NVIDIA GeForce RTX 5080 vs NVIDIA GeForce RTX 5090 Comparison

The NVIDIA GeForce RTX 5090 and RTX 5080 are both Blackwell 2.0 architecture GPUs, but they are positioned very differently in the GeForce 50-series lineup. The recorded benchmark data shows a clear hierarchy, with the RTX 5090 winning 9 of 10 head-to-head tests, while the RTX 5080 takes a single narrow victory. This split highlights distinct strengths: the RTX 5090 is the dominant force for raw compute and modern DirectX 12 workloads, while the RTX 5080 offers a more balanced profile with a slight edge in 2D performance.

Where Each One Wins

The RTX 5090 is the unequivocal winner in virtually every compute-heavy and 3D workload. In the 3DMark Steel Nomad DX12 test, it scores 18,355 points versus 8,637 points for the RTX 5080, a massive 112.5% advantage. This is the single largest performance gap in the entire dataset, making the RTX 5090 the clear choice for high-end gaming and demanding 3D rendering tasks that rely on DirectX 12. The lead extends to GPU compute as well: in Passmark GPU Compute, the RTX 5090 scores 26,756 against 21,789, a 22.8% margin, and in Geekbench OpenCL it posts 334,370 versus 235,901, a 41.7% lead. For users running scientific simulations, machine learning inference, or any workload that stresses raw FP32 throughput, the RTX 5090 is the superior option.

The RTX 5090 also dominates in legacy and API-specific tests, though with smaller margins. It wins Passmark DirectX 12 by 22.5% (185 vs 151), DirectX 10 by 8.7% (226 vs 208), DirectX 11 by 5.2% (341 vs 324), and DirectX 9 by just 1.5% (395 vs 389). The DirectX 9 gap is negligible, suggesting that older titles or lightweight workloads will not see a meaningful difference between the two cards. In Passmark G3D, the RTX 5090 scores 39,650 versus 36,565, an 8.4% advantage, which reflects its stronger overall 3D rasterization performance.

The RTX 5080's sole victory comes in Passmark G2D, where it scores 1,415 versus 1,413 for the RTX 5090. This is a razor-thin 0.1% margin, effectively a tie, but it does indicate that the RTX 5080 is not at a disadvantage in 2D desktop workloads, image blitting, or window management. The RTX 5080 also holds a higher percentile rank among all GPUs at 87, compared to 92 for the RTX 5090, but that reflects its position in the broader market, not a head-to-head advantage. For buyers who prioritize 3D performance, compute throughput, or modern API support, the RTX 5090 is the clear pick; for those who only care about 2D desktop responsiveness, the RTX 5080 is essentially equivalent.

Architecture Differences

Both GPUs are built on the Blackwell 2.0 architecture and use the same 5 nm process from TSMC, but they are fundamentally different chips. The RTX 5090 uses the GB202 chip, while the RTX 5080 uses the GB203 chip. This distinction drives nearly every specification difference. The GB202 is a much larger die at 750 mm², compared to 378 mm² for the GB203, and it packs 92,200 million transistors versus 45,600 million. Transistor density is similar, at 122.9 million per mm² for the RTX 5090 and 120.6 million per mm² for the RTX 5080, indicating that the RTX 5090's advantage comes from sheer die size rather than a denser design.

The RTX 5090 doubles the RTX 5080's core counts across the board. It has 21,760 shading units, 680 TMUs, and 176 ROPs, while the RTX 5080 has 10,752 shading units, 336 TMUs, and 112 ROPs. The ray tracing and tensor core counts follow the same pattern: 170 RT cores and 680 tensor cores on the RTX 5090, versus 84 RT cores and 336 tensor cores on the RTX 5080. This doubling in hardware resources directly explains the RTX 5090's commanding lead in compute-oriented benchmarks like Geekbench OpenCL and Vulkan.

Memory architecture is another major divergence. The RTX 5090 features 32 GB of GDDR7 memory on a 512-bit bus, delivering 1.79 TB/s of bandwidth. The RTX 5080 has 16 GB of GDDR7 on a 256-bit bus, with 960.0 GB/s of bandwidth. Both use the same memory type, but the RTX 5090's wider bus and larger capacity give it a significant edge in bandwidth-bound workloads, such as high-resolution textures or large dataset processing. Clock speeds differ as well: the RTX 5080 runs at a higher base clock of 2295 MHz and boost of 2617 MHz, compared to 2017 MHz and 2407 MHz for the RTX 5090. Despite the lower clocks, the RTX 5090's massive core count produces far higher peak throughput.

Power and physical specifications also separate the two. The RTX 5090 has a TDP of 575 W and requires a suggested 950 W PSU, while the RTX 5080 draws 360 W and needs a 750 W PSU. Both are dual-slot cards with identical dimensions (304 mm length, 137 mm height, 40 mm width) and use a single 16-pin power connector, plus the same display outputs: 1x HDMI 2.1b and 3x DisplayPort 2.1b. The RTX 5090's higher power draw is a direct consequence of its larger die and higher transistor count, but it also means more demanding cooling and power delivery requirements.

FAQ

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

A: The RTX 5090 is dramatically faster, scoring 18,355 versus 8,637 for the RTX 5080, a 112.5% advantage.

Q: Does the RTX 5080 win any benchmark?

