AMD Radeon RX 6650 XT vs NVIDIA GeForce RTX 2060 SUPER Comparison

Where Each One Wins

The benchmark split between these two cards is not a clean generational sweep. The AMD Radeon RX 6650 XT takes 6 of the 10 recorded head-to-head tests, while the NVIDIA GeForce RTX 2060 SUPER takes 4. The pattern of those wins, however, tells a clear story about workload suitability.

The RX 6650 XT is the stronger choice for modern DirectX 11 and DirectX 12 gaming workloads. Its PassMark DirectX 11 score of 169 beats the RTX 2060 SUPER's 130 by a substantial 30%, and its DirectX 12 result of 62 edges out the NVIDIA card's 61 by 1.6%. The AMD card also wins the PassMark G3D overall test, scoring 17166 against 16462, a 4.3% advantage. For raw compute tasks outside of gaming, the RX 6650 XT leads with a PassMark GPU compute score of 7532 versus 6721, a 12.1% margin. It also takes the 2D performance test, scoring 918 against 854, a 7.5% lead.

The RTX 2060 SUPER, meanwhile, wins in several specific and telling areas. Its 3DMark Steel Nomad DX12 score of 2011 is 7.6% ahead of the RX 6650 XT's 1859. The gap in Geekbench OpenCL is enormous: the NVIDIA card scores 76957 against the AMD card's 10111, a difference of 86.9%. This is not a small edge; it is a dominant victory in that particular compute benchmark. The RTX 2060 SUPER also wins the PassMark DirectX 10 test (111 vs 109, a 1.8% lead) and the PassMark DirectX 9 test (218 vs 215, a 1.4% lead). In Geekbench Vulkan, the AMD card wins with 81306 versus 77402, a 5% margin.

The practical interpretation is that the RX 6650 XT is tuned for contemporary rasterization and compute APIs, where its architectural efficiency pays off. The RTX 2060 SUPER retains strengths in legacy DirectX paths and shows overwhelming dominance in one specific OpenCL workload, which suggests it may be better suited for certain compute applications that rely heavily on that API.

Architecture Differences

The two cards come from different manufacturing generations and design philosophies. The AMD Radeon RX 6650 XT uses the Navi 23 chip built on RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC. It packs 11,060 million transistors into a 237 mm² die, yielding a transistor density of 46.7M per mm². The NVIDIA GeForce RTX 2060 SUPER uses the TU106 chip on Turing architecture, built on a 12 nm process, also at TSMC. It contains 10,800 million transistors but on a much larger 445 mm² die, giving a lower density of 24.3M per mm².

Clock speeds diverge sharply. The RX 6650 XT runs a base clock of 2055 MHz, a boost clock of 2635 MHz, and a game clock of 2410 MHz. Its memory runs at 2190 MHz with 17.5 Gbps effective. The RTX 2060 SUPER has a base clock of 1470 MHz and a boost of 1650 MHz, with memory at 1750 MHz and 14 Gbps effective. The AMD card's higher clocks are a direct consequence of the smaller, denser process node.

Memory configurations are similar in capacity but different in bus width. Both cards have 8 GB of GDDR6, but the RX 6650 XT uses a 128-bit bus delivering 280.3 GB/s, while the RTX 2060 SUPER uses a 256-bit bus delivering 448.0 GB/s. The NVIDIA card has a bandwidth advantage of 167.7 GB/s.

Compute unit counts are close but not identical. The RX 6650 XT has 2048 shading units, 128 TMUs, and 64 ROPs, with 32 ray tracing cores and no tensor cores. The RTX 2060 SUPER has 2176 shading units, 136 TMUs, and 64 ROPs, with 34 ray tracing cores and 272 tensor cores. The NVIDIA card's tensor cores are a notable feature absent from the AMD card. Pixel rate favors AMD at 168.6 GPixel/s versus 105.6 GPixel/s, and texture rate also favors AMD at 337.3 GTexel/s versus 224.4 GTexel/s. FP32 compute is 10.79 TFLOPS for AMD versus 7.181 TFLOPS for NVIDIA.

The bus interface differs: the RX 6650 XT uses PCIe 4.0 x8, while the RTX 2060 SUPER uses PCIe 3.0 x16. Display outputs also differ: AMD offers 1x HDMI 2.1 and 3x DisplayPort 1.4a, while NVIDIA offers 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, and 1x USB Type-C. Power requirements are nearly identical at 176 W for AMD and 175 W for NVIDIA, with both using a single 8-pin connector and a suggested 450 W PSU.

Head-to-Head Benchmarks

The largest single win for the RTX 2060 SUPER is in Geekbench OpenCL, where it scores 76957 against the RX 6650 XT's 10111. The delta is 86.9%, meaning the NVIDIA card is roughly 7.6 times faster in this specific test. This is the kind of outlier result that suggests a fundamental software or driver optimization difference rather than pure hardware capability.

The largest win for the RX 6650 XT is in PassMark DirectX 11, where it scores 169 against 130, a 30% advantage. This is a significant margin for a gaming API that remains widely used. The AMD card also wins PassMark GPU compute by 12.1% (7532 vs 6721), which indicates better general compute throughput in that benchmark suite.

In 3DMark Steel Nomad DX12, the RTX 2060 SUPER wins by 7.6% (2011 vs 1859). This is a modern DirectX 12 test, and the NVIDIA card's victory here is notable because it contradicts the pattern seen in PassMark DirectX 12, where AMD wins by 1.6%. The discrepancy suggests that benchmark methodology and specific workload characteristics matter greatly.

