NVIDIA GeForce RTX 2060 SUPER vs NVIDIA GeForce RTX 5090 D V2 Comparison

The Verdict

The database comparison between the NVIDIA GeForce RTX 2060 SUPER and the NVIDIA GeForce RTX 5090 D V2 is not a contest in the traditional sense; it is a generational chasm. The RTX 5090 D V2 wins the only head-to-head benchmark recorded, the 3dmark Steel Nomad DX12 test, with a score of 16504 against the RTX 2060 SUPER's 2011. This represents an 87.8% delta, meaning the newer card is drastically faster in this specific workload.

However, the data presents a curious anomaly: the RTX 5090 D V2's average benchmark score is 16504, placing it at the 59th percentile of all GPUs, while the older RTX 2060 SUPER, with an average score of 18093, sits at the 62nd percentile. This is because the RTX 2060 SUPER has a much richer benchmark history across multiple tests (DirectX 9, 10, 11, 12, OpenCL, Vulkan, and compute), while the RTX 5090 D V2 only has a single recorded score. The verdict must therefore be split: for raw, modern DirectX 12 performance, the RTX 5090 D V2 is the clear choice based on the recorded data. For users looking at a broader legacy software landscape where the RTX 2060 SUPER has demonstrated consistent results, the older card shows a more versatile, albeit slower, profile.

Given the available data, the RTX 5090 D V2 is for those who prioritize peak performance in current-generation DirectX 12 titles. The RTX 2060 SUPER is for those who need a card with proven results across a wider array of older APIs and workloads, and who are working within a system that cannot accommodate the newer card's power and physical requirements.

FAQ

Q: Which card is faster in the only benchmark they share?

A: The NVIDIA GeForce RTX 5090 D V2 is faster. In the 3dmark Steel Nomad DX12 test, it scored 16504, while the RTX 2060 SUPER scored 2011, a delta of -87.8% in favor of the newer card.

Q: How do their average benchmark scores compare?

A: The RTX 2060 SUPER has a higher average benchmark score of 18093 compared to the RTX 5090 D V2's 16504. This is because the RTX 2060 SUPER has scores from ten different tests, whereas the RTX 5090 D V2 only has one recorded score.

Q: What are the nearest rivals for each card according to the database?

A: For the RTX 2060 SUPER, the nearest rivals are the NVIDIA GeForce RTX 3060 Mobile (avg score 18159, delta -0.4%), AMD Radeon Pro 5700 (18189, -0.5%), AMD Radeon RX 460 (18373, -1.5%), and Intel Arc A770M (18383, -1.6%). For the RTX 5090 D V2, the nearest rivals are the NVIDIA T400 (16508, 0%), AMD Radeon PRO W7500 (16415, 0.5%), NVIDIA RTX PRO 6000 Blackwell (16408, 0.6%), and AMD Radeon RX 5700 XT (16361, 0.9%).

Q: What are the power connector requirements for each card?

A: The RTX 2060 SUPER requires a single 8-pin power connector and a suggested power supply of 450 W. The RTX 5090 D V2 requires a single 16-pin power connector and a suggested power supply of 950 W.

Q: Which card has more memory and what type is it?

A: The RTX 5090 D V2 has 24 GB of GDDR7 memory on a 384-bit bus, offering 1.34 TB/s of bandwidth. The RTX 2060 SUPER has 8 GB of GDDR6 memory on a 256-bit bus, offering 448.0 GB/s of bandwidth.

Architecture Differences

The architectural gap between these two GPUs is immense, reflecting a shift from the Turing era to the Blackwell 2.0 era. The RTX 2060 SUPER is built on the TU106 chip using a 12 nm process at TSMC, containing 10,800 million transistors on a 445 mm² die. This results in a transistor density of 24.3 million per mm². In contrast, the RTX 5090 D V2 uses the GB202 chip on a 5 nm process, also from TSMC, with a staggering 92,200 million transistors packed into a 750 mm² die, yielding a density of 122.9 million per mm².

The compute resources differ by an order of magnitude. The RTX 2060 SUPER has 2176 shading units, 136 texture mapping units (TMUs), and 64 raster output units (ROPs). The RTX 5090 D V2 has 21760 shading units, 680 TMUs, and 176 ROPs. For ray tracing, the RTX 2060 SUPER has 34 RT cores, while the RTX 5090 D V2 has 170. Tensor core counts also scale dramatically: 272 on the older card versus 680 on the newer one.

