AMD Radeon RX 570X vs NVIDIA GeForce GTX 1660 SUPER Comparison

# AMD Radeon RX 570X vs NVIDIA GeForce GTX 1660 SUPER

The NVIDIA GeForce GTX 1660 SUPER is the clear winner in every head-to-head benchmark, but the gap in raw 3D performance is far larger than the overall score difference suggests. The RX 570X trails by 84.9% in Passmark G3D, a decisive margin that places the two cards in entirely different performance tiers despite only a 2-point difference in percentile rank. This is not a close contest; it is a generational and architectural mismatch where the newer Turing-based GPU dominates the older Polaris design.

Head-to-Head Benchmarks

The data is unambiguous: the GTX 1660 SUPER wins all three comparable benchmarks, with the most significant victory coming in the 3D rendering test. In Passmark G3D, the GTX 1660 SUPER scores 12,699 against the RX 570X’s 1,923, a delta of -84.9% from the AMD card’s perspective. This means the NVIDIA card delivers roughly 6.6 times the 3D performance, a margin that would translate into playable frame rates at settings where the RX 570X would struggle to maintain minimums.

The OpenCL benchmark tells a similar but less extreme story. Here the GTX 1660 SUPER posts 52,490 versus 38,939 for the RX 570X, a 25.8% deficit for the AMD card. This is a substantial gap, but it is far narrower than the G3D result, suggesting that compute-heavy workloads see a smaller advantage for NVIDIA. The RX 570X’s older GCN architecture with its 2,048 shading units helps in raw compute throughput, though it cannot overcome the efficiency and feature advantages of Turing.

The 2D benchmark is the closest contest, with the GTX 1660 SUPER scoring 805 against 750 for the RX 570X, a modest 6.8% lead. This indicates that basic desktop rendering and 2D acceleration are nearly equivalent, with the NVIDIA card holding a slight edge but no decisive advantage. For users whose primary workload is not gaming or compute, the difference here is negligible.

Across all three tests, the GTX 1660 SUPER wins 3-0 with an aggregate pattern showing that the gap widens as workload intensity increases. The 2D test shows parity, the compute test shows a meaningful but not overwhelming NVIDIA lead, and the 3D test shows a landslide. This scaling suggests the architectural differences between GCN 4.0 and Turing become more pronounced under sustained 3D load.

Architecture Differences

The two cards represent fundamentally different design philosophies separated by roughly 18 months of development. The RX 570X uses AMD’s Polaris 20 chip built on GCN 4.0 architecture, manufactured on a 14 nm process at GlobalFoundries. The GTX 1660 SUPER uses NVIDIA’s TU116 chip with Turing architecture, built on a 12 nm process at TSMC. The process node difference is modest, but the architectural changes are substantial.

The RX 570X packs 5,700 million transistors on a 232 mm² die, giving a transistor density of 24.6 million per square millimeter. The GTX 1660 SUPER has 6,600 million transistors on a larger 284 mm² die, with a slightly lower density of 23.2 million per square millimeter. Despite having fewer transistors overall, the NVIDIA card achieves higher clock speeds and better performance, indicating that Turing’s design is more efficient per transistor than GCN 4.0.

Clock speeds tell a significant part of the story. The RX 570X runs at a base of 1168 MHz and boosts to 1244 MHz, while the GTX 1660 SUPER starts at 1530 MHz and boosts to 1785 MHz. That is a 43% higher boost clock for NVIDIA, which explains much of the 3D performance gap. The AMD card compensates with more shading units (2,048 versus 1,408) and more texture mapping units (128 versus 88), but the clock advantage proves decisive.

Memory configurations diverge sharply. The RX 570X has 8 GB of GDDR5 on a 256-bit bus, delivering 224.0 GB/s of bandwidth. The GTX 1660 SUPER has 6 GB of GDDR6 on a 192-bit bus, delivering 336.0 GB/s. Despite the narrower bus, the newer GDDR6 memory operates at 14 Gbps effective versus 7 Gbps for GDDR5, resulting in 50% more bandwidth from 25% less memory capacity. The NVIDIA card’s higher memory bandwidth is critical for modern games that increasingly rely on texture streaming.

Feature support also differs. The RX 570X supports DirectX 12 (12_0) and Vulkan 1.3, while the GTX 1660 SUPER supports DirectX 12 (12_1) and Vulkan 1.4. The higher DirectX feature level and newer Vulkan version give the NVIDIA card access to additional rendering techniques. Neither card has ray tracing or tensor cores, so this is not a factor, but the API differences matter for future game compatibility.

Power consumption and physical design show interesting trade-offs. The RX 570X has a 150 W TDP and requires a 450 W power supply, while the GTX 1660 SUPER has a 125 W TDP and needs only a 300 W PSU. The NVIDIA card also uses a more efficient single 8-pin connector versus the RX 570X’s 1x 6-pin. Both are dual-slot cards, but the GTX 1660 SUPER is shorter at 229 mm versus 241 mm for the RX 570X.

The Verdict

The GTX 1660 SUPER is the superior card by every measurable benchmark, and the margin is overwhelming in 3D workloads. The 84.9% G3D lead is not a minor advantage; it represents a different performance class entirely. If the primary use case is gaming or 3D rendering, the RX 570X is not competitive. The NVIDIA card’s higher clocks, faster memory, and more modern architecture combine to deliver results that the AMD card cannot approach.

