NVIDIA GeForce RTX 2080 SUPER vs NVIDIA GeForce RTX 3080 Comparison

The Verdict

The data separates these two cards cleanly. The NVIDIA GeForce RTX 3080 wins 9 of 10 head-to-head benchmarks, with its largest victories coming in modern DX12 workloads and compute. The RTX 2080 SUPER takes a single but spectacular win in Vulkan, where it leads by 231%. If you are building for current-generation gaming at high resolutions, the RTX 3080 is the obvious choice from the benchmark data alone. It delivers a 57.3% higher score in 3DMark Steel Nomad DX12, a 42.4% higher Passmark GPU compute score, and a 22.3% higher Passmark G3D score. The RTX 2080 SUPER is not without merit, but its one big win is in a legacy API scenario that matters less for future-proofing. The RTX 3080's average benchmark score of 23172 sits slightly below the 2080 SUPER's 24170, but that average is dragged down by the anomalous Vulkan result. In every other test, the 3080 leads by margins ranging from 12% to 57.3%.

For a builder choosing between these two end-of-life cards, the RTX 3080 is the pick for anyone prioritizing raw throughput, ray tracing, or compute-heavy tasks. The RTX 2080 SUPER remains viable only if you specifically need Vulkan performance or want to save the extra power draw and physical length — the 3080 runs 320 W versus 250 W and measures 285 mm versus 267 mm. Both cards launched at the same 699 USD MSRP, so the data shows the 3080 as the generational upgrade in nearly every measurable way.

Where Each One Wins

The RTX 3080 dominates the modern workload categories. In 3DMark Steel Nomad DX12, it scores 4407 against the 2080 SUPER's 1882 — a 57.3% advantage that reflects the Ampere architecture's strength in DirectX 12 Ultimate titles. The compute gap is equally stark: Passmark GPU compute shows 14397 versus 8290, a 42.4% delta. OpenCL performance follows the same pattern, with the 3080 hitting 152423 versus 99226, a 34.9% lead. These are the numbers that matter for 1440p and 4K gaming, where shader complexity and compute load scale with resolution.

Legacy DirectX performance also favors the 3080, though by smaller margins. Passmark DX11 shows 207 versus 165 (20.3% lead), DX12 shows 100 versus 75 (25% lead), and even DX9 shows 258 versus 227 (12% lead). The pattern is consistent: the 3080 is faster across every API generation except Vulkan.

The RTX 2080 SUPER's single win is in Geekbench Vulkan, where it scores 111284 against the 3080's 33620. That 231% delta is massive and suggests driver or architectural optimization for Vulkan workloads in the older Turing design. However, this is an outlier in an otherwise one-sided comparison. For real-world gaming, the Vulkan benchmark does not translate into a general advantage — it is a specific API test where the 2080 SUPER's driver stack and Turing compute units happen to excel. The 2080 SUPER also holds a higher percentile ranking at 69 versus the 3080's 68, but that difference is within noise given the 3080's lower average score is skewed by the Vulkan result.

Architecture Differences

The two cards come from different process nodes and foundries. The RTX 2080 SUPER uses the TU104 chip on TSMC's 12 nm process, packing 13,600 million transistors into a 545 mm² die. The RTX 3080 uses the GA102 chip on Samsung's 8 nm node, with 28,300 million transistors across a 628 mm² die. That is more than double the transistor count on a slightly larger die, yielding a transistor density of 45.1M per mm² versus 25.0M per mm². The architectural shift from Turing to Ampere is the core difference, and it shows in the specs.

Shading units jump from 3072 on the 2080 SUPER to 8704 on the 3080 — a 2.83x increase. Texture mapping units go from 192 to 272, and ROPs from 64 to 96. Ray tracing cores increase from 48 to 68, while tensor cores drop from 384 to 272 but gain efficiency in the Ampere design. FP32 performance nearly triples, from 11.15 TFLOPS to 29.77 TFLOPS. Notably, the 3080 runs FP16 at the same 29.77 TFLOPS (1:1 ratio), while the 2080 SUPER's FP16 is 22.30 TFLOPS at a 2:1 ratio — meaning the 3080 does not rely on rate doubling for half-precision work.

Memory differs substantially. The 2080 SUPER has 8 GB of GDDR6 on a 256-bit bus, delivering 495.9 GB/s bandwidth. The 3080 has 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s. That is a 53.4% bandwidth advantage for the 3080. Clock speeds are lower on the 3080 (1440 MHz base, 1710 MHz boost) versus the 2080 SUPER (1650 MHz base, 1815 MHz boost), but the much wider memory bus and higher shader count more than compensate.

Interface and power delivery also differ. The 3080 uses PCIe 4.0 x16, while the 2080 SUPER uses PCIe 3.0 x16. Power connectors are a single 12-pin on the 3080 versus 1x 6-pin + 1x 8-pin on the 2080 SUPER. The 3080's suggested PSU is 700 W versus 600 W for the 2080 SUPER. Display outputs are nearly identical, but the 3080 has HDMI 2.1 while the 2080 SUPER has HDMI 2.0 plus a USB Type-C port that the 3080 lacks.

FAQ

Q: Which card is faster in DirectX 12 Ultimate workloads?