The memory architecture has also evolved. The RTX 2060 SUPER uses 8 GB of GDDR6 with a 256-bit bus, while the RTX 5090 D V2 uses 24 GB of GDDR7 with a 384-bit bus. The FP32 compute performance tells the story of generational leap: the RTX 2060 SUPER delivers 7.181 TFLOPS, while the RTX 5090 D V2 delivers 104.8 TFLOPS. Notably, the FP16 performance differs in ratio: the RTX 2060 SUPER achieves 14.36 TFLOPS (2:1 ratio to FP32), while the RTX 5090 D V2 achieves 104.8 TFLOPS (1:1 ratio), indicating a shift in compute design priorities.

Specification Differences

The recorded specifications show clear divergences across nearly every field. The process node shifts from 12 nm for the RTX 2060 SUPER to 5 nm for the RTX 5090 D V2. Base clock speeds rise from 1470 MHz to 2017 MHz, and boost clocks from 1650 MHz to 2407 MHz. Memory speed increases from 14 Gbps effective (GDDR6) to 28 Gbps effective (GDDR7).

Memory capacity and bandwidth are major differentiators: 8 GB vs 24 GB, and 448.0 GB/s vs 1.34 TB/s. The bus interface upgrades from PCIe 3.0 x16 to PCIe 5.0 x16. Power consumption is listed as 175 W TDP for the RTX 2060 SUPER and 575 W TDP for the RTX 5090 D V2, with the power connector changing from 1x 8-pin to 1x 16-pin, and the suggested PSU rising from 450 W to 950 W.

Physical dimensions also differ: the RTX 2060 SUPER is 229 mm long, 113 mm high, and 35 mm wide, while the RTX 5090 D V2 is 304 mm long, 137 mm high, and 48 mm wide. Both are dual-slot cards. Display outputs vary, with the older card offering 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, and 1x USB Type-C, while the newer card offers 1x HDMI 2.1b and 3x DisplayPort 2.1b. The production status is End-of-life for the RTX 2060 SUPER and Active for the RTX 5090 D V2.

Head-to-Head Benchmarks

The only shared benchmark in the database is the 3dmark Steel Nomad DX12 test. Here, the NVIDIA GeForce RTX 5090 D V2 records a score of 16504, while the NVIDIA GeForce RTX 2060 SUPER records a score of 2011. The delta is -87.8% from the perspective of the RTX 2060 SUPER, meaning the newer card outperforms the older one by a massive margin in this DirectX 12 workload. This is the single most significant data point for direct comparison, and it strongly indicates that for modern, API-heavy gaming, the RTX 5090 D V2 is in a completely different performance class.

The RTX 2060 SUPER's other benchmark scores, such as 76957 in Geekbench OpenCL and 77402 in Geekbench Vulkan, along with its Passmark scores across DirectX 9 (218), DirectX 10 (111), DirectX 11 (130), and DirectX 12 (61), show it is a capable card for legacy and varied workloads. The RTX 5090 D V2 has no such historical data, so its performance in those specific tests remains unrecorded in this database.

Where Each One Wins

The NVIDIA GeForce RTX 5090 D V2 wins decisively in the single head-to-head benchmark, the 3dmark Steel Nomad DX12 test. This indicates a clear advantage in modern, DirectX 12-based gaming and compute scenarios. Its specifications also point to advantages in high-resolution texture workloads, given its 24 GB memory pool and 1.34 TB/s bandwidth, as well as in AI and ray tracing tasks due to its 680 tensor cores and 170 RT cores.

The NVIDIA GeForce RTX 2060 SUPER wins in the field of benchmark coverage and consistency. It has recorded scores across ten different tests, including legacy APIs like DirectX 9, DirectX 10, and DirectX 11, as well as compute workloads like Passmark GPU Compute and Geekbench OpenCL. Its average benchmark score of 18093 is higher than the RTX 5090 D V2's 16504, which is a direct result of this wider test suite. For users running older software or specific compute tasks that rely on these APIs, the RTX 2060 SUPER has demonstrated proven results, whereas the RTX 5090 D V2 has no recorded data for such workloads. Additionally, the RTX 2060 SUPER's lower power draw (175 W TDP vs 575 W TDP) and smaller physical footprint (229 mm vs 304 mm length) make it a more practical choice for systems with limited power supply headroom or smaller cases, based strictly on the specification data.