NVIDIA GeForce RTX 5090 D vs NVIDIA GeForce RTX 5090 SE Comparison

Head-to-Head Benchmarks

The recorded data presents a peculiar situation for the NVIDIA GeForce RTX 5090 D versus the NVIDIA GeForce RTX 5090 SE. The RTX 5090 D carries a full set of benchmark results across multiple test suites, while the RTX 5090 SE has an empty benchmark array in the database. This means no head-to-head averages, no per-test deltas, and no direct score comparisons exist between the two cards. The RTX 5090 D accumulates an average benchmark score of 77,712, but the RTX 5090 SE records an average of 0, reflecting the absence of any submitted results.

What can be analyzed, however, is the RTX 5090 D’s position against other hardware in the database. Its nearest rival by average score is the AMD Radeon RX 6650M XT at 76,904, a delta of 1.1%, meaning the RTX 5090 D sits 1.1% ahead of that mobile GPU. The next nearest rival is the AMD Radeon RX 6850M XT at 78,940, which leads the RTX 5090 D by 1.6%. Further down, the NVIDIA Tesla P100 PCIe 12 GB and 16 GB versions score 79,396 and 79,605, respectively, placing them 2.1% and 2.4% ahead of the RTX 5090 D. These deltas are modest, indicating that the RTX 5090 D’s aggregate performance in the database is closely clustered with those specific accelerators, despite the vast architectural differences.

Looking at individual test results for the RTX 5090 D, the Geekbench Vulkan score of 376,915 is the highest recorded among its benchmarks, followed by Geekbench OpenCL at 310,674. The 3DMark Steel Nomad DX12 test yields 14,326 points. Passmark results show a G3D score of 44,065, a GPU compute score of 28,396, and legacy DirectX tests ranging from 231 (DirectX 10) to 434 (DirectX 9). The G2D score sits at 1,487. Without corresponding RTX 5090 SE scores, these numbers serve only as absolute performance markers, not comparative deltas.

The RTX 5090 SE’s percentile ranking sits at 50, while the RTX 5090 D ranks at 92. This percentile gap suggests the SE’s expected position in the broader GPU landscape is far lower, but the database provides no measured scores to substantiate that rank. The absence of head-to-head benchmarks and wins counts (both zero) makes any direct statistical comparison impossible. The analysis must therefore rely on specification differences to infer the performance relationship.

Where Each One Wins

From the data as recorded, the RTX 5090 D wins in every measurable benchmark category because it is the only card with benchmark results. The RTX 5090 SE has no recorded wins, no scores, and no rival comparisons. The RTX 5090 D’s win count is zero in the head-to-head field, but that field is empty, not a true tally of zero victories. The practical interpretation is that the RTX 5090 D holds all available performance data points.

For use-case segmentation, the RTX 5090 D’s benchmark profile spans several API and workload types. The Geekbench Vulkan and OpenCL scores indicate strong general-purpose compute and graphics API performance. Passmark’s G3D score of 44,065 and GPU compute score of 28,396 suggest solid rasterization and compute throughput. The 3DMark Steel Nomad DX12 score of 14,326 points to DirectX 12 gaming workloads. The legacy DirectX 9 and 11 scores (434 and 371) show backward compatibility, while DirectX 10 (231) and DirectX 12 (219) are lower, which is typical for those specific test conditions.

The RTX 5090 SE, lacking any benchmark data, cannot be assigned wins in any workload category. Its specification sheet, however, implies a different positioning: fewer shading units (14,080 versus 21,760), fewer texture mapping units (440 versus 680), and fewer render output units (160 versus 176). The SE also has fewer ray tracing cores (110 versus 170) and tensor cores (440 versus 680). These reductions point to lower throughput in ray-traced scenes and AI-accelerated workloads, but without scores, the magnitude of the difference remains unquantified.

The memory subsystem differs notably as well. The RTX 5090 D uses 32 GB of GDDR7 on a 512-bit bus, yielding 1.79 TB/s of bandwidth. The RTX 5090 SE uses 24 GB of GDDR7 on a 384-bit bus, yielding 1.34 TB/s. For bandwidth-heavy tasks such as high-resolution texture streaming, large dataset compute, or multi-tasking with multiple displays, the D has an advantage in raw memory throughput. The SE’s smaller frame buffer may still handle most games, but the data does not provide test results to confirm.

The Verdict

The RTX 5090 D is the only card in this comparison with recorded benchmark results. Its average benchmark score of 77,712 places it in the 92nd percentile of all GPUs in the database, a strong positional indicator. The RTX 5090 SE has no recorded results, sits in the 50th percentile by its listed rank, and has an average score of zero. From the data alone, the RTX 5090 D is the demonstrably faster card across all tested workloads.

The nearest rivals to the RTX 5090 D are all within a narrow band: the AMD Radeon RX 6650M XT is 1.1% slower, the AMD Radeon RX 6850M XT is 1.6% faster, and the two NVIDIA Tesla P100 variants are 2.1% and 2.4% faster. This clustering is unusual for a flagship desktop GPU, suggesting that the database’s average score for the RTX 5090 D may be influenced by the mix of tests, some of which favor compute-heavy or legacy workloads. The RTX 5090 SE, with no such data, cannot be positioned relative to these rivals.

For a user selecting strictly on measured performance, the RTX 5090 D is the choice. It delivers verified scores in Vulkan, OpenCL, 3DMark, and Passmark suites. The RTX 5090 SE offers no measured evidence of performance in the database. Its specification sheet shows fewer cores, lower clocks, and reduced memory bandwidth, but those are architectural facts, not benchmark outcomes. The verdict is therefore one-sided: the RTX 5090 D wins by default and by specification, while the RTX 5090 SE remains an unverified entry.