However, the RX 570X is not without merit in specific contexts. Its 8 GB of VRAM doubles the GTX 1660 SUPER’s 6 GB, which could matter for workloads that require large memory footprints, such as certain compute tasks or high-resolution texture sets. The AMD card also has a higher transistor density and more shading units, which helps in compute-heavy scenarios where parallelism matters more than clock speed. The 25.8% OpenCL gap, while significant, is far smaller than the 3D gap, suggesting that the RX 570X is more competitive in general-purpose compute.

For most users, the choice is clear: the GTX 1660 SUPER is the better purchase, offering vastly superior 3D performance, higher memory bandwidth, better API support, and lower power requirements. The RX 570X only makes sense for users who specifically need 8 GB of VRAM and prioritize compute workloads over gaming performance. The data shows a 3-0 sweep for NVIDIA, and the verdict follows the evidence.

FAQ

Q: Which card has higher overall benchmark scores?

A: The GTX 1660 SUPER has an average benchmark score of 12,986 versus 13,871 for the RX 570X. Despite this, the GTX 1660 SUPER wins all three head-to-head benchmarks, with the largest margin in Passmark G3D (12,699 vs 1,923).

Q: How do the memory configurations compare?

A: The RX 570X has 8 GB of GDDR5 on a 256-bit bus with 224.0 GB/s bandwidth. The GTX 1660 SUPER has 6 GB of GDDR6 on a 192-bit bus with 336.0 GB/s bandwidth. The NVIDIA card delivers 50% more bandwidth despite having less capacity.

Q: Which card has higher clock speeds?

A: The GTX 1660 SUPER operates at 1530 MHz base and 1785 MHz boost, while the RX 570X runs at 1168 MHz base and 1244 MHz boost. The NVIDIA card’s boost clock is 43% higher.

Q: What are the power requirements for each card?

A: The RX 570X has a 150 W TDP and requires a 450 W power supply with a single 6-pin connector. The GTX 1660 SUPER has a 125 W TDP, needs only a 300 W PSU, and uses a single 8-pin connector.

Q: Do either of these cards support ray tracing?

A: No. Both the RX 570X and GTX 1660 SUPER have no ray tracing cores and no tensor cores. They rely on traditional rasterization with DirectX 12 (12_0 and 12_1 respectively) and Vulkan support.

Q: What is the production status of these cards?

A: Both cards are end-of-life. The RX 570X was released on April 10, 2018, and its successor is Vega. The GTX 1660 SUPER was released on October 28, 2019, with its successor being the GeForce 20 series.

Where Each One Wins

The GTX 1660 SUPER wins decisively in 3D rendering and gaming, as evidenced by its 84.9% lead in Passmark G3D. The NVIDIA card also wins in OpenCL compute by 25.8% and in 2D performance by 6.8%. Its higher memory bandwidth, faster clocks, and better API support make it the superior choice for any workload that stresses the GPU’s rendering pipeline.

The RX 570X’s only advantages are structural rather than benchmark-based. It has 8 GB of VRAM versus 6 GB, which can be beneficial for memory-intensive tasks like large texture packs or certain compute workloads. It also has more shading units (2,048 vs 1,408) and more TMUs (128 vs 88), which theoretically helps in parallel compute scenarios. The AMD card’s lower transistor density (24.6M vs 23.2M per mm²) suggests a less complex design that may have lower manufacturing costs, but this does not translate to performance advantages.

In practical terms, the RX 570X wins in scenarios where VRAM capacity is the limiting factor and where the workload does not heavily depend on memory bandwidth. The GTX 1660 SUPER wins everywhere else, which covers the vast majority of real-world use cases.

Specification Differences

| Specification | AMD Radeon RX 570X | NVIDIA GeForce GTX 1660 SUPER |

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

| Architecture | GCN 4.0 | Turing |

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

| Transistors | 5,700 million | 6,600 million |

| Die Size | 232 mm² | 284 mm² |

| Base Clock | 1168 MHz | 1530 MHz |

| Boost Clock | 1244 MHz | 1785 MHz |

| Memory Size | 8 GB GDDR5 | 6 GB GDDR6 |

| Memory Bus | 256 bit | 192 bit |

| Memory Bandwidth | 224.0 GB/s | 336.0 GB/s |

| Shading Units | 2048 | 1408 |

| TMUs | 128 | 88 |

| ROPs | 32 | 48 |

| FP32 Performance | 5.095 TFLOPS | 5.027 TFLOPS |

| FP16 Performance | 5.095 TFLOPS (1:1) | 10.05 TFLOPS (2:1) |

| Pixel Rate | 39.81 GPixel/s | 85.68 GPixel/s |

| Texture Rate | 159.2 GTexel/s | 157.1 GTexel/s |

| TDP | 150 W | 125 W |

| Power Connector | 1x 6-pin | 1x 8-pin |

| Suggested PSU | 450 W | 300 W |

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

| Vulkan Support | 1.3 | 1.4 |

| Dimensions | 241 mm (9.5 inches) | 229 mm (9 inches) |

| Release Date | 2018-04-10 | 2019-10-28 |

The GTX 1660 SUPER offers higher clocks, more bandwidth, better API support, and lower power draw. The RX 570X counters with more memory, more shading units, and a slightly higher FP32 throughput. The NVIDIA card’s 2:1 FP16 ratio versus the AMD card’s 1:1 ratio gives it a significant advantage in half-precision compute workloads, doubling its effective throughput to 10.05 TFLOPS. The RX 570X’s 5.095 TFLOPS FP16 matches its FP32, limiting its utility in AI and machine learning tasks that rely on reduced precision.