NVIDIA GeForce RTX 2060 SUPER vs NVIDIA GeForce RTX 2070 SUPER Comparison

Head-to-Head Benchmarks

The benchmark comparison between the NVIDIA GeForce RTX 2070 SUPER and the NVIDIA GeForce RTX 2060 SUPER is decisively lopsided in favor of the 2070 SUPER, which claims 9 of the 10 recorded tests. The single exception is a significant one, however: in the 3DMark Steel Nomad DX12 test, the 2060 SUPER posts a score of 2011 against the 2070 SUPER’s 1651, a 17.9% advantage. This is the largest delta in either direction and suggests that in this specific modern DirectX 12 workload, the lower-tier card punches above its class.

Everywhere else, the 2070 SUPER leads, but the margins vary widely. The smallest wins come in legacy and 2D-oriented tests: PassMark DirectX 9 shows only a 2.3% edge (223 vs. 218), and PassMark G2D shows a 2.8% edge (878 vs. 854). These are effectively negligible differences for practical purposes. The 2070 SUPER’s lead grows substantially in more demanding 3D and compute workloads. In PassMark G3D, it scores 18169 versus 16462, a 10.4% advantage. In PassMark GPU Compute, the gap is 12.4% (7557 vs. 6721). The Geekbench OpenCL result shows an 8.3% lead (83358 vs. 76957), while the Geekbench Vulkan test shows a much larger 17.1% lead (90637 vs. 77402), indicating that the 2070 SUPER scales better with modern graphics APIs.

The middle ground is occupied by the DirectX 10 and DirectX 11 PassMark tests, where the 2070 SUPER wins by 18.9% (132 vs. 111) and 16.2% (151 vs. 130) respectively. The DirectX 12 PassMark test shows a more modest 9.8% lead (67 vs. 61), which is curious given the 2060 SUPER’s strong showing in the Steel Nomad DX12 test. This discrepancy suggests that the 2060 SUPER’s advantage in that specific benchmark is not representative of broader DirectX 12 performance.

Looking at the aggregate data, the 2070 SUPER’s average benchmark score is 20282, which places it in the 65th percentile of all GPUs. The 2060 SUPER averages 18093, sitting in the 62nd percentile. The gap between them, roughly 12%, is consistent with the majority of individual test deltas. The 2070 SUPER’s nearest rivals in the database are the NVIDIA Quadro M4000M (avg score 20480, delta -1%), the NVIDIA GeForce RTX 3070 Mobile (20534, -1.2%), the Intel Arc B570 (20556, -1.3%), and the Intel Arc A750 (20582, -1.5%). The 2060 SUPER’s nearest rivals are the NVIDIA GeForce RTX 3060 Mobile (18159, -0.4%), the AMD Radeon Pro 5700 (18189, -0.5%), the AMD Radeon RX 460 (18373, -1.5%), and the Intel Arc A770M (18383, -1.6%). This contextualizes the pair: the 2070 SUPER competes in a higher performance tier, while the 2060 SUPER sits just below it.

Where Each One Wins

The data paints a clear picture for use-case selection. The 2070 SUPER is the default choice for anyone prioritizing compute-heavy tasks, modern API performance, and general 3D rendering. Its wins in Geekbench Vulkan (17.1% ahead), PassMark GPU Compute (12.4% ahead), and PassMark G3D (10.4% ahead) indicate that it handles shader-heavy workloads and parallel computation with more headroom. The PassMark DirectX 10 and DirectX 11 results, with leads of 18.9% and 16.2%, further reinforce its strength in traditional rasterization pipelines. For gaming at high resolutions or with demanding graphical settings, the 2070 SUPER’s superior texture rate and pixel rate (283.2 GTexel/s and 113.3 GPixel/s versus 224.4 GTexel/s and 105.6 GPixel/s) provide a measurable advantage.

