NVIDIA GeForce RTX 5090 SE vs NVIDIA N1X 40SM Comparison

Head-to-Head Benchmarks

The recorded database contains no direct head-to-head benchmark results for the NVIDIA GeForce RTX 5090 SE versus the NVIDIA N1X 40SM. Both entries show an average benchmark score of zero, and the wins counter for each side is zero. This does not mean the two are equivalent; rather, the data indicates that no standardized comparative testing has been logged between these specific SKUs.

What the data does provide is raw compute throughput figures that allow for a theoretical comparison. The RTX 5090 SE delivers 66.94 TFLOPS of FP32 performance, while the N1X 40SM delivers 24.02 TFLOPS. That places the RTX 5090 SE at roughly 2.79 times the FP32 throughput of the N1X 40SM. In FP16, the ratio is identical at 66.94 TFLOPS versus 24.02 TFLOPS, since both parts maintain a 1:1 FP16 to FP32 ratio.

The texture rate differential is narrower but still substantial. The RTX 5090 SE achieves 1,045.9 GTexel/s against the N1X 40SM's 750.7 GTexel/s, a margin of about 39%. The pixel rate gap is much larger, with the RTX 5090 SE producing 380.3 GPixel/s versus 93.84 GPixel/s for the N1X 40SM, a difference of roughly 4.05 times.

Memory bandwidth tells a similar story. The RTX 5090 SE has 1.34 TB/s of bandwidth, while the N1X 40SM has 273.2 GB/s. That is a 4.9 times advantage for the RTX 5090 SE. However, the N1X 40SM counters with a far larger memory pool: 128 GB of LPDDR5X versus 24 GB of GDDR7. For workloads that require massive resident datasets, the N1X 40SM holds a capacity advantage that raw bandwidth cannot overcome.

Both parts sit at the 50th percentile in the database's ranking of all GPUs, and both have zero logged benchmark scores. The absence of measured results means percentile rankings are provisional placeholders rather than evidence of parity.

Architecture Differences

Both GPUs are built on the Blackwell 2.0 architecture and use TSMC's 5 nm process node, but they diverge sharply in physical design and purpose. The RTX 5090 SE uses the GB202 chip with a die size of 750 mm² and a transistor count of 92,200 million, yielding a transistor density of 122.9 million per mm². The N1X 40SM uses the GB20B chip with a die size of 382 mm²; its transistor count is listed as unknown in the database.

The shading unit counts reflect the different design philosophies. The RTX 5090 SE carries 14,080 shading units, 440 TMUs, and 160 ROPs. The N1X 40SM has 5,120 shading units, 320 TMUs, and only 40 ROPs. The RTX 5090 SE also leads in ray tracing cores (110 versus 40) and tensor cores (440 versus 160).

Clock behavior differs significantly. The RTX 5090 SE has a base clock of 1740 MHz and a boost clock of 2377 MHz. The N1X 40SM starts at a much lower 741 MHz base but boosts to 2346 MHz, nearly matching the RTX 5090 SE's boost frequency. This suggests the N1X 40SM is designed to idle at very low power and ramp up aggressively when needed, consistent with its IGP (integrated graphics processor) classification.

Memory architecture is fundamentally different. The RTX 5090 SE uses 24 GB of GDDR7 on a 384-bit bus. The N1X 40SM uses 128 GB of LPDDR5X on a 256-bit bus. The memory clock also differs: 1750 MHz (28 Gbps effective) for the RTX 5090 SE versus 1067 MHz (8.5 Gbps effective) for the N1X 40SM.

The API support reveals a critical distinction. The RTX 5090 SE supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The N1X 40SM lists DirectX as N/A, OpenGL as N/A, and Vulkan as N/A. This indicates the N1X 40SM is not intended for conventional graphics API workloads, reinforcing its role as a specialized compute or IGP part rather than a general-purpose gaming GPU.

Form factor and power delivery also separate the two. The RTX 5090 SE is a dual-slot card with a 1x 16-pin power connector and a 500 W TDP, requiring a 900 W suggested PSU. The N1X 40SM is an IGP with no power connectors and an unknown TDP, meaning it draws power directly from the system board.

The production status for both is listed as Active. The RTX 5090 SE has a release date of 2025-12-31 and a successor in the GeForce 60 series, while the N1X 40SM has a release date of 2026-05-31 and no predecessor or successor listed.

The Verdict

The data paints a clear picture of two devices built for different tasks. The RTX 5090 SE is a discrete, high-throughput graphics card with full API support, designed for traditional rendering workloads such as gaming, 3D content creation, and GPU-accelerated compute that relies on DirectX, Vulkan, or OpenGL. Its 66.94 TFLOPS FP32, 1,045.9 GTexel/s texture rate, and 1.34 TB/s bandwidth place it firmly in the high-end discrete GPU tier.

The N1X 40SM, despite its Blackwell 2.0 architecture and 5 nm process, is fundamentally different. With no DirectX, OpenGL, or Vulkan support, it cannot run conventional graphics applications. Its 24.02 TFLOPS FP32 and 273.2 GB/s bandwidth are far below the RTX 5090 SE, but its 128 GB memory capacity and IGP form factor suggest a role in large-memory compute or server-side inference tasks where capacity matters more than throughput.

