NVIDIA GeForce RTX 4070 AD103 vs NVIDIA GeForce RTX 5090 Comparison

The Verdict

The recorded data positions the NVIDIA GeForce RTX 5090 as the definitive performance leader, with an average benchmark score of 79,842 and a 92nd percentile ranking across all GPUs. The RTX 4070 AD103, by contrast, holds a 50th percentile ranking with no recorded benchmark scores in the database, indicating a mid-range positioning. The RTX 5090’s measured results show it outperforming its nearest rivals by margins ranging from 0.3% to 1.4%, with the AMD Radeon Pro Vega 64X trailing by 1.4% and the Tesla P100 PCIe 16 GB nearly matching its score at a 0.3% delta. For users targeting maximum compute throughput, the RTX 5090 is the only choice supported by the data, as it delivers more than triple the FP32 throughput of the RTX 4070 AD103: 104.8 TFLOPS versus 29.15 TFLOPS. The RTX 4070 AD103 remains a capable mid-range option for those who prioritize lower power draw and physical footprint, but the benchmark evidence overwhelmingly favors the RTX 5090 in raw performance metrics.

FAQ

Q: What is the primary architectural difference between the two GPUs?

A: The RTX 4070 AD103 uses the Ada Lovelace architecture on a 5 nm TSMC process, while the RTX 5090 uses Blackwell 2.0, also on 5 nm TSMC. The RTX 5090 leverages a much larger chip: 92,200 million transistors on a 750 mm² die, versus 45,900 million transistors on 379 mm² for the RTX 4070 AD103.

Q: How do memory capacities and bandwidth compare?

A: The RTX 4070 AD103 has 12 GB of GDDR6X on a 192 bit bus, delivering 504.2 GB/s bandwidth. The RTX 5090 has 32 GB of GDDR7 on a 512 bit bus, delivering 1.79 TB/s bandwidth, a more than threefold increase in both capacity and bandwidth.

Q: Which GPU has higher clock speeds?

A: The RTX 5090 has a base clock of 2017 MHz and a boost clock of 2407 MHz. The RTX 4070 AD103 has a base clock of 1920 MHz and a boost clock of 2475 MHz, meaning the RTX 4070 AD103 actually boosts slightly higher, but the RTX 5090 maintains a higher base clock.

Q: What are the power requirements for each card?

A: The RTX 4070 AD103 carries a 200 W TDP with a suggested 550 W power supply, while the RTX 5090 carries a 575 W TDP with a suggested 950 W power supply. Both use a single 16-pin power connector.

Q: Which GPU has better API support?

A: Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 identically. The RTX 5090 adds newer display outputs with three DisplayPort 2.1b ports and one HDMI 2.1b, while the RTX 4070 AD103 has three DisplayPort 1.4a and one HDMI 2.1.

Q: What is the production status of each card?

A: The RTX 4070 AD103 is listed as end-of-life, with a release date of February 29, 2024, while the RTX 5090 is active, released on January 29, 2025.

Architecture Differences

The two GPUs represent successive generations of NVIDIA’s architecture. The RTX 4070 AD103 uses Ada Lovelace, the architecture behind the GeForce 40-series, while the RTX 5090 uses Blackwell 2.0, the foundation for the GeForce 50-series. Both are fabricated by TSMC on a 5 nm process, but the transistor counts diverge dramatically. The RTX 4070 AD103 packs 45,900 million transistors into a 379 mm² die, achieving a density of 121.1 million transistors per square millimeter. The RTX 5090’s GB202 chip contains 92,200 million transistors across a 750 mm² die, a density of 122.9 million transistors per square millimeter. The near-identical density suggests the architectural gains in the RTX 5090 come primarily from doubling the die size rather than process improvements.

The RTX 5090 also doubles the memory interface width to 512 bit and moves to GDDR7, a generational leap from the GDDR6X used by the RTX 4070 AD103. The newer card’s memory clock runs at 1750 MHz with 28 Gbps effective speed, compared to 1313 MHz with 21 Gbps effective on the older card. Feature-wise, the RTX 5090 supports PCIe 5.0 x16, while the RTX 4070 AD103 uses PCIe 4.0 x16. Display outputs also differ, with the RTX 5090 offering DisplayPort 2.1b and the RTX 4070 AD103 limited to DisplayPort 1.4a. Both support the same DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 API sets, so software compatibility remains equal.

