NVIDIA GeForce RTX 4090 D vs NVIDIA GeForce RTX 5070 SUPER Comparison

The Verdict

The recorded benchmark data separates these two NVIDIA GPUs decisively. The GeForce RTX 4090 D wins the only head-to-head test, the 3DMark Steel Nomad DX12 benchmark, by 219.2% over the GeForce RTX 5070 SUPER. The 4090 D posts a score of 8587 versus 2690 for the 5070 SUPER. With one benchmark win for the 4090 D and zero for the 5070 SUPER, the data points to the 4090 D as the stronger performer in this comparison.

The 4090 D sits in the 98th percentile of all GPUs in the database, while the 5070 SUPER sits in the 18th percentile. That 80-point percentile gap reflects a wide chasm in measured performance. However, the 5070 SUPER is an active product, uses newer architecture, and draws less power. The database shows the 4090 D as end-of-life, so availability and support favor the 5070 SUPER for new system builds.

Users who prioritize maximum measured performance should select the RTX 4090 D based on this data. Users who prioritize an active product with newer architecture, lower power draw, and a smaller physical footprint should select the RTX 5070 SUPER, accepting its significantly lower benchmark scores.

Architecture Differences

The two cards represent different architectural generations from NVIDIA. The RTX 4090 D uses the AD102 chip based on Ada Lovelace architecture, part of the GeForce 40-series. The RTX 5070 SUPER uses the GB205 chip based on Blackwell 2.0 architecture, part of the GeForce 50-series. Both are manufactured by TSMC on a 5 nm process node.

The chip sizes differ substantially. The AD102 die measures 609 mm² and contains 76,300 million transistors, resulting in a transistor density of 125.3 million per mm². The GB205 die measures 263 mm² and contains 31,100 million transistors, resulting in a density of 118.3 million per mm². The 4090 D's chip is more than twice the physical size of the 5070 SUPER's chip, and it packs more than double the transistor count.

Core configurations follow the same pattern. The 4090 D has 14,592 shading units, 456 texture mapping units, 176 raster output pipelines, 114 ray tracing cores, and 456 tensor cores. The 5070 SUPER has 6,400 shading units, 200 texture mapping units, 80 raster output pipelines, 50 ray tracing cores, and 200 tensor cores. Every major compute block on the 4090 D exceeds the corresponding block on the 5070 SUPER by a wide margin.

Memory configurations also diverge. The 4090 D uses 24 GB of GDDR6X memory on a 384-bit bus, delivering 1.01 TB/s of bandwidth. The 5070 SUPER uses 18 GB of GDDR7 memory on a 192-bit bus, delivering 672.0 GB/s of bandwidth. While the 5070 SUPER uses a newer memory type, the 4090 D's wider bus and larger capacity give it a clear bandwidth advantage.

Clock speeds are similar at the boost state. The 4090 D boosts to 2520 MHz from a base of 2280 MHz. The 5070 SUPER boosts to 2512 MHz from a base of 2325 MHz. The 5070 SUPER has a slightly higher base clock, but the 4090 D has a slightly higher boost clock.

The 4090 D delivers higher theoretical throughput rates. Its pixel rate is 443.5 GPixel/s versus 201.0 GPixel/s for the 5070 SUPER. Its texture rate is 1,149.1 GTexel/s versus 502.4 GTexel/s. Its FP32 compute is 73.54 TFLOPS versus 32.15 TFLOPS, and its FP16 compute also measures 73.54 TFLOPS (1:1) versus 32.15 TFLOPS (1:1).

Power and physical specifications differ as well. The 4090 D has a TDP of 425 W, uses a triple-slot cooler, and requires a 1x 16-pin power connector with a suggested 800 W power supply. The 5070 SUPER has a TDP of 275 W, uses a dual-slot cooler, and also uses a 1x 16-pin power connector. The 4090 D measures 304 mm long, 137 mm tall, and 61 mm wide. The 5070 SUPER measures 245 mm long, 115 mm tall, and 40 mm wide, making it substantially more compact.

Interface and output differences exist. The 4090 D uses PCIe 4.0 x16, while the 5070 SUPER uses PCIe 5.0 x16. The 4090 D offers 1x HDMI 2.1 and 3x DisplayPort 1.4a outputs. The 5070 SUPER offers 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The 4090 D launched on 2023-12-27 with a launch MSRP of 1,599 USD and has reached end-of-life status. The 5070 SUPER has a release date of 2025-12-31, has no launch MSRP listed, and remains active in production. The 4090 D's predecessor is GeForce 30 and its successor is GeForce 50, placing the 5070 SUPER in the successor generation.

Head-to-Head Benchmarks

The database contains one direct comparison between these two cards: the 3DMark Steel Nomad DX12 test. The RTX 4090 D scored 8587, while the RTX 5070 SUPER scored 2690. The 4090 D wins this test by 219.2%, a margin of 5897 points.

This result aligns with the broader benchmark context. The 4090 D has an average benchmark score of 178050 across all recorded tests, while the 5070 SUPER has an average benchmark score of 2690. The 4090 D also has additional benchmark entries in the database: a Geekbench OpenCL score of 278621 and a Geekbench Vulkan score of 246941. The 5070 SUPER has no additional benchmark entries beyond the Steel Nomad result.

