIA at 1.4, AMD at 1.3.

FAQ

Q: Can the NVIDIA P106-090 output video to a monitor?

A: No. The P106-090 lists "No outputs" for display outputs in the database. The AMD Radeon RX 550 provides 1x DVI, 1x HDMI 2.0b, and 1x DisplayPort 1.4a.

Q: Which card has higher raw compute performance?

A: The NVIDIA P106-090 wins all three head-to-head benchmarks. Its average benchmark score is 13470, compared to 11075 for the RX 550. The NVIDIA card also sits at the 54th percentile versus the 50th percentile for AMD.

Q: Is the RX 550 more power-efficient?

A: The RX 550 has a 50 W TDP and requires no power connectors, while the P106-090 has a 75 W TDP and needs a single 6-pin connector. Both share the same suggested PSU rating of 250 W.

Q: How do the memory configurations compare?

A: The P106-090 has 3 GB of GDDR5 on a 192-bit bus with 192.2 GB/s bandwidth. The RX 550 has 2 GB of GDDR5 on a 128-bit bus with 112.0 GB/s bandwidth. Memory clocks run at 2002 MHz (8 Gbps effective) for NVIDIA and 1750 MHz (7 Gbps effective) for AMD.

Q: What is the bus interface difference?

A: The P106-090 uses PCIe 1.0 x1, which is a severely limited connection. The RX 550 uses PCIe 3.0 x8, offering substantially more bandwidth for host communication.

Q: Which card has a smaller physical footprint?

A: The RX 550 is 145 mm (5.7 inches) long, while the P106-090 is 250 mm (9.8 inches) long. Both are dual-slot cards.

Specification Differences

| Specification | NVIDIA P106-090 | AMD Radeon RX 550 |

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

| Chip | GP106 | Lexa |

| Architecture | Pascal | GCN 4.0 |

| Generation | Mining GPUs | Polaris (RX 500) |

| Process Node | 16 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 4,400 million | 2,200 million |

| Die Size | 200 mm² | 103 mm² |

| Transistor Density | 22.0M / mm² | 21.4M / mm² |

| Base Clock | 1354 MHz | 1100 MHz |

| Boost Clock | 1531 MHz | 1183 MHz |

| Memory Clock | 2002 MHz, 8 Gbps effective | 1750 MHz, 7 Gbps effective |

| Memory Size | 3 GB | 2 GB |

| Memory Bus Width | 192 bit | 128 bit |

| Memory Bandwidth | 192.2 GB/s | 112.0 GB/s |

| Shading Units | 768 | 512 |

| TMUs | 48 | 32 |

| ROPs | 48 | 16 |

| Pixel Rate | 73.49 GPixel/s | 18.93 GPixel/s |

| Texture Rate | 73.49 GTexel/s | 37.86 GTexel/s |

| FP32 | 2.352 TFLOPS | 1,211.4 GFLOPS |

| FP16 | 36.74 GFLOPS (1:64) | 1,211.4 GFLOPS (1:1) |

| TDP | 75 W | 50 W |

| Power Connectors | 1x 6-pin | None |

| Bus Interface | PCIe 1.0 x1 | PCIe 3.0 x8 |

| Display Outputs | No outputs | 1x DVI, 1x HDMI 2.0b, 1x DisplayPort 1.4a |

| DirectX | 12 (12_1) | 12 (12_0) |

| Vulkan | 1.4 | 1.3 |

| Length | 250 mm (9.8 inches) | 145 mm (5.7 inches) |

| Release Date | 2017-07-30 | 2017-04-19 |

| Launch MSRP | None | 79 USD |

Head-to-Head Benchmarks

The database records three direct comparisons between these cards, and the NVIDIA P106-090 wins every one of them.

The largest margin appears in 3DMark Steel Nomad DX12. The P106-090 scores 509, while the RX 550 manages only 127. That is a 300.8% advantage for NVIDIA, a gap of roughly 4x. This test appears to stress the full GPU pipeline, and the P106-090's advantages in shading units, ROPs, memory bandwidth, and clock speed all compound into an overwhelming result. The RX 550's 16 ROPs and 112.0 GB/s bandwidth simply cannot keep pace with 48 ROPs and 192.2 GB/s.

In Geekbench OpenCL, the P106-090 scores 21304 against 11063 for the RX 550. The delta is 92.6%, meaning the NVIDIA card nearly doubles the AMD card's compute output. This aligns with the FP32 figures, where 2.352 TFLOPS versus 1,211.4 GFLOPS predicts roughly a 94% advantage. The measured result is consistent with that theoretical ceiling.

The closest contest is Geekbench Vulkan. The P106-090 scores 18596, and the RX 550 scores 12270, a 51.6% difference. While still a clear win for NVIDIA, the Vulkan result shows the RX 550 in its best light. The GCN architecture's Vulkan driver support appears to narrow the gap somewhat, and the AMD card's 1:1 FP16 ratio may help in compute-heavy Vulkan workloads. Still, the P106-090's higher clock speeds and wider memory bus carry the day.

The overall benchmark averages reinforce the pattern. The P106-090 sits at 13470, while the RX 550 sits at 11075. That is a 21.6% gap in average score, placing the NVIDIA card at the 54th percentile of all GPUs and the AMD card at the 50th percentile. The rival sets in the database reflect the positioning: the P106-090's nearest rivals include the GeForce GTX 570 (13515, 0.3% behind) and the Radeon Pro 555 (13407, 0.5% ahead), while the RX 550's nearest rivals include the GeForce GTX 1650 SUPER (11047, 0.3% ahead) and the FirePro W4300 (11225, 1.3% behind).

The record shows a straightforward hierarchy. The P106-090 is the faster card by every measured metric, but it is a headless compute device. The RX 550 is slower yet functional as a display adapter. The choice depends entirely on whether the system needs video output.