NVIDIA GeForce RTX 4070 SUPER vs NVIDIA GeForce RTX 5090 Mobile Comparison

# Head-to-Head Benchmarks

The benchmark data splits evenly at five wins apiece, but the magnitude of those wins tells a more nuanced story. The RTX 5090 Mobile takes decisive victories in modern DirectX 12 workloads, while the RTX 4070 SUPER counters with equally commanding leads in compute and legacy API tests.

The largest margin of the entire comparison belongs to the RTX 5090 Mobile in 3DMark Steel Nomad DX12, where it scores 5,871 against the RTX 4070 SUPER’s 4,627 — a 26.9% advantage. This is the clearest signal that Blackwell’s architecture scales aggressively in contemporary rendering pipelines. The DirectX 12 PassMark result reinforces this pattern, with the mobile GPU posting 138 versus 110, a 25.5% gap in its favor.

OpenCL performance also favors the RTX 5090 Mobile substantially, with a score of 201,834 compared to 172,795 — a 16.8% lead. This suggests the mobile part carries meaningful advantages in general-purpose compute tasks that leverage this API. The RTX 5090 Mobile also edges ahead in PassMark DirectX 10 (183 vs 167, +9.6%) and PassMark G3D (30,034 vs 29,995, +0.1%), though the latter is essentially a statistical tie.

The RTX 4070 SUPER, however, dominates in compute-heavy and legacy scenarios. Its PassMark GPU Compute score of 17,108 versus 13,401 represents a 21.7% lead — the second-largest delta in the entire dataset. This is a substantial reversal of the OpenCL result and indicates that the desktop card’s higher clock speeds and dedicated power budget pay off in sustained compute workloads. The Vulkan benchmark also goes to the RTX 4070 SUPER, with 205,624 against 198,405, a 3.5% edge. Legacy DirectX 9 performance favors the desktop card by 5.8% (344 vs 324), and DirectX 11 shows a narrow 1.5% advantage (273 vs 269). The 2D graphics score also goes to the RTX 4070 SUPER, 1,184 versus 1,057, a 10.7% margin.

# Architecture Differences

The two GPUs represent entirely different architectural generations from the same manufacturer. The RTX 5090 Mobile is built on the Blackwell 2.0 architecture using the GB203 chip, fabricated on a 5 nm process at TSMC. The RTX 4070 SUPER uses the Ada Lovelace architecture with the AD104 chip, also on TSMC’s 5 nm node. Both use the same foundry and process node, which makes the performance differences particularly interesting from a design perspective rather than a fabrication one.

Transistor counts differ significantly. The GB203 packs 45,600 million transistors on a 378 mm² die, yielding a density of 120.6 million transistors per mm². The AD104 contains 35,800 million transistors on a 294 mm² die, with a density of 121.8 million per mm². Despite the larger absolute transistor budget on the mobile chip, the desktop chip actually has a marginally higher transistor density, suggesting similar design efficiency per area.

Memory subsystems diverge sharply. The RTX 5090 Mobile ships with 24 GB of GDDR7 memory on a 256-bit bus, delivering 896.0 GB/s of bandwidth. The RTX 4070 SUPER offers 12 GB of GDDR6X on a 192-bit bus, with 504.2 GB/s bandwidth. This is a 77.7% bandwidth advantage for the mobile part and double the capacity. Memory clock rates also differ: the mobile chip runs at 1750 MHz (28 Gbps effective) while the desktop card runs at 1313 MHz (21 Gbps effective).

Core configurations are heavily lopsided. The RTX 5090 Mobile has 10,496 shading units, 328 TMUs, 112 ROPs, 82 RT cores, and 328 tensor cores. The RTX 4070 SUPER has 7,168 shading units, 224 TMUs, 80 ROPs, 56 RT cores, and 224 tensor cores. The mobile chip carries roughly 46% more shading units and 46% more RT cores. Yet the desktop card posts higher raw FP32 throughput at 35.48 TFLOPS versus 31.80 TFLOPS, due to its much higher clock speeds of 2475 MHz boost versus 1515 MHz boost. Pixel fill rates favor the desktop card at 198.0 GPixel/s versus 169.7 GPixel/s, while texture rates are closer: 554.4 GTexel/s versus 496.9 GTexel/s.

Power and physical characteristics could hardly be more different. The RTX 5090 Mobile is rated at 95 W TDP, is an integrated graphics package (IGP) with no power connectors, and draws power directly from the laptop motherboard. The RTX 4070 SUPER consumes 220 W, requires a 16-pin power connector, suggests a 550 W PSU, and occupies dual slots with dimensions of 267 mm x 112 mm x 42 mm. The desktop card uses PCIe 4.0 x16 while the mobile part uses PCIe 5.0 x16.

# Where Each One Wins

The RTX 5090 Mobile is the clear choice for modern DirectX 12 gaming and general OpenCL compute workloads. Its 26.9% lead in 3DMark Steel Nomad DX12 and 25.5% lead in PassMark DirectX 12 indicate that games built around current rendering APIs will see substantial benefits. The 16.8% OpenCL advantage also positions it well for applications that leverage OpenCL for acceleration, such as certain content creation and scientific workloads. With 24 GB of GDDR7 memory and 896.0 GB/s bandwidth, it also handles large datasets and high-resolution textures with far more headroom.

