NVIDIA GeForce RTX 2080 Ti vs NVIDIA GeForce RTX 3090 Comparison

FAQ

Q: Which card wins the majority of head-to-head benchmarks?

A: The NVIDIA GeForce RTX 3090 wins 9 of the 10 head-to-head tests, with the only exception being Geekbench Vulkan, where the RTX 2080 Ti takes the win by a 146.5% margin.

Q: How large is the RTX 3090's advantage in the 3DMark Steel Nomad DX12 test?

A: The RTX 3090 scores 5,118 versus the RTX 2080 Ti's 3,537, a delta of -30.9% from the perspective of the older card — meaning the RTX 3090 is roughly 31% ahead in that test.

Q: Does the RTX 2080 Ti have any compute advantage?

A: In Geekbench Vulkan, yes — the RTX 2080 Ti scores 132,930 versus 53,927 for the RTX 3090, a 146.5% lead. However, in Geekbench OpenCL the RTX 3090 wins decisively with 172,758 versus 128,171.

Q: What about memory capacity and bandwidth differences?

A: The RTX 3090 has 24 GB of GDDR6X on a 384-bit bus with 936.2 GB/s bandwidth, while the RTX 2080 Ti has 11 GB of GDDR6 on a 352-bit bus with 616.0 GB/s bandwidth.

Q: Are both cards still in production?

A: No. Both the RTX 2080 Ti and RTX 3090 are listed as end-of-life products in the data.

Q: How do their overall percentile rankings compare?

A: The RTX 2080 Ti sits at the 75th percentile among all GPUs, while the RTX 3090 is at the 73rd percentile — despite the RTX 3090 winning nearly all direct comparisons.

Architecture Differences

The RTX 2080 Ti is built on NVIDIA's Turing architecture using the TU102 chip, fabricated on TSMC's 12 nm process. It packs 18,600 million transistors across a 754 mm² die, yielding a transistor density of 24.7 million per square millimeter. The RTX 3090 moves to the Ampere architecture with the GA102 chip, manufactured by Samsung on an 8 nm process. It contains 28,300 million transistors on a smaller 628 mm² die, giving a much higher density of 45.1 million per square millimeter.

The two cards diverge significantly in their compute resources. The RTX 2080 Ti has 4,352 shading units, 272 texture mapping units, and 88 ROPs. The RTX 3090 nearly doubles the shading units to 10,496, while also increasing TMUs to 328 and ROPs to 112. Ray tracing cores jump from 68 on the Turing card to 82 on Ampere, and tensor cores go from 544 to 328 — though the Ampere tensor cores are architecturally different, the raw count is lower.

Clock speeds tell a partial story: the RTX 2080 Ti has a 1350 MHz base and 1545 MHz boost, while the RTX 3090 runs at 1395 MHz base and 1695 MHz boost. The memory subsystem also differs fundamentally — GDDR6 on a 352-bit bus for the older card versus GDDR6X on a 384-bit bus for the newer one. The RTX 2080 Ti supports PCIe 3.0 x16, whereas the RTX 3090 uses PCIe 4.0 x16. Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Physical design changes are substantial. The RTX 2080 Ti is a dual-slot card measuring 267 mm in length, 116 mm in height, and 35 mm in width, drawing power from two 8-pin connectors. The RTX 3090 is a triple-slot card at 336 mm by 140 mm by 61 mm, using a single 12-pin connector. The RTX 2080 Ti includes a USB Type-C output alongside 1x HDMI 2.0 and 3x DisplayPort 1.4a, while the RTX 3090 offers 1x HDMI 2.1 and 3x DisplayPort 1.4a.

Where Each One Wins

The RTX 3090 dominates in almost every measured workload. Its largest victory comes in Passmark GPU Compute, where it scores 15,356 against 10,056 — a 34.5% margin. This points to raw compute throughput as the Ampere card's primary strength, consistent with its 35.58 TFLOPS FP32 versus 13.45 TFLOPS for the RTX 2080 Ti. The RTX 3090 also wins clearly in 3DMark Steel Nomad DX12 (5,118 vs 3,537, a 30.9% lead), indicating superior modern DirectX 12 gaming performance.

The RTX 3090 further takes Geekbench OpenCL (172,758 vs 128,171, a 25.8% lead), Passmark DirectX 12 (110 vs 84, a 23.6% lead), and Passmark G3D (26,645 vs 21,548, a 19.1% lead). In legacy DirectX tests, the RTX 3090 wins by smaller margins: DirectX 11 at 220 vs 190 (13.6%), DirectX 10 at 182 vs 153 (15.9%), and DirectX 9 at 268 vs 235 (12.3%). Even in 2D performance, the RTX 3090 leads with 1,063 versus 927 in Passmark G2D.

The RTX 2080 Ti's sole victory is striking: Geekbench Vulkan at 132,930 versus 53,927, a 146.5% advantage. This suggests that in Vulkan-specific workloads, the Turing architecture retains a significant edge — likely reflecting driver or architectural optimizations that favor the older design in that particular API path. For users running Vulkan-based applications or games, the RTX 2080 Ti would be the better choice based on this data.

Specification Differences

The two cards differ on nearly every specification line. Process node moves from 12 nm (TSMC) to 8 nm (Samsung), with transistor count rising from 18,600 million to 28,300 million. Die size actually shrinks from 754 mm² to 628 mm², while transistor density nearly doubles from 24.7M/mm² to 45.1M/mm².

