NVIDIA GeForce RTX 4090 vs NVIDIA GeForce RTX 5090 D Comparison

The NVIDIA GeForce RTX 5090 D and the RTX 4090 represent two distinct generations of flagship graphics hardware. The recorded data shows a decisive performance shift, with the newer card winning every single head-to-head benchmark in the database. This analysis examines the scale of that advantage, the architectural reasons behind it, and what the numbers imply for different usage scenarios.

Head-to-Head Benchmarks

The benchmark results are unambiguous: the RTX 5090 D wins all 10 recorded comparisons, with no test favoring the RTX 4090. The most dramatic gap appears in the 3DMark Steel Nomad DX12 test, a demanding modern workload. Here, the RTX 5090 D scores 14,326 points against the RTX 4090's 9,223, a delta of 55.3%. This is the largest relative difference in the entire dataset, suggesting that the newer architecture scales particularly well in DirectX 12 rendering.

The Vulkan API shows a similarly substantial lead. The RTX 5090 D posts 376,915 in Geekbench Vulkan, while the RTX 4090 manages 271,631, representing a 38.8% advantage. DirectX 12 in Passmark tells a comparable story, with the 5090 D scoring 219 versus 150, a 46% delta. These results indicate that the 5090 D's advantage is not limited to a single API; it consistently outperforms across modern graphics interfaces.

Compute-oriented workloads also favor the newer card, though by a smaller margin. In Geekbench OpenCL, the RTX 5090 D scores 310,674 against the RTX 4090's 255,416, a 21.6% improvement. Passmark GPU Compute shows 28,396 versus 26,613, a narrower 6.7% delta. This pattern suggests that while the 5090 D excels in rasterization and ray tracing, its compute advantage, while present, is less pronounced than its rendering lead.

Legacy DirectX performance reveals interesting behavior. In Passmark DirectX 10, the 5090 D leads by only 3.1% (231 vs 224). DirectX 9 shows a 9.3% gap (434 vs 397). These older API tests compress the performance difference, likely due to driver overhead and the diminishing returns of raw hardware resources on legacy workloads. DirectX 11 falls in between, with the 5090 D ahead by 13.8% (371 vs 326).

The remaining tests round out the picture. Passmark G3D, a general 3D performance metric, shows the 5090 D at 44,065 versus 38,194, a 15.4% lead. Passmark G2D, which measures 2D graphics operations, shows a 14.5% advantage (1,487 vs 1,299). Across the entire benchmark suite, the average score for the RTX 5090 D is 77,712, placing it in the 92nd percentile of all GPUs. The RTX 4090 averages 60,347, sitting in the 88th percentile.

FAQ

Q: Does the RTX 5090 D win every benchmark against the RTX 4090?

A: Yes. The database records 10 head-to-head comparisons, and the RTX 5090 D wins all 10. The RTX 4090 records zero wins in the dataset.

Q: What is the largest performance gap between the two cards?

A: The biggest delta is in 3DMark Steel Nomad DX12, where the RTX 5090 D scores 14,326 versus 9,223, a 55.3% advantage. The smallest gap is in Passmark DirectX 10, with a 3.1% delta.

Q: How do the cards compare in compute workloads?

A: The RTX 5090 D leads in Geekbench OpenCL by 21.6% (310,674 vs 255,416) and in Passmark GPU Compute by 6.7% (28,396 vs 26,613). The compute advantage is real but smaller than the rendering lead.

Q: Which card has higher memory bandwidth?

A: The RTX 5090 D has a memory bandwidth of 1.79 TB/s, while the RTX 4090 offers 1.01 TB/s. The 5090 D also has more memory capacity at 32 GB versus 24 GB.

Q: What are the power requirements for each card?

A: The RTX 5090 D has a TDP of 575 W and requires a suggested PSU of 950 W. The RTX 4090 has a TDP of 450 W with a suggested PSU of 850 W.

Q: Are the cards the same physical size?

A: Both cards are 304 mm long and 137 mm high. The RTX 5090 D is 48 mm wide and dual-slot, while the RTX 4090 is 61 mm wide and triple-slot.

Architecture Differences

The two GPUs are built on fundamentally different architectures. The RTX 5090 D uses the Blackwell 2.0 architecture with the GB202 chip, while the RTX 4090 relies on Ada Lovelace with the AD102 chip. Both are fabricated on a 5 nm process at TSMC, yet the transistor counts differ significantly. The GB202 packs 92,200 million transistors on a 750 mm² die, whereas the AD102 contains 76,300 million transistors on a 609 mm² die. Interestingly, the transistor density is slightly higher on the older chip: 125.3M per mm² for the RTX 4090 versus 122.9M per mm² for the RTX 5090 D.

The core configuration scales dramatically. The RTX 5090 D features 21,760 shading units, 680 texture mapping units, 176 ROPs, 170 RT cores, and 680 tensor cores. The RTX 4090 has 16,384 shading units, 512 TMUs, 176 ROPs, 128 RT cores, and 512 tensor cores. This represents a 32.8% increase in shading units and a 32.8% increase in both RT and tensor cores, while ROPs remain identical at 176.

Memory architecture presents another clear divergence. The RTX 5090 D uses 32 GB of GDDR7 memory on a 512-bit bus, achieving 1.79 TB/s bandwidth. The RTX 4090 uses 24 GB of GDDR6X on a 384-bit bus, reaching 1.01 TB/s. The newer card offers 33% more capacity and 77% more bandwidth, a substantial upgrade that likely contributes to its rendering advantage.