A: The RTX 3080 wins decisively. In 3DMark Steel Nomad DX12, it scores 4407 versus 1882, a 57.3% advantage. Passmark DX12 also shows the 3080 ahead at 100 versus 75.

Q: Does the RTX 2080 SUPER beat the RTX 3080 in anything?

A: Yes, in Geekbench Vulkan. The 2080 SUPER scores 111284 against the 3080's 33620, a 231% lead. This is the only benchmark where the older card wins.

Q: How do the memory systems compare?

A: The 3080 has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. The 2080 SUPER has 8 GB of GDDR6 on a 256-bit bus with 495.9 GB/s. The 3080 delivers 53.4% more bandwidth.

Q: Which card has better compute performance?

A: The RTX 3080 by a wide margin. Passmark GPU compute shows 14397 versus 8290, a 42.4% lead. OpenCL also favors the 3080 at 152423 versus 99226, a 34.9% delta.

Q: What are the power requirements?

A: The 3080 has a 320 W TDP and requires a 700 W suggested PSU with a single 12-pin connector. The 2080 SUPER has a 250 W TDP and requires a 600 W suggested PSU with 1x 6-pin + 1x 8-pin connectors.

Q: Do both cards support the same APIs?

A: Yes, both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The 3080 adds HDMI 2.1 output, while the 2080 SUPER has a USB Type-C port.

Head-to-Head Benchmarks

The biggest win for the RTX 3080 comes in 3DMark Steel Nomad DX12, where it scores 4407 against the 2080 SUPER's 1882. That 57.3% delta is the largest across all modern benchmarks and reflects the Ampere architecture's fundamental advantage in DX12. The 3080's 8704 shading units versus 3072 explain much of this — more than double the compute units processing the same workload.

Passmark GPU compute shows the second-largest gap at 42.4%. The 3080 scores 14397, the 2080 SUPER 8290. This test stresses raw FP32 throughput, where the 3080's 29.77 TFLOPS versus 11.15 TFLOPS gives it a 2.67x theoretical advantage that translates to the measured delta. OpenCL follows closely at 34.9% (152423 versus 99226), confirming the compute dominance.

Passmark G3D shows a 22.3% lead for the 3080 (25086 versus 19490). This is the broadest gaming-oriented metric, and the margin is consistent with the DX11 and DX12 results. Passmark DX12 shows 100 versus 75 (25% lead), DX11 shows 207 versus 165 (20.3% lead), and even the older DX10 and DX9 tests favor the 3080 by 17.6% and 12% respectively. The 2D test also goes to the 3080 at 1054 versus 920, a 12.7% margin.

The one reverse result — Geekbench Vulkan — is striking. The 2080 SUPER scores 111284, the 3080 just 33620. That 231% delta is the largest in either direction across all benchmarks. It suggests the Turing architecture's Vulkan driver path is significantly more optimized than Ampere's early implementation, or that the specific test workload favors the 2080 SUPER's tensor core layout and memory subsystem. Regardless, this single result does not offset the 3080's wins in nine other tests.

Specification Differences

The table below lists only the fields where the two cards differ.

| Specification | RTX 2080 SUPER | RTX 3080 |

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

| Architecture | Turing | Ampere |

| Process Node | 12 nm (TSMC) | 8 nm (Samsung) |

| Transistors | 13,600 million | 28,300 million |

| Die Size | 545 mm² | 628 mm² |

| Transistor Density | 25.0M / mm² | 45.1M / mm² |

| Base Clock | 1650 MHz | 1440 MHz |

| Boost Clock | 1815 MHz | 1710 MHz |

| Memory Clock | 1937 MHz (15.5 Gbps) | 1188 MHz (19 Gbps) |

| Memory Size | 8 GB | 10 GB |

| Memory Type | GDDR6 | GDDR6X |

| Memory Bus | 256 bit | 320 bit |

| Memory Bandwidth | 495.9 GB/s | 760.3 GB/s |

| Shading Units | 3072 | 8704 |

| TMUs | 192 | 272 |

| ROPs | 64 | 96 |

| RT Cores | 48 | 68 |

| Tensor Cores | 384 | 272 |

| Pixel Rate | 116.2 GPixel/s | 164.2 GPixel/s |

| Texture Rate | 348.5 GTexel/s | 465.1 GTexel/s |

| FP32 | 11.15 TFLOPS | 29.77 TFLOPS |

| FP16 | 22.30 TFLOPS (2:1) | 29.77 TFLOPS (1:1) |

| TDP | 250 W | 320 W |

| Power Connectors | 1x 6-pin + 1x 8-pin | 1x 12-pin |

| Suggested PSU | 600 W | 700 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x16 |

| Display Outputs | 1x HDMI 2.0, 3x DP 1.4a, 1x USB-C | 1x HDMI 2.1, 3x DP 1.4a |

| Length | 267 mm (10.5 in) | 285 mm (11.2 in) |

| Height | 116 mm (4.6 in) | 112 mm (4.4 in) |

| Width | 35 mm (1.4 in) | 40 mm (1.6 in) |

| Release Date | 2019-07-22 | 2020-08-31 |

| Successor | GeForce 30 | GeForce 40 |