The 2060 SUPER’s single win in 3DMark Steel Nomad DX12 cannot be ignored, but it is an outlier. That test favors the 2060 SUPER by 17.9%, which is the only instance where the lower-tier card leads by double digits. This could indicate a workload that is particularly sensitive to memory bandwidth or specific scheduling characteristics, but since both cards share the same 8 GB GDDR6 memory configuration, 256-bit bus width, and 448.0 GB/s bandwidth, the difference must stem from other architectural factors. The 2060 SUPER also has a lower TDP at 175 W versus 215 W, so it is the more power-efficient option in absolute terms. For users with a smaller power supply (450 W suggested versus 550 W) or those prioritizing lower power draw, the 2060 SUPER is the safer choice.

For legacy DirectX 9 content, the two are nearly identical: the 2070 SUPER leads by just 2.3% (223 vs. 218). The same applies to 2D workloads, where G2D scores are within 2.8% (878 vs. 854). Neither card offers a meaningful advantage for older games or desktop applications. The 2060 SUPER also holds a minor edge in physical dimensions, being shorter at 229 mm versus 267 mm, which matters for compact cases. However, both are dual-slot cards with identical 35 mm widths.

Architecture Differences

Both cards are built on NVIDIA’s Turing architecture using TSMC’s 12 nm process, but they use different chips. The 2070 SUPER is based on the TU104 die, which measures 545 mm² and contains 13,600 million transistors. The 2060 SUPER uses the TU106 die, which is smaller at 445 mm² with 10,800 million transistors. The transistor density is nearly identical: 25.0M per mm² for the TU104 versus 24.3M per mm² for the TU106. This means the 2070 SUPER’s performance advantage comes primarily from having more silicon area and more execution resources rather than from a denser or more efficient design.

The core configurations differ substantially. The 2070 SUPER has 2560 shading units, 160 texture mapping units, and 64 raster output units. The 2060 SUPER has 2176 shading units, 136 TMUs, and also 64 ROPs. The 2070 SUPER therefore has 384 more shaders and 24 more TMUs, which directly explains its higher texture rate and FP32 throughput. The ray tracing and tensor core counts also scale: the 2070 SUPER has 40 RT cores and 320 tensor cores, while the 2060 SUPER has 34 RT cores and 272 tensor cores. This gives the 2070 SUPER a 17.6% advantage in RT core count and a 17.6% advantage in tensor core count, aligning closely with its 17.1% lead in the Geekbench Vulkan test.

Clock speeds favor the 2070 SUPER as well. Its base clock is 1605 MHz and boost clock is 1770 MHz, versus 1470 MHz base and 1650 MHz boost for the 2060 SUPER. Combined with the higher core counts, this yields a FP32 performance of 9.062 TFLOPS for the 2070 SUPER versus 7.181 TFLOPS for the 2060 SUPER, a 26.2% gap. FP16 performance follows the same 2:1 ratio, with the 2070 SUPER delivering 18.12 TFLOPS versus 14.36 TFLOPS. Memory is a non-factor in differentiation: both use 8 GB of GDDR6 on a 256-bit bus with 1750 MHz memory clock and 448.0 GB/s bandwidth.

The display outputs differ slightly. The 2070 SUPER offers 1x HDMI 2.0 and 3x DisplayPort 1.4a, plus 1x USB Type-C. The 2060 SUPER replaces one DisplayPort with a DVI connector, offering 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, and 1x USB Type-C. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Power delivery also differs: the 2070 SUPER requires a 1x 6-pin plus 1x 8-pin connector, while the 2060 SUPER only needs a single 8-pin. The suggested PSU is 550 W for the 2070 SUPER and 450 W for the 2060 SUPER.

FAQ

Q: Which card has a higher average benchmark score?

A: The NVIDIA GeForce RTX 2070 SUPER has an average benchmark score of 20282, compared to 18093 for the NVIDIA GeForce RTX 2060 SUPER.

Q: Is the RTX 2060 SUPER ever faster than the RTX 2070 SUPER?

A: Yes, in one test. The RTX 2060 SUPER wins the 3DMark Steel Nomad DX12 benchmark with a score of 2011 versus 1651, a 17.9% advantage.

Q: How do the two cards compare in ray tracing capability?

A: The RTX 2070 SUPER has 40 RT cores, while the RTX 2060 SUPER has 34 RT cores. This gives the 2070 SUPER a 17.6% advantage in RT core count.

Q: What are the power requirements for each card?

