NVIDIA GeForce RTX 4090 Mobile vs NVIDIA H20 Comparison

Where Each One Wins

The database records only one of these two accelerators with benchmark results. The NVIDIA GeForce RTX 4090 Mobile has a full set of nine benchmark scores across DirectX, OpenCL, Vulkan, and compute workloads, while the NVIDIA H20 has no recorded benchmark entries. The RTX 4090 Mobile's average benchmark score of 43,667 places it in the 84th percentile of all GPUs tracked in the database. The H20 sits at the 50th percentile with an average score of zero, indicating a lack of measured performance data rather than a literal absence of capability.

The RTX 4090 Mobile's benchmark wins are concentrated in graphics-oriented workloads. Its PassMark G3D score of 27,212 reflects strong rasterization performance, while the PassMark DirectX 11 result of 262 and DirectX 9 result of 310 show broad legacy API competence. The OpenCL score of 180,831 and Vulkan score of 170,774 indicate substantial compute throughput for a mobile part. The GPU compute score of 12,347 in PassMark's compute test rounds out a profile that excels across both graphics and general-purpose workloads.

The H20, by contrast, has no wins because the database contains no measured results for it. Its positioning as a server Hopper part with a 500 W TDP and SXM module form factor suggests a different use case entirely: data center inference and training rather than client-side rendering. The H20's architecture targets throughput with tensor cores and high memory bandwidth, but without benchmark data, its relative standing cannot be quantified. The RTX 4090 Mobile therefore wins every recorded comparison by default, though the lack of H20 results means these wins reflect data availability as much as actual performance differences.

The percentile gap is telling. The RTX 4090 Mobile's 84th percentile ranking places it among the top tier of all GPUs in the database, whereas the H20's 50th percentile is the median position assigned to parts with no scores. This is not a statement about the H20's real-world capability; it is a statement about the completeness of the recorded data. The RTX 4090 Mobile's nearest rivals in the database include the NVIDIA Quadro M6000 at 43,301 average score (0.8% behind), the GeForce RTX 5050 Mobile at 43,268 (0.9% behind), the Quadro M6000 24 GB at 43,262 (0.9% behind), and the RTX A6000 at 44,075 (0.9% ahead). These deltas are small, indicating that the RTX 4090 Mobile sits in a tightly clustered performance band.

Architecture Differences

The two accelerators share a manufacturer and a process node but diverge sharply in nearly every other architectural dimension. The RTX 4090 Mobile uses the AD103 chip built on the Ada Lovelace architecture, fabricated by TSMC on a 5 nm process. The H20 uses the GH100 chip built on the Hopper architecture, also fabricated by TSMC on 5 nm. Both use the same process node, which is where the similarities end.

The chip scale differs substantially. The H20's GH100 die contains 80,000 million transistors on a 814 mm² die, while the RTX 4090 Mobile's AD103 contains 45,900 million transistors on a 379 mm² die. The H20's die is more than twice the physical size and carries nearly twice the transistor count. Transistor density tells a different story: the RTX 4090 Mobile achieves 121.1 million transistors per square millimeter, while the H20 manages 98.3 million per square millimeter. The smaller, denser AD103 die reflects a design optimized for mobile power envelopes, whereas the sprawling GH100 die prioritizes raw throughput.

Memory subsystems diverge completely. The RTX 4090 Mobile uses 16 GB of GDDR6 on a 256-bit bus, delivering 576.0 GB/s of bandwidth. The H20 uses 96 GB of HBM3 on a 6144-bit bus, delivering 4.03 TB/s of bandwidth. The H20's memory capacity is six times larger, and its bandwidth is roughly seven times higher. The bus width difference is stark: 256 bits versus 6144 bits, a 24-fold gap that reflects the fundamentally different memory architectures of client GPUs versus data center accelerators.

Compute resources show both similarities and differences. The RTX 4090 Mobile has 9,728 shading units, 304 TMUs, and 112 ROPs. The H20 has 9,984 shading units, 312 TMUs, and only 24 ROPs. The shading unit and TMU counts are nearly identical, but the ROP count is dramatically different: the H20 has less than a quarter of the RTX 4090 Mobile's ROPs. This reflects the H20's orientation toward compute rather than rasterization, where ROP throughput matters less. The H20 also lacks recorded ray tracing cores, while the RTX 4090 Mobile has 76 of them. Both have tensor cores, with the RTX 4090 Mobile carrying 304 and the H20 carrying 312.