The RTX 4070 SUPER wins in Vulkan, legacy DirectX 9 and 11 titles, 2D desktop workloads, and GPU compute benchmarks. Its 21.7% lead in PassMark GPU Compute is particularly notable, suggesting that compute shaders and GPGPU tasks that use this benchmark’s methodology will run noticeably faster. The 3.5% Vulkan advantage is relevant for games that use Vulkan as their primary API, such as many Linux ports and certain AAA titles. The 10.7% lead in PassMark G2D indicates better 2D rendering performance, which matters for desktop environments and productivity apps. The DirectX 9 and 11 wins, at 5.8% and 1.5% respectively, make it the better option for older game libraries.

# FAQ

Q: Which GPU has the higher average benchmark score?

A: The RTX 5090 Mobile has an average benchmark score of 45,152, which is 4.5% higher than the RTX 4070 SUPER’s 43,223. The RTX 5090 Mobile also ranks in the 84th percentile of all GPUs, versus the 83rd percentile for the RTX 4070 SUPER.

Q: How does the RTX 5090 Mobile compare to the RTX 4070 Ti?

A: The RTX 5090 Mobile has an average score of 45,152, which is 0.8% ahead of the RTX 4070 Ti’s 44,795. This places it in the same performance tier as the desktop RTX 4070 Ti.

Q: What is the launch MSRP of the RTX 4070 SUPER?

A: The RTX 4070 SUPER has a launch MSRP of 599 USD. The RTX 5090 Mobile has no listed launch MSRP in the data.

Q: Which card has better Vulkan performance?

A: The RTX 4070 SUPER wins in Geekbench Vulkan with a score of 205,624, which is 3.5% higher than the RTX 5090 Mobile’s 198,405.

Q: How do the memory capacities compare?

A: The RTX 5090 Mobile has 24 GB of GDDR7 memory on a 256-bit bus with 896.0 GB/s bandwidth. The RTX 4070 SUPER has 12 GB of GDDR6X on a 192-bit bus with 504.2 GB/s bandwidth.

Q: Which GPU is more power-efficient?

A: The RTX 5090 Mobile is rated at 95 W TDP, while the RTX 4070 SUPER is rated at 220 W. Despite the lower power envelope, the mobile chip delivers higher average benchmark scores.

# The Verdict

The data points to a split decision based on use case. For gamers focused on modern DirectX 12 titles, the RTX 5090 Mobile is the clear winner, offering a 26.9% lead in 3DMark Steel Nomad DX12 and a 25.5% lead in PassMark DirectX 12. Its 24 GB memory capacity and 896.0 GB/s bandwidth also future-proof it for high-resolution textures and large assets. The 84th percentile ranking and average score of 45,152, sitting just above the RTX 4070 Ti, confirm it as a high-end performer despite its 95 W mobile power envelope.

The RTX 4070 SUPER, however, is the better choice for compute-centric workflows and users who prioritize Vulkan or legacy API performance. Its 21.7% lead in PassMark GPU Compute and 3.5% advantage in Vulkan make it more suitable for applications that stress these paths. The 10.7% lead in 2D graphics also matters for productivity work. With an 83rd percentile ranking and average score of 43,223, it remains a strong performer, particularly given its 220 W desktop power budget that allows sustained high clocks.

Pick the RTX 5090 Mobile if you need maximum performance in modern DirectX 12 games, require large memory capacity, or want a high-end GPU in a laptop form factor with only a 95 W TDP. Pick the RTX 4070 SUPER if your workload leans on compute shaders, Vulkan rendering, legacy DirectX titles, or 2D productivity tasks, and you have a desktop chassis with 550 W PSU capacity. The equal 5-5 win split in benchmarks underscores that neither GPU is universally dominant; the correct choice depends entirely on which APIs and workloads matter most to you.

# Specification Differences

| Specification | NVIDIA GeForce RTX 5090 Mobile | NVIDIA GeForce RTX 4070 SUPER |

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

| Architecture | Blackwell 2.0 | Ada Lovelace |

| Chip | GB203 | AD104 |

| Process Node | 5 nm | 5 nm |

| Transistors | 45,600 million | 35,800 million |

| Die Size | 378 mm² | 294 mm² |

| Base Clock | 990 MHz | 1980 MHz |

| Boost Clock | 1515 MHz | 2475 MHz |

| Memory Size | 24 GB | 12 GB |

| Memory Type | GDDR7 | GDDR6X |

| Memory Bus | 256 bit | 192 bit |

| Memory Bandwidth | 896.0 GB/s | 504.2 GB/s |

| Shading Units | 10496 | 7168 |

| TMUs | 328 | 224 |

| ROPs | 112 | 80 |

| RT Cores | 82 | 56 |

| Tensor Cores | 328 | 224 |

| FP32 Performance | 31.80 TFLOPS | 35.48 TFLOPS |

| Pixel Rate | 169.7 GPixel/s | 198.0 GPixel/s |

| Texture Rate | 496.9 GTexel/s | 554.4 GTexel/s |

| TDP | 95 W | 220 W |

| Slot Width | IGP | Dual-slot |

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

| Bus Interface | PCIe 5.0 x16 | PCIe 4.0 x16 |

| Production Status | Active | End-of-life |

| Release Date | 2025-03-26 | 2024-01-16 |

| Launch MSRP | — | 599 USD |