The 4090 D's nearest rivals in the database, based on average scores, are the NVIDIA RTX PRO 5000 Blackwell at 182109, the NVIDIA A100 SXM4 80 GB at 183725, the NVIDIA RTX 5000 Ada Generation at 184664, and the NVIDIA A100 SXM4 40 GB at 187147. The 4090 D trails each of these cards by between 2.2% and 4.9%. These deltas indicate the 4090 D sits in a competitive performance tier, close to professional and datacenter-class accelerators.

The 5070 SUPER's nearest rivals tell a very different story. Its average score of 2690 places it near the NVIDIA Quadro K1100M at 2664, the NVIDIA GeForce GT 1030 at 2662, the Intel Arc Pro B50 at 2660, and the NVIDIA GeForce GT 440 at 2645. The 5070 SUPER leads these cards by between 1.0% and 1.7%. These are modest margins over much older or lower-tier products, indicating the 5070 SUPER's measured performance places it in a much lower performance bracket than the 4090 D.

The performance gap between the two cards is not incremental. The 4090 D's Steel Nomad score is more than triple the 5070 SUPER's score. The percentile rankings confirm this: 98th percentile for the 4090 D versus 18th percentile for the 5070 SUPER. In the database's ranking system, the 4090 D outperforms approximately 80% more of the GPU population than the 5070 SUPER does.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce RTX 4090 D has an average benchmark score of 178050, while the NVIDIA GeForce RTX 5070 SUPER has an average benchmark score of 2690.

Q: How large is the performance gap in the 3DMark Steel Nomad DX12 test?

A: The RTX 4090 D scores 8587 compared to 2690 for the RTX 5070 SUPER, giving the 4090 D a 219.2% advantage in that test.

Q: Which GPU uses more memory and has higher bandwidth?

A: The RTX 4090 D uses 24 GB of GDDR6X memory with 1.01 TB/s bandwidth. The RTX 5070 SUPER uses 18 GB of GDDR7 memory with 672.0 GB/s bandwidth.

Q: What are the power consumption figures for each card?

A: The RTX 4090 D has a TDP of 425 W with a suggested 800 W power supply. The RTX 5070 SUPER has a TDP of 275 W, and no suggested power supply is listed in the database.

Q: Which card is still in active production?

A: The RTX 5070 SUPER has active production status. The RTX 4090 D is marked as end-of-life.

Q: How does each card rank among all GPUs in the database?

A: The RTX 4090 D ranks in the 98th percentile of all GPUs. The RTX 5070 SUPER ranks in the 18th percentile.

Where Each One Wins

The RTX 4090 D wins the only direct benchmark comparison by a wide margin. Its 219.2% lead in 3DMark Steel Nomad DX12 establishes it as the stronger choice for applications that stress raw graphics throughput. The card's higher pixel rate of 443.5 GPixel/s, texture rate of 1,149.1 GTexel/s, and FP32 compute of 73.54 TFLOPS all support this outcome. Users running GPU-intensive workloads that resemble the Steel Nomad test should expect the 4090 D to deliver substantially higher frame rates or compute throughput.

The 4090 D also wins in memory capacity and bandwidth. Its 24 GB of GDDR6X memory on a 384-bit bus provides 1.01 TB/s of bandwidth, which suits large datasets, high-resolution textures, and memory-hungry rendering tasks. The 5070 SUPER's 18 GB of GDDR7 memory on a 192-bit bus provides 672.0 GB/s, which is lower in both capacity and bandwidth.

The 4090 D wins in compute resource counts across the board. Its 14,592 shading units, 456 TMUs, 176 ROPs, 114 RT cores, and 456 tensor cores all exceed the 5070 SUPER's 6,400 shading units, 200 TMUs, 80 ROPs, 50 RT cores, and 200 tensor cores. For workloads that scale with these resources, such as ray tracing, machine learning inference, or parallel compute, the 4090 D's larger configuration provides more headroom.

The 5070 SUPER wins in areas related to efficiency and physical design. Its TDP of 275 W is 150 W lower than the 4090 D's 425 W, meaning it draws less power under load. Its dual-slot cooler is narrower than the 4090 D's triple-slot cooler. Its dimensions of 245 mm by 115 mm by 40 mm are smaller than the 4090 D's 304 mm by 137 mm by 61 mm, making it easier to fit in compact cases. Its PCIe 5.0 x16 interface is a newer generation than the 4090 D's PCIe 4.0 x16. Its DisplayPort 2.1b outputs are newer than the 4090 D's DisplayPort 1.4a outputs. Its GDDR7 memory type is newer than the 4090 D's GDDR6X.

The 5070 SUPER also wins on product lifecycle status. It is active, while the 4090 D is end-of-life. The 5070 SUPER has a newer release date and belongs to the current GeForce 50-series generation. The 4090 D's successor is GeForce 50, which directly includes the 5070 SUPER.

The database shows no benchmark tests where the 5070 SUPER beats the 4090 D. The wins tally stands at 1 for the 4090 D and 0 for the 5070 SUPER. In scenarios where measured performance is the sole criterion, the 4090 D is the clear winner. In scenarios where power efficiency, physical size, modern interfaces, or active production status matter more than raw performance, the 5070 SUPER has advantages that the benchmark scores do not capture.