NVIDIA GeForce RTX 5090 SE vs NVIDIA N1 20SM Comparison

Head-to-Head Benchmarks

The recorded database contains no benchmark scores for either the NVIDIA GeForce RTX 5090 SE or the NVIDIA N1 20SM. Both entries list an average benchmark score of zero and a percentile rank of 50 against all GPUs. With zero wins recorded for each part, the head-to-head comparison rests entirely on architectural specifications and theoretical performance ceilings derived from clock rates and unit counts.

The RTX 5090 SE delivers 66.94 TFLOPS of FP32 compute, while the N1 20SM delivers 12.01 TFLOPS. That difference places the RTX 5090 SE at roughly 5.57 times the raw shader throughput of the N1 20SM. Pixel throughput shows a similar gap: the RTX 5090 SE reaches 380.3 GPixel/s against 56.30 GPixel/s for the N1 20SM, a margin of approximately 6.75 times. Texture fill rates differ by a factor of about 2.79, with the RTX 5090 SE producing 1,045.9 GTexel/s versus 375.4 GTexel/s.

Memory bandwidth strongly favors the RTX 5090 SE. Its 1.34 TB/s over a 384-bit GDDR7 interface compares to 273.2 GB/s over a 256-bit LPDDR5X bus on the N1 20SM. The RTX 5090 SE carries 24 GB of GDDR7, while the N1 20SM carries 128 GB of LPDDR5X. Capacity favors the N1 20SM by a wide margin, but bandwidth favors the RTX 5090 SE by nearly 4.9 times.

Clock behavior also differs substantially. The RTX 5090 SE bases at 1740 MHz and boosts to 2377 MHz. The N1 20SM bases at 741 MHz but boosts to 2346 MHz, a much larger relative boost window. Effective memory speed on the RTX 5090 SE is 28 Gbps, compared to 8.5 Gbps on the N1 20SM.

Architecture Differences

Both parts use the Blackwell 2.0 architecture and are fabricated on a 5 nm process at TSMC. The RTX 5090 SE uses the GB202 chip, while the N1 20SM uses the GB20B chip. Die size differs significantly: the RTX 5090 SE measures 750 mm² with 92,200 million transistors, yielding a transistor density of 122.9M per mm². The N1 20SM measures 382 mm², with transistor count listed as unknown and density not recorded.

Core configurations diverge sharply. The RTX 5090 SE contains 14,080 shading units, 440 texture mapping units, 160 ROPs, 110 RT cores, and 440 tensor cores. The N1 20SM contains 2,560 shading units, 160 TMUs, 24 ROPs, 20 RT cores, and 80 tensor cores. The RTX 5090 SE thus has 5.5 times the shading units, 2.75 times the TMUs, and 6.67 times the ROPs.

The N1 20SM is classified as an integrated graphics processor (IGP) with no slot width, no power connectors, and unknown TDP. The RTX 5090 SE is a dual-slot discrete card with a 500 W TDP, a single 16-pin power connector, and a suggested 900 W PSU. The N1 20SM draws no external power and has no PSU recommendation.

API support separates the two as well. The RTX 5090 SE supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The N1 20SM lists N/A for all three APIs. Display outputs also differ: the RTX 5090 SE provides 1x HDMI 2.1b and 3x DisplayPort 2.1b, while the N1 20SM provides only 1x HDMI.

The bus interface is identical at PCIe 5.0 x16. Both parts are currently active in production. The RTX 5090 SE belongs to the GeForce 50 series with a generation tag of GeForce 50, while the N1 20SM belongs to the Blackwell IGP (N1x) generation. The RTX 5090 SE has a predecessor (GeForce 40) and successor (GeForce 60); the N1 20SM lists neither.

Where Each One Wins

The RTX 5090 SE wins decisively in every compute and rendering metric recorded. Its FP32 throughput of 66.94 TFLOPS dwarfs the 12.01 TFLOPS of the N1 20SM, which indicates a 5.5 times advantage in general shader work. Pixel fill rate favors the RTX 5090 SE at 380.3 GPixel/s versus 56.30 GPixel/s, making it the stronger choice for high-resolution rasterization and heavy overdraw scenarios. Texture rate of 1,045.9 GTexel/s against 375.4 GTexel/s gives the RTX 5090 SE a clear edge in texture-bound workloads.

RT and tensor core counts reinforce this pattern. With 110 RT cores versus 20, and 440 tensor cores versus 80, the RTX 5090 SE provides substantially more dedicated hardware for ray tracing and AI inference. The memory subsystem of the RTX 5090 SE, with 1.34 TB/s of bandwidth, supports these compute advantages by feeding data to the shader array much faster than the N1 20SM's 273.2 GB/s.

