NVIDIA GeForce RTX 3090 Ti vs NVIDIA GeForce RTX 5090 D Comparison

Head-to-Head Benchmarks

The benchmark data is unambiguous: the NVIDIA GeForce RTX 5090 D wins all three recorded head-to-head tests by substantial margins. In 3DMark Steel Nomad DX12, the RTX 5090 D scores 14,326 points against the RTX 3090 Ti's 5,741 points, a delta of 59.9% in favor of the newer card. This is the largest relative gap in the comparison and indicates a generational leap in raw rasterization performance under this workload.

Geekbench results follow the same pattern. In the OpenCL test, the RTX 5090 D posts 310,674 against 174,441 for the RTX 3090 Ti, a 43.9% advantage. The Vulkan test shows a similar spread: 376,915 versus 215,633, a 42.8% lead. Across all three recorded benchmarks, the RTX 5090 D wins every single test, giving it a perfect 3-0 record in the head-to-head comparison.

These deltas matter beyond simple wins. A 59.9% lead in 3DMark Steel Nomad represents more than just a faster card; it suggests a fundamentally different performance tier. The RTX 3090 Ti's average benchmark score of 131,938 places it at the 95th percentile of all GPUs in the database, with nearest rivals such as the AMD Radeon PRO W6800 (135,396) and NVIDIA RTX 4000 Ada Generation (135,218) sitting within 2.6% and 2.4% respectively. The RTX 5090 D, by contrast, carries an average benchmark score of 77,712 and sits at the 92nd percentile, though this figure is skewed by its inclusion of multiple PassMark sub-tests that drag the average downward.

The PassMark data for the RTX 5090 D, while not directly comparable to the RTX 3090 Ti (which lacks those entries), shows internal consistency: G3D score of 44,065, GPU compute of 28,396, and DirectX 11 at 371. These numbers, combined with the Geekbench and 3DMark results, paint a picture of a card that dominates its predecessor across every measurable dimension included in the database.

Architecture Differences

The two cards represent entirely different architectural eras. The RTX 3090 Ti uses the GA102 chip built on Ampere architecture, fabricated by Samsung on an 8 nm process. It integrates 28,300 million transistors across a 628 mm² die, yielding a transistor density of 45.1 million per square millimeter. The RTX 5090 D moves to the GB202 chip under Blackwell 2.0 architecture, produced by TSMC on a 5 nm node. This newer chip packs 92,200 million transistors into a 750 mm² die, achieving a density of 122.9 million transistors per square millimeter, nearly triple the density of the older part.

Memory configuration differs sharply. The RTX 3090 Ti uses 24 GB of GDDR6X on a 384-bit bus, delivering 1.01 TB/s of bandwidth. The RTX 5090 D expands to 32 GB of GDDR7 on a 512-bit bus, pushing bandwidth to 1.79 TB/s, a 77% increase. Effective memory speed jumps from 21 Gbps to 28 Gbps, and the memory clock rises from 1313 MHz to 1750 MHz.

The compute resources scale accordingly. Shading units increase from 10,752 to 21,760, texture mapping units from 336 to 680, and render output units from 112 to 176. Ray tracing cores double from 84 to 170, and tensor cores double from 336 to 680. Raw throughput figures reflect these changes: FP32 performance climbs from 40.00 TFLOPS to 104.8 TFLOPS, texture rate from 625.0 GTexel/s to 1,636.8 GTexel/s, and pixel rate from 208.3 GPixel/s to 423.6 GPixel/s.

Clock speeds also differ significantly. The RTX 3090 Ti runs at a base of 1560 MHz and boost of 1860 MHz. The RTX 5090 D starts higher at 2017 MHz base and reaches 2407 MHz boost, a 547 MHz advantage at boost. Power consumption rises from 450 W TDP on the RTX 3090 Ti to 575 W on the RTX 5090 D, with suggested PSU ratings climbing from 850 W to 950 W. The newer card is also physically smaller in every dimension: 304 mm length versus 336 mm, 137 mm height versus 140 mm, and 48 mm width versus 61 mm, while simultaneously shifting from a triple-slot to a dual-slot design.

Connectivity improves as well. The RTX 3090 Ti uses PCIe 4.0 x16 and outputs 1x HDMI 2.1 with 3x DisplayPort 1.4a. The RTX 5090 D moves to PCIe 5.0 x16 and offers 1x HDMI 2.1b with 3x DisplayPort 2.1b. Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The RTX 3090 Ti is end-of-life, launched in January 2022, while the RTX 5090 D remains active, released in January 2025.

FAQ

Q: Which card is faster in 3DMark Steel Nomad DX12?

A: The RTX 5090 D scores 14,326 points versus 5,741 points for the RTX 3090 Ti, giving it a 59.9% advantage in this test.

Q: How much more memory does the RTX 5090 D offer?

A: The RTX 5090 D has 32 GB of GDDR7 memory on a 512-bit bus, compared to 24 GB of GDDR6X on a 384-bit bus for the RTX 3090 Ti. Bandwidth increases from 1.01 TB/s to 1.79 TB/s.

Q: What is the transistor count difference between the two GPUs?

A: The RTX 5090 D contains 92,200 million transistors on a 750 mm² die, while the RTX 3090 Ti contains 28,300 million transistors on a 628 mm² die. This represents a transistor density of 122.9M per mm² versus 45.1M per mm².

Q: Which card has higher boost clock speeds?

A: The RTX 5090 D boosts to 2407 MHz, while the RTX 3090 Ti boosts to 1860 MHz. The base clocks are 2017 MHz and 1560 MHz respectively.

Q: Does the RTX 5090 D generate more heat or require more power?

A: Yes. The RTX 5090 D has a TDP of 575 W with a suggested PSU of 950 W, versus 450 W TDP and 850 W suggested PSU for the RTX 3090 Ti.

