AMD Radeon RX 550 vs NVIDIA GeForce GTX 1660 SUPER Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 1660 SUPER has an average benchmark score of 12986, while the AMD Radeon RX 550 scores 11075. The NVIDIA card sits at the 53rd percentile of all GPUs, compared to the 50th percentile for the AMD card.

Q: How large is the performance gap in the 3DMark Steel Nomad DX12 test?

A: The GTX 1660 SUPER scores 1278, which is 906.3% higher than the RX 550's score of 127. This is the largest single-test delta between the two cards.

Q: What are the memory configurations of the two cards?

A: The GTX 1660 SUPER has 6 GB of GDDR6 memory on a 192-bit bus with 336.0 GB/s bandwidth. The RX 550 has 2 GB of GDDR5 memory on a 128-bit bus with 112.0 GB/s bandwidth.

Q: Which card requires an external power connector?

A: The GTX 1660 SUPER uses a single 8-pin power connector and has a 125 W TDP. The RX 550 has no power connectors and a 50 W TDP.

Q: What is the architectural generation for each GPU?

A: The GTX 1660 SUPER is based on NVIDIA's Turing architecture (TU116 chip) from the GeForce 16 generation. The RX 550 uses AMD's GCN 4.0 architecture (Lexa chip) from the Polaris RX 500 generation.

Q: Do both cards support DirectX 12 and Vulkan?

A: Yes, both support DirectX 12 and Vulkan, but with differences: the GTX 1660 SUPER supports DirectX 12 (12_1) and Vulkan 1.4, while the RX 550 supports DirectX 12 (12_0) and Vulkan 1.3.

Architecture Differences

The GTX 1660 SUPER and RX 550 represent two very different design philosophies and manufacturing eras. The NVIDIA card uses the TU116 chip built on TSMC's 12 nm process, packing 6,600 million transistors into a 284 mm² die. The AMD card uses the Lexa chip fabricated by GlobalFoundries on a 14 nm process, with 2,200 million transistors on a 103 mm² die. Transistor density is similar (23.2M / mm² for NVIDIA, 21.4M / mm² for AMD), but the NVIDIA chip is nearly three times larger in physical size and transistor count.

The compute pipelines diverge sharply. The GTX 1660 SUPER has 1,408 shading units, 88 texture mapping units, and 48 raster output units. The RX 550 has 512 shading units, 32 TMUs, and 16 ROPs. Neither card includes ray tracing or tensor cores, but their FP32 and FP16 capabilities differ. NVIDIA's card delivers 5.027 TFLOPS FP32 and 10.05 TFLOPS FP16 (2:1 ratio), while AMD's card delivers 1,211.4 GFLOPS in both FP32 and FP16 (1:1 ratio). The NVIDIA architecture uses a split FP16 path that doubles throughput, whereas AMD's GCN 4.0 processes both formats at the same rate.

Memory architecture is another fundamental split. The GTX 1660 SUPER uses 6 GB of GDDR6 running at 14 Gbps effective over a 192-bit bus, yielding 336.0 GB/s of bandwidth. The RX 550 uses 2 GB of GDDR5 at 7 Gbps effective on a 128-bit bus, producing 112.0 GB/s. That is a 3x difference in capacity and exactly 3x in bandwidth. Pixel and texture rates follow the same pattern: 85.68 GPixel/s and 157.1 GTexel/s for NVIDIA versus 18.93 GPixel/s and 37.86 GTexel/s for AMD.

The interface and physical design also differ. The GTX 1660 SUPER connects via PCIe 3.0 x16, while the RX 550 uses PCIe 3.0 x8. NVIDIA's card is 229 mm long, 111 mm tall, and 35 mm wide, requiring a dual-slot bracket and a 300 W suggested PSU. AMD's card is 145 mm long with a dual-slot design and a 250 W suggested PSU. Both cards are end-of-life products, but their release dates are more than two years apart: the RX 550 launched in April 2017, and the GTX 1660 SUPER launched in October 2019.

Head-to-Head Benchmarks

The recorded data shows a decisive sweep: the GTX 1660 SUPER wins all three shared benchmark tests, with no wins for the RX 550. The margins are not subtle, they are generational.

In 3DMark Steel Nomad DX12, the GTX 1660 SUPER scores 1278 against the RX 550's 127. That is a 906.3% delta, the largest advantage in any shared workload. This test stresses modern DirectX 12 rendering paths, and the NVIDIA card's Turing architecture handles it with overwhelming efficiency. For context, the RX 550's score of 127 is roughly one-tenth of the NVIDIA card's result.

Geekbench OpenCL shows a 374.5% advantage. The GTX 1660 SUPER records 52490 points, while the RX 550 manages 11063. OpenCL exercises general-purpose compute across the shading units, and the NVIDIA card's 1,408 cores versus 512 cores, combined with higher clocks and memory bandwidth, produces a massive gap. The RX 550's FP32 throughput of 1,211.4 GFLOPS is about 24% of the GTX 1660 SUPER's 5.027 TFLOPS, which aligns closely with the observed score ratio.