The N1 20SM wins on memory capacity with 128 GB versus 24 GB. This capacity advantage can matter for workloads that need to hold very large datasets locally, such as certain inference or memory-resident processing tasks, even though the available bandwidth is far lower. The N1 20SM also wins on power efficiency by design: it requires no power connectors and has unknown TDP, whereas the RTX 5090 SE demands 500 W and a 900 W PSU. Physical footprint favors the N1 20SM as an IGP with no dimensions recorded, while the RTX 5090 SE occupies 267 mm in length, 111 mm in height, and 40 mm in width.

Boost clocks are close between the two (2377 MHz for the RTX 5090 SE, 2346 MHz for the N1 20SM), but the RTX 5090 SE maintains a much higher base clock of 1740 MHz versus 741 MHz. This indicates the N1 20SM relies on a wide boost range to reach near-parity in peak frequency, while the RTX 5090 SE sustains higher clocks under load.

The Verdict

The data points to a clear performance hierarchy. The RTX 5090 SE is the higher-performance part by every computational metric in the database: FP32, pixel rate, texture rate, RT core count, tensor core count, and memory bandwidth. Its 66.94 TFLOPS FP32 output and 1.34 TB/s bandwidth position it as the appropriate choice for graphics-intensive workloads, ray tracing, and compute tasks that demand sustained throughput. The 500 W TDP and dual-slot cooler reflect this capability.

The N1 20SM serves a fundamentally different role. As an IGP with no external power, no slot width, and no API support listed, it appears designed for integrated, low-power environments rather than discrete rendering. Its 128 GB memory capacity is notable, and its 12.01 TFLOPS FP32 output, while far below the RTX 5090 SE, remains meaningful for an integrated part with a 2346 MHz boost clock. The 273.2 GB/s bandwidth limits its ability to exploit that capacity in bandwidth-sensitive tasks.

For a user prioritizing rendering performance, ray tracing, or high-FPS gaming, the RTX 5090 SE is the only choice between these two. For a user needing massive memory capacity in a power-constrained, space-constrained, or integrated form factor, the N1 20SM offers a capability the RTX 5090 SE cannot match. The two parts do not compete in the same market segment; their specification sheets place them at opposite ends of the GPU spectrum.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The NVIDIA GeForce RTX 5090 SE delivers 66.94 TFLOPS of FP32, while the NVIDIA N1 20SM delivers 12.01 TFLOPS.

Q: How much memory does each GPU have?

A: The RTX 5090 SE has 24 GB of GDDR7 on a 384-bit bus. The N1 20SM has 128 GB of LPDDR5X on a 256-bit bus.

Q: What are the memory bandwidth figures?

A: The RTX 5090 SE reaches 1.34 TB/s, while the N1 20SM reaches 273.2 GB/s.

Q: Do both GPUs use the same architecture and process node?

A: Yes, both use Blackwell 2.0 architecture and a 5 nm TSMC process. The RTX 5090 SE uses the GB202 chip, and the N1 20SM uses the GB20B chip.

Q: What are the power requirements?

A: The RTX 5090 SE has a 500 W TDP with a 1x 16-pin connector and a suggested 900 W PSU. The N1 20SM has unknown TDP, no power connectors, and no PSU suggestion.

Q: Which GPU has more RT cores and tensor cores?

A: The RTX 5090 SE has 110 RT cores and 440 tensor cores. The N1 20SM has 20 RT cores and 80 tensor cores.

Specification Differences

| Specification | NVIDIA GeForce RTX 5090 SE | NVIDIA N1 20SM |

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

| Chip | GB202 | GB20B |

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

| Process Node | 5 nm | 5 nm |

| Die Size | 750 mm² | 382 mm² |

| Transistors | 92,200 million | unknown |

| Transistor Density | 122.9M / mm² | null |

| 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 Width | 384 bit | 256 bit |

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

| Shading Units | 14080 | 2560 |

| TMUs | 440 | 160 |

| ROPs | 160 | 24 |

| RT Cores | 110 | 20 |

| Tensor Cores | 440 | 80 |

| Pixel Rate | 380.3 GPixel/s | 56.30 GPixel/s |

| Texture Rate | 1,045.9 GTexel/s | 375.4 GTexel/s |

| FP32 | 66.94 TFLOPS | 12.01 TFLOPS |

| FP16 | 66.94 TFLOPS (1:1) | 12.01 TFLOPS (1:1) |

| TDP | 500 W | unknown |

| Slot Width | Dual-slot | IGP |

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

| Suggested PSU | 900 W | null |

| Bus Interface | PCIe 5.0 x16 | PCIe 5.0 x16 |

| 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 |