For anyone selecting between these two, the decision is not about performance per dollar or benchmark scores. It is about whether the workload requires a full graphics API stack and high-bandwidth rendering. If it does, the RTX 5090 SE is the only viable option. If the workload is memory-capacity-bound and does not use conventional graphics APIs, the N1X 40SM offers a vastly larger memory pool in an integrated form factor. The data does not support a single winner; it supports two distinct winners for two distinct use cases.

Specification Differences

| Field | RTX 5090 SE | N1X 40SM |

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

| Chip | GB202 | GB20B |

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

| Die Size | 750 mm² | 382 mm² |

| Transistors | 92,200 million | unknown |

| Base Clock | 1740 MHz | 741 MHz |

| Boost Clock | 2377 MHz | 2346 MHz |

| Memory Clock | 1750 MHz (28 Gbps effective) | 1067 MHz (8.5 Gbps effective) |

| Memory Size | 24 GB | 128 GB |

| Memory Type | GDDR7 | LPDDR5X |

| Memory Bus | 384 bit | 256 bit |

| Memory Bandwidth | 1.34 TB/s | 273.2 GB/s |

| Shading Units | 14,080 | 5,120 |

| TMUs | 440 | 320 |

| ROPs | 160 | 40 |

| RT Cores | 110 | 40 |

| Tensor Cores | 440 | 160 |

| Pixel Rate | 380.3 GPixel/s | 93.84 GPixel/s |

| Texture Rate | 1,045.9 GTexel/s | 750.7 GTexel/s |

| FP32 | 66.94 TFLOPS | 24.02 TFLOPS |

| FP16 | 66.94 TFLOPS | 24.02 TFLOPS |

| TDP | 500 W | unknown |

| Slot Width | Dual-slot | IGP |

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

| Suggested PSU | 900 W | null |

| 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 |

| Dimensions | 267 mm x 111 mm x 40 mm | null |

| Release Date | 2025-12-31 | 2026-05-31 |

| Launch MSRP | 1,499 USD | null |

FAQ

Q: Which GPU has higher raw FP32 compute performance?

A: The NVIDIA GeForce RTX 5090 SE delivers 66.94 TFLOPS, which is about 2.79 times the 24.02 TFLOPS of the NVIDIA N1X 40SM.

Q: How do the memory capacities compare?

A: The N1X 40SM has 128 GB of LPDDR5X memory, more than five times the 24 GB of GDDR7 found on the RTX 5090 SE.

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

A: No. The database lists DirectX, OpenGL, and Vulkan support as N/A for the N1X 40SM. The RTX 5090 SE supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: What is the memory bandwidth of each GPU?

A: The RTX 5090 SE has 1.34 TB/s of bandwidth, while the N1X 40SM has 273.2 GB/s. The RTX 5090 SE offers roughly 4.9 times the bandwidth.

Q: Are both GPUs based on the same architecture?

A: Yes, both use Blackwell 2.0 architecture and are manufactured on TSMC's 5 nm process node, but they use different chips: GB202 for the RTX 5090 SE and GB20B for the N1X 40SM.

Q: What is the die size difference?

A: The RTX 5090 SE has a 750 mm² die, while the N1X 40SM has a 382 mm² die.

Where Each One Wins

The RTX 5090 SE wins in every throughput metric recorded. Its FP32 and FP16 performance are both 66.94 TFLOPS, which is 2.79 times the N1X 40SM's 24.02 TFLOPS. The pixel rate advantage is even more pronounced at 380.3 GPixel/s versus 93.84 GPixel/s, a 4.05 times margin. Texture rate favors the RTX 5090 SE at 1,045.9 GTexel/s versus 750.7 GTexel/s, a 39% lead. Memory bandwidth is decisively in favor of the RTX 5090 SE at 1.34 TB/s versus 273.2 GB/s, a 4.9 times difference. The RTX 5090 SE also has more shading units (14,080 versus 5,120), more ROPs (160 versus 40), more RT cores (110 versus 40), and more tensor cores (440 versus 160). For any workload that depends on raw compute throughput, fill rates, or memory bandwidth, the RTX 5090 SE is the clear choice.

The N1X 40SM wins in memory capacity, offering 128 GB versus 24 GB, a 5.3 times advantage. This matters for workloads with very large working sets that must reside in GPU memory. The N1X 40SM also has a lower base clock (741 MHz versus 1740 MHz), which suggests lower idle power draw, and its IGP form factor with no power connectors means it requires no external power cabling. The N1X 40SM also has a smaller die (382 mm² versus 750 mm²), which could imply lower manufacturing cost per wafer, though the database does not provide pricing data for the N1X 40SM.

The release dates show the N1X 40SM is a later product, arriving 2026-05-31 versus the RTX 5090 SE's 2025-12-31. The RTX 5090 SE has a listed launch MSRP of 1,499 USD, while the N1X 40SM has no launch MSRP in the database. The RTX 5090 SE has a successor (GeForce 60) while the N1X 40SM does not.

For conventional graphics rendering with full API support, the RTX 5090 SE wins outright. For large-memory compute workloads that do not rely on DirectX, OpenGL, or Vulkan, the N1X 40SM provides a capacity advantage that the RTX 5090 SE cannot match. The data supports a functional split: rendering and throughput-bound tasks favor the RTX 5090 SE; capacity-bound tasks with no graphics API requirements favor the N1X 40SM.