A: Yes, it wins Passmark G2D by a tiny margin, 1,415 versus 1,413, a 0.1% difference. It loses the other 9 head-to-head tests.

Q: How much memory do the two cards have?

A: The RTX 5090 has 32 GB of GDDR7 on a 512-bit bus, while the RTX 5080 has 16 GB of GDDR7 on a 256-bit bus.

Q: What is the power consumption difference?

A: The RTX 5090 has a TDP of 575 W and suggests a 950 W PSU, while the RTX 5080 has a TDP of 360 W and suggests a 750 W PSU.

Q: Are the cards the same physical size?

A: Yes, both are dual-slot cards with identical dimensions of 304 mm length, 137 mm height, and 40 mm width.

Q: Which card has better compute performance?

A: The RTX 5090 leads in Geekbench OpenCL by 41.7% (334,370 vs 235,901) and in Passmark GPU Compute by 22.8% (26,756 vs 21,789).

Specification Differences

The two cards differ in nearly every major specification, reflecting their distinct positions in the lineup. The RTX 5090 uses the GB202 chip, while the RTX 5080 uses the GB203 chip. The RTX 5090 has 92,200 million transistors on a 750 mm² die, versus 45,600 million on a 378 mm² die for the RTX 5080. Transistor density is nearly identical, at 122.9M per mm² for the RTX 5090 and 120.6M per mm² for the RTX 5080.

Core counts are doubled on the RTX 5090: 21,760 shading units, 680 TMUs, and 176 ROPs, versus 10,752 shading units, 336 TMUs, and 112 ROPs on the RTX 5080. The RTX 5090 also has 170 RT cores and 680 tensor cores, compared to 84 RT cores and 336 tensor cores on the RTX 5080. Clock speeds favor the RTX 5080, with a base of 2295 MHz and boost of 2617 MHz, versus 2017 MHz base and 2407 MHz boost on the RTX 5090.

Memory is a major differentiator: the RTX 5090 has 32 GB of GDDR7 with a 512-bit bus and 1.79 TB/s bandwidth, while the RTX 5080 has 16 GB of GDDR7 with a 256-bit bus and 960.0 GB/s bandwidth. Pixel rate and texture rate also reflect the core count advantage, with the RTX 5090 at 423.6 GPixel/s and 1,636.8 GTexel/s, versus 293.1 GPixel/s and 879.3 GTexel/s for the RTX 5080. FP32 and FP16 throughput are both 104.8 TFLOPS on the RTX 5090 and 56.28 TFLOPS on the RTX 5080.

Power requirements differ significantly: the RTX 5090 has a 575 W TDP and suggests a 950 W PSU, while the RTX 5080 has a 360 W TDP and suggests a 750 W PSU. Both cards are dual-slot, use a single 16-pin connector, and share identical dimensions and display outputs. Both support PCIe 5.0 x16, DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The launch MSRP for the RTX 5090 is 1,999 USD, and for the RTX 5080 it is 999 USD. Both were released on the same date and share the same predecessor (GeForce 40) and successor (GeForce 60).

Head-to-Head Benchmarks

The most striking result is in 3DMark Steel Nomad DX12, where the RTX 5090 scores 18,355 versus 8,637 for the RTX 5080, a 112.5% delta. This is the largest win for either card in the entire dataset, and it underscores the RTX 5090's suitability for next-generation DirectX 12 titles that can fully utilize its doubled core configuration. The RTX 5080's score is still respectable, but it trails by more than half, which is a decisive gap for users targeting high refresh rates at 4K or beyond.

In Geekbench Vulkan, the RTX 5090 posts 376,728 against 255,450, a 47.5% lead. This test stresses compute and async execution, and the RTX 5090's 680 tensor cores and 170 RT cores provide a clear advantage. The OpenCL result is similar: 334,370 versus 235,901, a 41.7% margin. These two tests together confirm that the RTX 5090 is the superior compute card, whether the workload is graphics-oriented or general-purpose GPU computing.

Passmark GPU Compute shows a 22.8% win for the RTX 5090 (26,756 vs 21,789), and Passmark DirectX 12 shows a 22.5% win (185 vs 151). Both tests measure the ability to handle modern API workloads, and the RTX 5090's margin is consistent. In Passmark G3D, the RTX 5090 leads by 8.4% (39,650 vs 36,565), a smaller but still meaningful gap in overall 3D rendering. The older DirectX 10 and 11 tests show narrower margins: 8.7% (226 vs 208) and 5.2% (341 vs 324), respectively, and the DirectX 9 result is nearly a tie at 1.5% (395 vs 389).

The only RTX 5080 win is in Passmark G2D, where it scores 1,415 versus 1,413, a 0.1% delta. This is effectively noise, but it does mean the RTX 5080 is not inferior in 2D desktop tasks. The RTX 5080's average benchmark score of 56,083 is far below the RTX 5090's 79,842, and its percentile rank of 87 versus 92 confirms that the RTX 5090 sits higher in the overall GPU hierarchy. The nearest rivals in the database also reflect this gap: the RTX 5090's closest competitor is the Tesla P100 PCIe 16 GB, with a 0.3% delta, while the RTX 5080's nearest rival is the AMD Radeon 8060S, with a 0.6% delta. Neither card is closely challenged by the other, and the RTX 5090's performance is in a different tier entirely.