Geekbench Vulkan goes to AMD by 5% (81306 vs 77402). This is a moderate win and shows the RX 6650 XT handles Vulkan well. PassMark G3D goes to AMD by 4.3% (17166 vs 16462), which is a general 3D performance indicator. PassMark G2D goes to AMD by 7.5% (918 vs 854), indicating better 2D and desktop compositing performance.

The legacy DirectX tests are close. PassMark DirectX 10 goes to NVIDIA by 1.8% (111 vs 109), and PassMark DirectX 9 goes to NVIDIA by 1.4% (218 vs 215). These margins are small and suggest near-parity in older APIs. The overall average benchmark score favors AMD: the RX 6650 XT averages 19765 across its recorded benchmarks, while the RTX 2060 SUPER averages 18093. This is a 9.2% difference in average score, though the average is skewed by the outlier OpenCL result.

FAQ

Q: Which card is faster in modern DirectX 12 gaming?

A: It depends on the benchmark. In 3DMark Steel Nomad DX12, the RTX 2060 SUPER is 7.6% ahead. In PassMark DirectX 12, the RX 6650 XT is 1.6% ahead. The results are mixed, but the overall G3D score favors AMD by 4.3%.

Q: Is the RTX 2060 SUPER significantly better for compute workloads?

A: In Geekbench OpenCL, the RTX 2060 SUPER is dramatically ahead, scoring 76957 versus 10111, a difference of 86.9%. However, in PassMark GPU compute, the RX 6650 XT wins by 12.1% with a score of 7532 versus 6721. The answer depends heavily on which compute benchmark and API you trust.

Q: Which card has better memory bandwidth?

A: The RTX 2060 SUPER has a 256-bit bus and delivers 448.0 GB/s. The RX 6650 XT has a 128-bit bus and delivers 280.3 GB/s. NVIDIA's card has a 167.7 GB/s bandwidth advantage.

Q: Do both cards support ray tracing?

A: Yes. The RX 6650 XT has 32 ray tracing cores, and the RTX 2060 SUPER has 34. The NVIDIA card additionally has 272 tensor cores, which the AMD card lacks.

Q: Which card has a higher boost clock?

A: The RX 6650 XT has a boost clock of 2635 MHz, while the RTX 2060 SUPER has a boost clock of 1650 MHz. The AMD card runs nearly 1000 MHz higher.

Q: How do their overall benchmark averages compare?

A: The RX 6650 XT has an average benchmark score of 19765, while the RTX 2060 SUPER averages 18093. The RX 6650 XT also sits at the 65th percentile of all GPUs, compared to the 62nd percentile for the RTX 2060 SUPER.

The Verdict

The data presents a nuanced picture. The AMD Radeon RX 6650 XT is the better all-around card based on average benchmark score (19765 vs 18093), and it wins more head-to-head tests (6 out of 10). Its wins in PassMark DirectX 11 (30% ahead), GPU compute (12.1% ahead), and G3D (4.3% ahead) make it the stronger choice for general gaming and compute workloads measured by that suite. Its 7 nm process node and higher clocks (2635 MHz boost vs 1650 MHz) deliver better texture and pixel rates (337.3 GTexel/s vs 224.4 GTexel/s, 168.6 GPixel/s vs 105.6 GPixel/s).

However, the RTX 2060 SUPER should not be dismissed. Its 86.9% victory in Geekbench OpenCL is a massive advantage for any workload that relies on that API. Its 7.6% win in 3DMark Steel Nomad DX12 shows it can still hold its own in a modern DirectX 12 test. It also has double the memory bus width (256-bit vs 128-bit) and significantly higher bandwidth (448.0 GB/s vs 280.3 GB/s), plus tensor cores for AI-accelerated tasks.

For a PC builder prioritizing raw gaming performance across a broad range of APIs, the RX 6650 XT is the safer bet. For someone with specific OpenCL compute needs or who values the NVIDIA feature set including tensor cores, the RTX 2060 SUPER has clear justification. Both cards are end-of-life products, and both launched at the same MSRP of 399 USD. The choice comes down to workload, not price.

Specification Differences

| Specification | AMD Radeon RX 6650 XT | NVIDIA GeForce RTX 2060 SUPER |

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

| Process node | 7 nm | 12 nm |

| Die size | 237 mm² | 445 mm² |

| Transistor density | 46.7M / mm² | 24.3M / mm² |

| Base clock | 2055 MHz | 1470 MHz |

| Boost clock | 2635 MHz | 1650 MHz |

| Memory clock | 2190 MHz (17.5 Gbps effective) | 1750 MHz (14 Gbps effective) |

| Memory bus width | 128 bit | 256 bit |

| Memory bandwidth | 280.3 GB/s | 448.0 GB/s |

| Shading units | 2048 | 2176 |

| TMUs | 128 | 136 |

| Ray tracing cores | 32 | 34 |

| Tensor cores | None | 272 |

| Pixel rate | 168.6 GPixel/s | 105.6 GPixel/s |

| Texture rate | 337.3 GTexel/s | 224.4 GTexel/s |

| FP32 compute | 10.79 TFLOPS | 7.181 TFLOPS |

| FP16 compute | 21.59 TFLOPS (2:1) | 14.36 TFLOPS (2:1) |

| Bus interface | PCIe 4.0 x8 | PCIe 3.0 x16 |

| Display outputs | 1x HDMI 2.1, 3x DisplayPort 1.4a | 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, 1x USB Type-C |

| Dimensions | Not recorded | 229 mm length, 113 mm height, 35 mm width |

| Release date | 2022-05-09 | 2019-07-08 |