Q: Are there any benchmark categories where the RTX 3090 Ti wins?

A: No. In the recorded head-to-head tests, the RTX 5090 D wins all three: 3DMark Steel Nomad DX12, Geekbench OpenCL, and Geekbench Vulkan.

The Verdict

The data leads to a straightforward conclusion: the RTX 5090 D is the superior performer across every recorded benchmark. Its 59.9% lead in 3DMark Steel Nomad DX12 and 42.8-43.9% leads in Geekbench tests are decisive, not marginal. The architectural overhaul from Ampere to Blackwell 2.0, combined with the shift from Samsung 8 nm to TSMC 5 nm, delivers more than double the shading units, double the ray tracing cores, and 1.77 times the memory bandwidth. The RTX 5090 D also carries a larger memory pool at 32 GB versus 24 GB, which benefits workloads that exceed the older card's capacity.

However, the RTX 3090 Ti retains relevance in specific contexts. Its 95th percentile standing among all GPUs, with an average benchmark score of 131,938, places it above rivals like the NVIDIA RTX 4000 Ada Generation and AMD Radeon PRO W6800 by 2.4-2.6%. It remains a capable card for users already invested in the Ampere ecosystem, particularly those who prioritize the lower 450 W TDP and 850 W PSU requirement. The RTX 3090 Ti's end-of-life status and 1,999 USD launch MSRP (versus 2,299 USD for the RTX 5090 D) also factor into procurement decisions, though the performance gap is stark enough that new purchases should favor the RTX 5090 D unless power constraints or existing platform compatibility override raw performance needs.

Specification Differences

| Specification | RTX 3090 Ti | RTX 5090 D |

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

| Architecture | Ampere | Blackwell 2.0 |

| Process Node | 8 nm | 5 nm |

| Foundry | Samsung | TSMC |

| Transistors | 28,300 million | 92,200 million |

| Die Size | 628 mm² | 750 mm² |

| Transistor Density | 45.1M / mm² | 122.9M / mm² |

| Base Clock | 1560 MHz | 2017 MHz |

| Boost Clock | 1860 MHz | 2407 MHz |

| Memory Clock | 1313 MHz (21 Gbps effective) | 1750 MHz (28 Gbps effective) |

| Memory Size | 24 GB | 32 GB |

| Memory Type | GDDR6X | GDDR7 |

| Memory Bus Width | 384 bit | 512 bit |

| Memory Bandwidth | 1.01 TB/s | 1.79 TB/s |

| Shading Units | 10752 | 21760 |

| TMUs | 336 | 680 |

| ROPs | 112 | 176 |

| RT Cores | 84 | 170 |

| Tensor Cores | 336 | 680 |

| Pixel Rate | 208.3 GPixel/s | 423.6 GPixel/s |

| Texture Rate | 625.0 GTexel/s | 1,636.8 GTexel/s |

| FP32 | 40.00 TFLOPS | 104.8 TFLOPS |

| FP16 | 40.00 TFLOPS (1:1) | 104.8 TFLOPS (1:1) |

| TDP | 450 W | 575 W |

| Slot Width | Triple-slot | Dual-slot |

| Suggested PSU | 850 W | 950 W |

| Bus Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |

| Display Outputs | 1x HDMI 2.1, 3x DisplayPort 1.4a | 1x HDMI 2.1b, 3x DisplayPort 2.1b |

| Dimensions | 336 mm x 140 mm x 61 mm | 304 mm x 137 mm x 48 mm |

| Production Status | End-of-life | Active |

| Release Date | 2022-01-26 | 2025-01-29 |

Where Each One Wins

The RTX 5090 D wins in every benchmark category recorded in the database. Its 3DMark Steel Nomad DX12 score of 14,326 outperforms the RTX 3090 Ti by 59.9%, indicating superior DirectX 12 rasterization performance. Geekbench OpenCL and Vulkan results show 43.9% and 42.8% leads respectively, confirming strong compute and graphics API performance across the board. The card also offers double the FP32 throughput at 104.8 TFLOPS versus 40.00 TFLOPS, which translates to faster compute workloads in general-purpose GPU tasks. Its 32 GB GDDR7 memory with 1.79 TB/s bandwidth gives it a clear edge in memory-intensive applications such as large dataset processing, high-resolution texture streaming, and AI inference workloads that benefit from the doubled tensor core count.

The RTX 3090 Ti wins in the domain of efficiency relative to its performance tier. Its 450 W TDP is 125 W lower than the RTX 5090 D's 575 W, and its suggested PSU of 850 W is 100 W lower. The card is also physically larger at 336 mm length versus 304 mm, which may be a factor in cases designed for older, bulkier GPUs. Its 1,999 USD launch MSRP undercuts the RTX 5090 D's 2,299 USD. For users running legacy applications that do not leverage Blackwell-specific optimizations, or for systems with power delivery limits, the RTX 3090 Ti remains a viable option. Its 95th percentile standing among all GPUs means it still outperforms the vast majority of installed hardware, and its nearest rivals (NVIDIA L4, RTX 4000 Ada Generation, A10M, Radeon PRO W6800) sit within 2.6% of its average score, indicating that it remains competitive within its own generation.

For new builds or upgrades where performance is the primary criterion, the RTX 5090 D is the clear recommendation. For constrained environments where power draw and PSU requirements are limiting factors, the RTX 3090 Ti offers a meaningful alternative, though users should expect roughly 43-60% lower performance in the recorded workloads. The RTX 5090 D's active production status and newer release date also mean it will likely receive longer driver support, an advantage that cannot be quantified in the current benchmark data but follows from its 2025 launch versus the 2022 launch of the RTX 3090 Ti.