A: The RTX 2070 SUPER has a TDP of 215 W and requires a 550 W power supply, using a 1x 6-pin plus 1x 8-pin connector. The RTX 2060 SUPER has a TDP of 175 W and requires a 450 W power supply, using a single 8-pin connector.

Q: Do both cards have the same memory configuration?

A: Yes, both have 8 GB of GDDR6 memory on a 256-bit bus with a memory clock of 1750 MHz (14 Gbps effective) and bandwidth of 448.0 GB/s.

Q: Which card is better for compute workloads?

A: The RTX 2070 SUPER is better, scoring 7557 in PassMark GPU Compute versus 6721 for the RTX 2060 SUPER, a 12.4% advantage. It also leads in Geekbench OpenCL by 8.3% (83358 vs. 76957).

Specification Differences

| Specification | NVIDIA GeForce RTX 2070 SUPER | NVIDIA GeForce RTX 2060 SUPER |

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

| Chip | TU104 | TU106 |

| Die Size | 545 mm² | 445 mm² |

| Transistors | 13,600 million | 10,800 million |

| Transistor Density | 25.0M / mm² | 24.3M / mm² |

| Base Clock | 1605 MHz | 1470 MHz |

| Boost Clock | 1770 MHz | 1650 MHz |

| Shading Units | 2560 | 2176 |

| TMUs | 160 | 136 |

| RT Cores | 40 | 34 |

| Tensor Cores | 320 | 272 |

| Pixel Rate | 113.3 GPixel/s | 105.6 GPixel/s |

| Texture Rate | 283.2 GTexel/s | 224.4 GTexel/s |

| FP32 | 9.062 TFLOPS | 7.181 TFLOPS |

| FP16 | 18.12 TFLOPS (2:1) | 14.36 TFLOPS (2:1) |

| TDP | 215 W | 175 W |

| Power Connectors | 1x 6-pin + 1x 8-pin | 1x 8-pin |

| Suggested PSU | 550 W | 450 W |

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

| Length | 267 mm | 229 mm |

| Height | 116 mm | 113 mm |

| Launch MSRP | 499 USD | 399 USD |

The two cards share the same architecture (Turing), process node (12 nm), foundry (TSMC), memory size (8 GB), memory type (GDDR6), bus width (256 bit), memory clock (1750 MHz), bandwidth (448.0 GB/s), ROPs (64), slot width (Dual-slot), bus interface (PCIe 3.0 x16), and API support (DirectX 12 Ultimate, OpenGL 4.6, Vulkan 1.4). They were also released on the same date, share the same predecessor (GeForce 10) and successor (GeForce 30), and both are marked as end-of-life.

The Verdict

The data is unambiguous for most users: the NVIDIA GeForce RTX 2070 SUPER is the stronger card across the vast majority of workloads. Its average benchmark score of 20282 is 12.1% higher than the 2060 SUPER’s 18093, and it wins 9 out of 10 head-to-head tests. For gaming, 3D rendering, and compute tasks, the 2070 SUPER’s higher shading unit count (2560 vs. 2176), faster boost clock (1770 MHz vs. 1650 MHz), and greater FP32 throughput (9.062 TFLOPS vs. 7.181 TFLOPS) translate directly into measurable performance gains. Its 17.1% lead in Geekbench Vulkan and 16.2% lead in PassMark DirectX 11 suggest that it will hold up better in both modern and legacy API workloads.

The 2060 SUPER is not without its merits. Its single win in 3DMark Steel Nomad DX12 is notable, and it achieves this with a lower TDP of 175 W versus 215 W, requiring only a 450 W power supply instead of 550 W. It is also 38 mm shorter, making it easier to fit into compact builds. However, its performance deficit in every other test, particularly the 18.9% and 16.2% losses in PassMark DirectX 10 and 11, means that it is only the better choice for users who prioritize power efficiency, smaller physical footprint, or the specific Steel Nomad workload. The 2060 SUPER’s launch MSRP of 399 USD versus 499 USD for the 2070 SUPER reflects this positioning, but the performance gap is substantial enough that the 2070 SUPER is the clear pick for uncompromising performance. The 2060 SUPER is the pragmatic alternative for constrained systems, while the 2070 SUPER is the definitive answer for those who want the best Turing-based performance in this comparison.