NVIDIA GeForce RTX 5090 D vs NVIDIA N1X 40SM Comparison

Where Each One Wins

The data reveals a stark contrast in the intended roles of these two NVIDIA parts. The GeForce RTX 5090 D is a dedicated discrete graphics card with a full suite of benchmark results across DirectX, OpenCL, and Vulkan workloads. The N1X 40SM, classified as an IGP (integrated graphics processor) in the Blackwell N1x generation, has no recorded benchmark scores in the database, an average score of zero, and sits at the 50th percentile of all GPUs by default. This makes a direct use-case split straightforward: the RTX 5090 D wins every measurable performance contest, while the N1X 40SM is positioned for systems where a separate graphics card is not present or not needed.

The RTX 5090 D delivers a 92nd percentile standing among all GPUs, with an average benchmark score of 77,712 across its ten recorded tests. The N1X 40SM has no tests to compare, so its 50th percentile is a placeholder rather than a measured result. In practical terms, the 5090 D is designed for high-end desktop workloads, while the N1X 40SM is an integrated solution with a much smaller footprint: it uses an IGP slot width, draws no external power connectors, and has a die size of 382 mm² compared to the 5090 D's 750 mm². The N1X 40SM also carries a substantially larger memory pool of 128 GB, which points toward a memory-capacity-centric use case, likely for AI or data processing where bandwidth and capacity matter more than raw pixel pushing.

Architecture Differences

Both parts share the same Blackwell 2.0 architecture and are fabricated on a 5 nm process at TSMC, but the similarities end there. The RTX 5090 D uses the GB202 chip, while the N1X 40SM uses the GB20B chip. The 5090 D packs 92,200 million transistors into a 750 mm² die, yielding a transistor density of 122.9 million per square millimeter. The N1X 40SM's transistor count is listed as unknown, but its die is nearly half the size at 382 mm², and no density figure is recorded.

Core counts diverge dramatically. The 5090 D has 21,760 shading units, 680 texture mapping units, 176 ROPs, 170 RT cores, and 680 tensor cores. The N1X 40SM has 5,120 shading units, 320 TMUs, 40 ROPs, 40 RT cores, and 160 tensor cores. That is roughly a 4.25x difference in shading units, 2.125x in TMUs, 4.4x in ROPs, 4.25x in RT cores, and 4.25x in tensor cores. Clock speeds tell a more nuanced story: the 5090 D has a base clock of 2017 MHz and a boost of 2407 MHz, while the N1X 40SM runs a much lower base of 741 MHz but boosts to 2346 MHz. The integrated part's boost clock nearly matches the discrete card, but its sustained base clock is far lower, suggesting power and thermal constraints for the IGP form factor.

Memory architecture is another differentiator. The 5090 D uses 32 GB of GDDR7 on a 512-bit bus, delivering 1.79 TB/s of bandwidth. The N1X 40SM uses 128 GB of LPDDR5X on a 256-bit bus, with 273.2 GB/s of bandwidth. The bandwidth gap is enormous, roughly 6.5x in favor of the 5090 D, but the capacity advantage goes to the N1X 40SM by 4x. The 5090 D's memory clock is 1750 MHz (28 Gbps effective), while the N1X 40SM runs at 1067 MHz (8.5 Gbps effective). The N1X 40SM also lists no DirectX, OpenGL, or Vulkan API support, whereas the 5090 D supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Head-to-Head Benchmarks

The database records no head-to-head benchmark entries between these two parts, and the N1X 40SM has an empty benchmark array. All comparisons must therefore rely on the 5090 D's measured scores and the architectural figures for the N1X 40SM.

The RTX 5090 D's strongest recorded result is in Geekbench Vulkan, where it scores 376,915. Its Geekbench OpenCL score is 310,674. In 3DMark Steel Nomad DX12, it posts 14,326. Passmark results show 44,065 in G3D, 28,396 in GPU compute, 1,487 in G2D, 434 in DirectX 9, 371 in DirectX 11, 231 in DirectX 10, and 219 in DirectX 12. The N1X 40SM has no comparable numbers, so the performance delta is incalculable from recorded data. However, the raw throughput figures are telling: the 5090 D delivers 104.8 TFLOPS of FP32 and FP16 (1:1), while the N1X 40SM delivers 24.02 TFLOPS for both precision formats. That is a 4.36x difference in compute throughput. Pixel rate on the 5090 D is 423.6 GPixel/s versus 93.84 GPixel/s on the N1X 40SM, a 4.5x gap. Texture rate is 1,636.8 GTexel/s versus 750.7 GTexel/s, a 2.18x difference.