Memory specifications are substantially different: 11 GB GDDR6 versus 24 GB GDDR6X, 352-bit versus 384-bit bus, 616.0 GB/s versus 936.2 GB/s bandwidth, and 1750 MHz (14 Gbps effective) versus 1219 MHz (19.5 Gbps effective) memory clocks. Shading units jump from 4,352 to 10,496, TMUs from 272 to 328, and ROPs from 88 to 112. Ray tracing cores increase from 68 to 82, while tensor cores decrease from 544 to 328.

Compute throughput metrics all favor the RTX 3090: FP32 at 35.58 TFLOPS versus 13.45 TFLOPS, FP16 at 35.58 TFLOPS (1:1) versus 26.90 TFLOPS (2:1), pixel rate at 189.8 GPixel/s versus 136.0 GPixel/s, and texture rate at 556.0 GTexel/s versus 420.2 GTexel/s. Power draw rises from 250 W to 350 W, with suggested PSU going from 600 W to 750 W. The bus interface upgrades from PCIe 3.0 x16 to PCIe 4.0 x16.

Physical dimensions grow considerably: length from 267 mm to 336 mm, height from 116 mm to 140 mm, and width from 35 mm to 61 mm. The slot width expands from dual-slot to triple-slot, and power connectors change from 2x 8-pin to 1x 12-pin. Display outputs change from 1x HDMI 2.0, 3x DisplayPort 1.4a, and 1x USB Type-C to 1x HDMI 2.1 and 3x DisplayPort 1.4a.

Head-to-Head Benchmarks

The benchmark data tells a consistent story with one dramatic outlier. In Passmark GPU Compute, the RTX 3090 leads by the widest margin at 34.5% (15,356 vs 10,056). This aligns with its much higher FP32 throughput and suggests the Ampere card is substantially better suited for compute-heavy tasks like rendering or scientific workloads.

The 3DMark Steel Nomad DX12 result shows a 30.9% advantage for the RTX 3090 (5,118 vs 3,537), indicating that modern DirectX 12 gaming performance strongly favors the newer architecture. Geekbench OpenCL follows at 25.8% (172,758 vs 128,171), reinforcing the compute-oriented strength of the RTX 3090.

Passmark DirectX 12 shows a 23.6% lead for the RTX 3090 (110 vs 84), while Passmark G3D demonstrates a 19.1% advantage (26,645 vs 21,548). The older DirectX versions show smaller but still clear wins: DirectX 11 at 13.6% (220 vs 190), DirectX 10 at 15.9% (182 vs 153), and DirectX 9 at 12.3% (268 vs 235). Even Passmark G2D favors the RTX 3090 by 12.8% (1,063 vs 927).

The outlier is Geekbench Vulkan, where the RTX 2080 Ti wins by an enormous 146.5% (132,930 vs 53,927). This is the only test where the older card wins, but the margin is so large that it cannot be dismissed as noise. For Vulkan-specific workloads, the RTX 2080 Ti appears to have a fundamental architectural advantage in this benchmark.

The average benchmark score also favors the RTX 3090: 27,565 versus 29,783 for the RTX 2080 Ti — wait, that is reversed. The RTX 2080 Ti has a slightly higher average benchmark score of 29,783 compared to the RTX 3090's 27,565. This is notable given that the RTX 3090 wins 9 of 10 head-to-head tests, suggesting the RTX 2080 Ti's massive Vulkan win inflates its average more than the RTX 3090's consistent but smaller wins across the other tests.

The Verdict

The data points overwhelmingly toward the RTX 3090 for most use cases. It wins 9 of 10 head-to-head benchmarks, with margins ranging from 12.3% in Passmark DirectX 9 to 34.5% in Passmark GPU Compute. Its 24 GB of GDDR6X memory and 936.2 GB/s bandwidth are double the capacity and over 50% more bandwidth than the RTX 2080 Ti. For modern DirectX 12 gaming, compute workloads, and OpenCL applications, the RTX 3090 is clearly the superior card.

However, the RTX 2080 Ti deserves consideration for one specific scenario: Vulkan workloads. Its 146.5% advantage in Geekbench Vulkan is the single largest margin in any comparison. Users running Vulkan-based applications or games that rely heavily on that API may find the older card performs better, despite its older architecture and lower raw specifications.

The percentile rankings complicate the picture: the RTX 2080 Ti sits at the 75th percentile among all GPUs, while the RTX 3090 is at the 73rd. This, combined with the RTX 2080 Ti's higher average benchmark score (29,783 vs 27,565), suggests that the RTX 2080 Ti's overall standing in the broader GPU landscape is slightly higher than the RTX 3090's, even though the RTX 3090 dominates their direct comparison.

For buyers choosing between these two, the RTX 3090 is the default recommendation for gaming and general compute. The RTX 2080 Ti is only advisable if Vulkan performance is the primary requirement — and even then, the 146.5% lead in that single test must be weighed against the RTX 3090's wins in every other category. The RTX 3090 is also the only one of the two with 24 GB memory, making it more future-proof for large datasets and high-resolution textures. Ultimately, the RTX 3090 is the more capable card in nearly all measurable respects, with the RTX 2080 Ti holding a narrow but significant niche in Vulkan.