Geekbench Vulkan follows a similar pattern: 57102 for NVIDIA versus 12270 for AMD, a 365.4% delta. Vulkan is a low-level API that rewards raw hardware throughput, and the GTX 1660 SUPER's advantage in shading units, texture units, and bandwidth shows up clearly. The RX 550's Vulkan support is version 1.3, while NVIDIA offers 1.4, but the score gap is driven primarily by hardware resources rather than API version differences.

The average benchmark scores tell the same story. The GTX 1660 SUPER averages 12986, placing it at the 53rd percentile of all GPUs. The RX 550 averages 11075, at the 50th percentile. The nearest rival data for the GTX 1660 SUPER includes the RTX 3050 Ti Mobile at 12940 (0.4% lower), the RX 580 at 12928 (0.4% lower), and the Tesla M2090 at 13075 (0.7% higher). The RX 550's nearest rivals include the GTX 1650 SUPER at 11047 (0.3% lower) and the FirePro W4300 at 11225 (1.3% higher). Both cards sit within a tight cluster of similarly performing GPUs, but the clusters themselves are far apart.

Specification Differences

| Specification | NVIDIA GeForce GTX 1660 SUPER | AMD Radeon RX 550 |

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

| Architecture | Turing | GCN 4.0 |

| Chip | TU116 | Lexa |

| Process node | 12 nm (TSMC) | 14 nm (GlobalFoundries) |

| Transistors | 6,600 million | 2,200 million |

| Die size | 284 mm² | 103 mm² |

| Transistor density | 23.2M / mm² | 21.4M / mm² |

| Base clock | 1530 MHz | 1100 MHz |

| Boost clock | 1785 MHz | 1183 MHz |

| Memory clock | 14 Gbps effective | 7 Gbps effective |

| Memory size | 6 GB GDDR6 | 2 GB GDDR5 |

| Memory bus width | 192 bit | 128 bit |

| Memory bandwidth | 336.0 GB/s | 112.0 GB/s |

| Shading units | 1408 | 512 |

| TMUs | 88 | 32 |

| ROPs | 48 | 16 |

| Pixel rate | 85.68 GPixel/s | 18.93 GPixel/s |

| Texture rate | 157.1 GTexel/s | 37.86 GTexel/s |

| FP32 | 5.027 TFLOPS | 1,211.4 GFLOPS |

| FP16 | 10.05 TFLOPS (2:1) | 1,211.4 GFLOPS (1:1) |

| TDP | 125 W | 50 W |

| Power connectors | 1x 8-pin | None |

| Suggested PSU | 300 W | 250 W |

| Bus interface | PCIe 3.0 x16 | PCIe 3.0 x8 |

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

| Vulkan | 1.4 | 1.3 |

| Length | 229 mm | 145 mm |

| Height | 111 mm | Not specified |

| Width | 35 mm | Not specified |

| Release date | 2019-10-28 | 2017-04-19 |

| Launch MSRP | 229 USD | 79 USD |

Where Each One Wins

The GTX 1660 SUPER wins in every measured category, and its advantages are most pronounced in heavy 3D rendering and compute workloads. The 3DMark Steel Nomad DX12 result, with a 906.3% lead, indicates that modern DirectX 12 games with complex geometry and shading will run far smoother on the NVIDIA card. The 6 GB GDDR6 frame buffer also enables higher texture detail and larger resolutions than the 2 GB GDDR5 buffer on the RX 550. For users targeting 1080p gaming with recent titles, the GTX 1660 SUPER is the clear choice based on the recorded data.

The RX 550 does not win any benchmark category, but its profile suits a different class of use case. Its 50 W TDP and lack of power connectors mean it can run in systems with minimal power delivery, and its 145 mm length fits in compact cases where the 229 mm NVIDIA card would not. The 250 W suggested PSU requirement is lower than the GTX 1660 SUPER's 300 W, making the RX 550 viable for small form factor or low-budget builds that prioritize physical compatibility over performance. The PCIe 3.0 x8 interface is sufficient for its bandwidth needs, though the GTX 1660 SUPER's x16 interface provides more headroom.

For compute tasks that rely on OpenCL or Vulkan, the GTX 1660 SUPER delivers 374.5% and 365.4% higher scores respectively. That makes it suitable for GPU-accelerated productivity workloads, while the RX 550's 1,211.4 GFLOPS FP32 performance is modest by comparison. The RX 550's 1:1 FP16 ratio matches its FP32 throughput, but that parity does not help against the GTX 1660 SUPER's 2:1 FP16 advantage.

The percentile data contextualizes both cards within the broader GPU landscape. The GTX 1660 SUPER at the 53rd percentile sits just above the median of all GPUs, trading blows with the RTX 3050 Ti Mobile and RX 580 within 0.4%. The RX 550 at the 50th percentile is squarely median, with the GTX 1650 SUPER just 0.3% ahead and the RTX PRO 6000 series just 0.1% behind. In practical terms, the GTX 1660 SUPER is a mid-range card that can handle contemporary gaming, while the RX 550 is an entry-level card suited for basic display output, legacy titles, or low-demand applications. The measured data shows no scenario where the RX 550 outpaces the GTX 1660 SUPER, so the choice between them depends entirely on system constraints and workload expectations rather than performance parity.