The nearest rivals for the 5090 D in the database are all AMD or older NVIDIA parts: the AMD Radeon RX 6650M XT sits 1.1% below in average score, the RX 6850M XT is 1.6% above, and the NVIDIA Tesla P100 PCIe 12 GB and 16 GB variants are 2.1% and 2.4% above respectively. This places the 5090 D in a tight cluster of high-end parts, with its 77,712 average score nearly matching those comparison points. The N1X 40SM has no nearest rivals recorded, so its competitive positioning cannot be assessed beyond its architectural limits.

FAQ

Q: Which part has a higher boost clock?

A: The RTX 5090 D boosts to 2407 MHz, while the N1X 40SM boosts to 2346 MHz. The discrete card is 61 MHz higher.

Q: How much memory bandwidth does each part offer?

A: The RTX 5090 D provides 1.79 TB/s over a 512-bit GDDR7 interface. The N1X 40SM offers 273.2 GB/s over a 256-bit LPDDR5X interface.

Q: Does the N1X 40SM support DirectX or Vulkan?

A: No. The database lists DirectX as N/A, OpenGL as N/A, and Vulkan as N/A for the N1X 40SM. The RTX 5090 D supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

Q: What is the transistor count difference?

A: The RTX 5090 D has 92,200 million transistors on a 750 mm² die. The N1X 40SM's transistor count is listed as unknown, and its die is 382 mm².

Q: Which part has a higher pixel fill rate?

A: The RTX 5090 D achieves 423.6 GPixel/s, while the N1X 40SM achieves 93.84 GPixel/s.

Q: Are there any recorded benchmark scores for the N1X 40SM?

A: No. Its benchmark array is empty, its average score is zero, and it holds no wins in head-to-head comparisons.

Specification Differences

| Specification | NVIDIA GeForce RTX 5090 D | NVIDIA N1X 40SM |

|---|---|---|

| Chip | GB202 | GB20B |

| Generation | GeForce 50 | Blackwell IGP (N1x) |

| Die Size | 750 mm² | 382 mm² |

| Base Clock | 2017 MHz | 741 MHz |

| Boost Clock | 2407 MHz | 2346 MHz |

| Memory Size | 32 GB | 128 GB |

| Memory Type | GDDR7 | LPDDR5X |

| Memory Bus Width | 512 bit | 256 bit |

| Memory Bandwidth | 1.79 TB/s | 273.2 GB/s |

| Shading Units | 21,760 | 5,120 |

| TMUs | 680 | 320 |

| ROPs | 176 | 40 |

| RT Cores | 170 | 40 |

| Tensor Cores | 680 | 160 |

| Pixel Rate | 423.6 GPixel/s | 93.84 GPixel/s |

| Texture Rate | 1,636.8 GTexel/s | 750.7 GTexel/s |

| FP32 / FP16 | 104.8 TFLOPS | 24.02 TFLOPS |

| TDP | 575 W | Unknown |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 1x 16-pin | None |

| Suggested PSU | 950 W | None listed |

| Display Outputs | 1x HDMI 2.1b, 3x DisplayPort 2.1b | 1x HDMI |

| DirectX | 12 Ultimate (12_2) | N/A |

| OpenGL | 4.6 | N/A |

| Vulkan | 1.4 | N/A |

| Release Date | 2025-01-29 | 2026-05-31 |

| Launch MSRP | 2,299 USD | None listed |

The Verdict

The recorded data leaves no ambiguity about which part delivers superior graphics performance. The RTX 5090 D is a fully benchmarked, active production card with a 92nd percentile standing, ten measured scores across multiple APIs, and a launch MSRP of 2,299 USD. Its nearest rivals in the database are within roughly 2.4% of its average score, confirming it sits in a competitive tier of high-end hardware. The N1X 40SM, by contrast, has no benchmarks, no wins, no rivals, and no API support listed. It is an integrated part with a much smaller die, fewer cores, and lower throughput across every measurable compute metric.

The N1X 40SM does hold one clear advantage: memory capacity. Its 128 GB of LPDDR5X is four times the 5090 D's 32 GB, and its 382 mm² die is roughly half the size. For a workload that needs large resident datasets and does not require DirectX or Vulkan rendering, the integrated part could be relevant. But for any graphics, gaming, or general compute task that relies on the APIs the 5090 D supports, the discrete card is the only viable option from this data set. The 5090 D also carries a 575 W TDP and requires a 950 W suggested PSU, while the N1X 40SM's power draw is unknown and it uses no external connectors. The choice comes down to whether the system needs a removable, high-throughput graphics solution or an integrated memory-heavy processor with no measured rendering capability.