The RTX 5090’s tensor core count of 680 versus 184 on the RTX 4070 AD103, and RT core count of 170 versus 46, indicates a substantial increase in AI and ray tracing hardware. The shading unit count jumps from 5,888 to 21,760, and texture mapping units scale from 184 to 680. These architectural changes, combined with the larger memory subsystem, define the performance gap observed in the benchmark data.

Specification Differences

The specification sheets differ across nearly every measurable field. The RTX 4070 AD103 has 5,888 shading units, 184 TMUs, and 64 ROPs, while the RTX 5090 has 21,760 shading units, 680 TMUs, and 176 ROPs. RT cores increase from 46 to 170, and tensor cores from 184 to 680. The pixel rate climbs from 158.4 GPixel/s to 423.6 GPixel/s, and the texture rate from 455.4 GTexel/s to 1,636.8 GTexel/s. FP32 and FP16 throughput both scale from 29.15 TFLOPS to 104.8 TFLOPS.

Memory specifications show the RTX 4070 AD103 with 12 GB GDDR6X, a 192 bit bus, and 504.2 GB/s bandwidth. The RTX 5090 uses 32 GB GDDR7, a 512 bit bus, and 1.79 TB/s bandwidth. Clock speeds differ modestly: the RTX 4070 AD103 runs at 1920 MHz base and 2475 MHz boost, while the RTX 5090 runs at 2017 MHz base and 2407 MHz boost. Power consumption rises from 200 W to 575 W TDP, with suggested PSU ratings of 550 W and 950 W respectively. Physical dimensions grow from 240 mm by 110 mm by 40 mm on the RTX 4070 AD103 to 304 mm by 137 mm by 40 mm on the RTX 5090, though both remain dual-slot cards with a single 16-pin connector. The bus interface advances from PCIe 4.0 x16 to PCIe 5.0 x16. Release dates sit roughly 11 months apart, with the RTX 4070 AD103 launching on February 29, 2024, and the RTX 5090 on January 29, 2025.

Head-to-Head Benchmarks

The database records no direct head-to-head benchmark results between the RTX 4070 AD103 and the RTX 5090, and the RTX 4070 AD103 has no individual benchmark scores listed. The RTX 5090, however, has a full suite of recorded measurements. In 3DMark Steel Nomad DX12, it scores 18,355. Geekbench results show 334,370 in OpenCL and 376,728 in Vulkan. PassMark scores span multiple DirectX versions: 226 in DirectX 10, 341 in DirectX 11, 185 in DirectX 12, and 395 in DirectX 9. The PassMark G2D score is 1,413, while the G3D score reaches 39,650, and the GPU compute score hits 26,756.

The RTX 5090’s average benchmark score of 79,842 places it at the 92nd percentile of all GPUs. Its nearest rivals in the database include the NVIDIA Tesla P100 PCIe 16 GB with an average score of 79,605, a 0.3% delta, and the Tesla P100 PCIe 12 GB at 79,396, a 0.6% delta. The AMD Radeon RX 6850M XT scores 78,940, 1.1% behind, while the AMD Radeon Pro Vega 64X scores 80,959, which is 1.4% ahead. These margins are narrow, indicating the RTX 5090 trades closely with high-end workstation and laptop GPUs rather than dominating them outright.

Without recorded benchmarks for the RTX 4070 AD103, direct comparisons rely on theoretical specifications. The RTX 5090’s FP32 throughput of 104.8 TFLOPS is 3.6 times the RTX 4070 AD103’s 29.15 TFLOPS. Texture rate scales by a similar factor, 1,636.8 GTexel/s versus 455.4 GTexel/s. Memory bandwidth jumps from 504.2 GB/s to 1.79 TB/s, a 3.55 times increase. The RTX 5090’s 176 ROPs versus 64 ROPs yields a pixel rate of 423.6 GPixel/s, 2.67 times the RTX 4070 AD103’s 158.4 GPixel/s. These specification ratios align with the RTX 5090’s top-tier percentile ranking, while the RTX 4070 AD103’s mid-range 50th percentile suggests it delivers roughly half the market’s performance envelope. The data confirms the RTX 5090 as the clear performance winner, with the RTX 4070 AD103 offering a lower-power alternative that sacrifices most compute metrics.