FAQ

Q: Does the RTX 5090 SE have any benchmark scores in the database?

A: No, the RTX 5090 SE has an empty benchmarks array, an average benchmark score of 0, and no nearest rivals listed.

Q: What is the RTX 5090 D’s average benchmark score?

A: The RTX 5090 D records an average benchmark score of 77,712 across its submitted tests.

Q: How does the RTX 5090 D compare to its nearest rival by average score?

A: The AMD Radeon RX 6650M XT scores 76,904, putting the RTX 5090 D 1.1% ahead. The AMD Radeon RX 6850M XT scores 78,940, putting the RTX 5090 D 1.6% behind.

Q: What is the percentile ranking for each card?

A: The RTX 5090 D is in the 92nd percentile of all GPUs, while the RTX 5090 SE is listed in the 50th percentile.

Q: Which card has more memory and bandwidth?

A: The RTX 5090 D has 32 GB of GDDR7 on a 512-bit bus with 1.79 TB/s, while the RTX 5090 SE has 24 GB on a 384-bit bus with 1.34 TB/s.

Q: Are the boost clocks different between the two?

A: Yes, the RTX 5090 D boosts to 2407 MHz, while the RTX 5090 SE boosts to 2377 MHz. The base clocks are 2017 MHz and 1740 MHz, respectively.

Architecture Differences

Both cards share the same fundamental architecture: Blackwell 2.0 on the GB202 chip, fabricated on a 5 nm process at TSMC, with 92,200 million transistors on a 750 mm² die. The transistor density is identical at 122.9M per mm². The core configurations diverge significantly. The RTX 5090 D has 21,760 shading units, 680 texture mapping units, and 176 render output units. The RTX 5090 SE reduces these to 14,080 shading units, 440 texture mapping units, and 160 render output units.

Ray tracing and tensor core counts follow the same pattern. The RTX 5090 D carries 170 ray tracing cores and 680 tensor cores. The RTX 5090 SE has 110 ray tracing cores and 440 tensor cores. This represents a 35% reduction in RT cores and a 35% reduction in tensor cores, based on the recorded figures. The shading unit reduction is more severe, at roughly 35% fewer, which directly impacts pixel and texture throughput.

The memory architecture is another major split. The RTX 5090 D uses a 512-bit memory bus with 32 GB of GDDR7, achieving 1.79 TB/s of bandwidth. The RTX 5090 SE narrows the bus to 384-bit with 24 GB of GDDR7, achieving 1.34 TB/s. Both use the same memory clock of 1750 MHz, translating to 28 Gbps effective. The compute rates reflect these differences: the RTX 5090 D delivers 104.8 TFLOPS of FP32 and FP16 (1:1), while the RTX 5090 SE delivers 66.94 TFLOPS for both. Texture rate drops from 1,636.8 GTexel/s to 1,045.9 GTexel/s, and pixel rate drops from 423.6 GPixel/s to 380.3 GPixel/s.

Power and cooling requirements also differ. The RTX 5090 D has a TDP of 575 W with a suggested PSU of 950 W. The RTX 5090 SE has a TDP of 500 W with a suggested PSU of 900 W. Both use a single 16-pin power connector and are dual-slot cards. The physical dimensions vary: the RTX 5090 D measures 304 mm in length, 137 mm in height, and 48 mm in width, while the RTX 5090 SE is shorter at 267 mm, narrower at 111 mm, and slimmer at 40 mm. The release dates differ as well, with the RTX 5090 D launching on 2025-01-29 and the RTX 5090 SE on 2025-12-31.

Specification Differences

The two cards differ in several measurable fields. The RTX 5090 D has a base clock of 2017 MHz and a boost clock of 2407 MHz, while the RTX 5090 SE has a base clock of 1740 MHz and a boost clock of 2377 MHz. Memory size differs: 32 GB versus 24 GB. Memory bus width differs: 512 bit versus 384 bit. Memory bandwidth differs: 1.79 TB/s versus 1.34 TB/s. The core counts differ across shading units (21,760 versus 14,080), TMUs (680 versus 440), ROPs (176 versus 160), RT cores (170 versus 110), and tensor cores (680 versus 440).

Pixel rate differs: 423.6 GPixel/s versus 380.3 GPixel/s. Texture rate differs: 1,636.8 GTexel/s versus 1,045.9 GTexel/s. FP32 and FP16 compute differ: 104.8 TFLOPS versus 66.94 TFLOPS. TDP differs: 575 W versus 500 W. Suggested PSU differs: 950 W versus 900 W. Dimensions differ: the RTX 5090 D is 304 mm by 137 mm by 48 mm, while the RTX 5090 SE is 267 mm by 111 mm by 40 mm. Release dates differ: 2025-01-29 versus 2025-12-31. The launch MSRP values are 2,299 USD for the RTX 5090 D and 1,499 USD for the RTX 5090 SE.

Fields that are identical include the chip (GB202), architecture (Blackwell 2.0), process node (5 nm), foundry (TSMC), transistor count (92,200 million), die size (750 mm²), memory type (GDDR7), memory clock (1750 MHz, 28 Gbps effective), slot width (dual-slot), power connector (1x 16-pin), bus interface (PCIe 5.0 x16), display outputs (1x HDMI 2.1b, 3x DisplayPort 2.1b), API support (DirectX 12 Ultimate, OpenGL 4.6, Vulkan 1.4), production status (Active), predecessor (GeForce 40), and successor (GeForce 60).