Clock speeds tell a nuanced story. The RTX 4090 has higher base and boost clocks at 2235 MHz and 2520 MHz respectively. The RTX 5090 D runs at 2017 MHz base and 2407 MHz boost, which is lower. However, the memory clock is much higher on the newer card: 1750 MHz (28 Gbps effective) versus 1313 MHz (21 Gbps effective). The performance lead, despite lower core clocks, underscores the architectural efficiency of Blackwell 2.0.

Interface and connectivity also differ. The RTX 5090 D uses PCIe 5.0 x16, while the RTX 4090 uses PCIe 4.0 x16. Display outputs are updated as well, with the 5090 D offering 1x HDMI 2.1b and 3x DisplayPort 2.1b, compared to the 4090's 1x HDMI 2.1 and 3x DisplayPort 1.4a. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Specification Differences

The recorded specifications reveal several key differences between the two cards. The RTX 5090 D has a TDP of 575 W versus 450 W for the RTX 4090, a 125 W increase. The suggested PSU scales accordingly, from 850 W to 950 W. Physical dimensions differ in width only: the 5090 D is 48 mm thick and dual-slot, while the 4090 is 61 mm thick and triple-slot. Length and height are identical at 304 mm and 137 mm respectively.

Memory specifications show the 5090 D with 32 GB GDDR7, 512-bit bus, and 1.79 TB/s bandwidth. The 4090 has 24 GB GDDR6X, 384-bit bus, and 1.01 TB/s bandwidth. The newer card has a higher memory clock at 1750 MHz versus 1313 MHz. Core counts are universally higher on the 5090 D except for ROPs, which are equal at 176.

The pixel rate is slightly lower on the 5090 D at 423.6 GPixel/s versus 443.5 GPixel/s for the 4090, likely due to the lower boost clock. However, texture rate is higher: 1,636.8 GTexel/s versus 1,290.2 GTexel/s. FP32 and FP16 performance are both 104.8 TFLOPS on the 5090 D, versus 82.58 TFLOPS on the 4090, a 26.9% increase.

Production status differs notably. The RTX 5090 D is listed as Active, while the RTX 4090 is End-of-life. Release dates are 2025-01-29 for the 5090 D and 2022-09-19 for the 4090. The launch MSRP for the 5090 D is 2,299 USD, and for the 4090 it is 1,599 USD.

The Verdict

The benchmark data strongly favors the RTX 5090 D across every measurable workload. With a 55.3% lead in 3DMark Steel Nomad DX12 and a 38.8% lead in Vulkan, the newer card is clearly the superior performer for modern graphics applications. Its average benchmark score of 77,712 versus 60,347 places it in the 92nd percentile of all GPUs, compared to the 4090's 88th percentile.

However, the verdict is not without nuance. The RTX 4090 maintains higher clock speeds (2235 MHz base, 2520 MHz boost) and a higher pixel rate (443.5 GPixel/s). It also has a lower TDP of 450 W versus 575 W, and a slimmer profile is not a factor since the 4090 is actually wider. For users constrained by power supply capacity, the 4090's 850 W suggested PSU is more accessible than the 5090 D's 950 W requirement.

The data also suggests diminishing returns in legacy workloads. In DirectX 10, the 5090 D leads by only 3.1%, and in DirectX 9 by 9.3%. This indicates that for older games or applications using these APIs, the performance difference is minor. The compute gap is also narrower, with only a 6.7% delta in Passmark GPU Compute.

The RTX 4090's end-of-life status and the 5090 D's active production status further reinforce the generational shift. The newer card offers more memory, higher bandwidth, and superior core counts, but requires more power and a larger PSU. The RTX 5090 D is the data-driven choice for state-of-the-art performance, while the RTX 4090 remains a capable option with lower power demands.

Where Each One Wins

The RTX 5090 D wins in all 10 recorded benchmarks, but the magnitude of its victories varies by workload. Its strongest showings are in modern, API-heavy tests: 3DMark Steel Nomad DX12 (55.3% ahead), Passmark DirectX 12 (46% ahead), and Geekbench Vulkan (38.8% ahead). These workloads leverage the full architectural advantages of Blackwell 2.0, including higher shading unit counts and faster memory.

The RTX 4090's relative strengths appear in legacy and compute scenarios where the gap narrows. In Passmark DirectX 10, the 4090 trails by just 3.1%, and in DirectX 9 by 9.3%. For users running older software titles, the performance difference becomes nearly negligible. The compute workloads also show a compressed delta: 21.6% in OpenCL and 6.7% in GPU Compute, suggesting the 4090's compute efficiency remains competitive.

The 4090 also holds advantages in power efficiency and physical design considerations. Its 450 W TDP and 850 W PSU requirement are lower than the 5090 D's 575 W and 950 W. Its triple-slot width is larger, but its lower power draw may suit systems with smaller PSUs. The 4090's higher base and boost clocks (2235 MHz and 2520 MHz) also indicate that it can sustain high frequencies, even if the newer architecture ultimately outperforms it.

For users prioritizing maximum frame rates in the latest DirectX 12 and Vulkan titles, the RTX 5090 D is the clear winner based on recorded data. For those with power constraints or running legacy applications, the RTX 4090 remains a viable choice, though the data shows it cannot match the 5090 D in any single benchmark. The choice ultimately depends on whether the user values the 5090 D's substantial modern performance lead or the 4090's lower power requirements and tighter legacy performance gap.