Clock speeds favor the H20. The H20 has a base clock of 1830 MHz and a boost clock of 1980 MHz, compared to the RTX 4090 Mobile's 1335 MHz base and 1695 MHz boost. The H20's boost clock is nearly 17% higher. Memory clocks also differ: the RTX 4090 Mobile runs at 2250 MHz with 18 Gbps effective, while the H20 runs at 1313 MHz with 5.3 Gbps effective. The HBM3's wider bus compensates for the lower clock rate.

The form factors and interfaces reflect their intended environments. The RTX 4090 Mobile is an IGP (integrated graphics processor) with no power connectors, designed to be soldered into laptops. The H20 is an SXM module with a 500 W TDP and a suggested PSU of 900 W. The RTX 4090 Mobile's TDP is 120 W, less than a quarter of the H20's. The bus interfaces also differ: the RTX 4090 Mobile uses PCIe 4.0 x16, while the H20 uses PCIe 5.0 x16.

Head-to-Head Benchmarks

The database contains no head-to-head benchmark entries for these two parts. This is a direct consequence of the H20 having no recorded benchmark scores at all. The RTX 4090 Mobile's individual results stand alone, and the H20's performance cannot be compared numerically.

What the data does show is the RTX 4090 Mobile's absolute performance levels. Its strongest results come in the compute-oriented tests: Geekbench OpenCL at 180,831 and Geekbench Vulkan at 170,774. These scores are more than six times higher than the PassMark G3D score of 27,212, illustrating that the part's compute throughput far exceeds its rasterization score in the PassMark suite. The PassMark DirectX results are comparatively modest: 310 for DirectX 9, 262 for DirectX 11, 173 for DirectX 10, and 107 for DirectX 12. The DirectX 12 score being the lowest of the four suggests that the PassMark DirectX 12 test may not fully utilize the Ada Lovelace architecture's features, or that the test's workload favors different GPU generations.

The RTX 4090 Mobile's average benchmark score of 43,667 is remarkably close to its nearest rivals. The Quadro M6000 trails by 0.8%, the RTX 5050 Mobile by 0.9%, and the Quadro M6000 24 GB by 0.9%. The RTX A6000 leads by 0.9%. These sub-one-percent deltas indicate that the RTX 4090 Mobile sits in an extremely tight competitive cluster, where the difference between first and fourth place is less than 2%. The RTX A6000's status as the only rival ahead of the RTX 4090 Mobile by any margin suggests that the mobile part punches above its power envelope in the database's metrics.

Without H20 benchmark data, any head-to-head analysis must remain qualitative. The H20's architecture suggests strengths in FP16 compute, where it delivers 79.07 TFLOPS at a 2:1 ratio, and in memory bandwidth, where its 4.03 TB/s dwarfs the RTX 4090 Mobile's 576.0 GB/s. The RTX 4090 Mobile's strengths lie in graphics APIs: it supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, while the H20 lists no graphics API support. The H20 has no display outputs, while the RTX 4090 Mobile's outputs are portable device dependent.

FAQ

Q: Why does the NVIDIA H20 have an average benchmark score of zero?

A: The database contains no recorded benchmark entries for the H20. Its average score is therefore calculated as zero, and its percentile ranking of 50 reflects the median position assigned to GPUs without measured performance data.

Q: How does the RTX 4090 Mobile compare to its nearest rivals in the database?

A: The RTX 4090 Mobile's average score of 43,667 is 0.8% ahead of the Quadro M6000, 0.9% ahead of the GeForce RTX 5050 Mobile, and 0.9% ahead of the Quadro M6000 24 GB. The RTX A6000 is 0.9% ahead of the RTX 4090 Mobile.

Q: What are the key memory differences between the two GPUs?

A: The RTX 4090 Mobile has 16 GB of GDDR6 on a 256-bit bus with 576.0 GB/s bandwidth. The H20 has 96 GB of HBM3 on a 6144-bit bus with 4.03 TB/s bandwidth. The H20's capacity is six times larger and its bandwidth is about seven times higher.

Q: Which GPU has higher clock speeds?

A: The H20 has a base clock of 1830 MHz and a boost clock of 1980 MHz. The RTX 4090 Mobile has a base clock of 1335 MHz and a boost clock of 1695 MHz. The H20's boost clock is approximately 17% higher.

Q: Do both GPUs support the same graphics APIs?

A: No. The RTX 4090 Mobile supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The H20 lists no graphics API support, with its DirectX, OpenGL, and Vulkan entries all marked as N/A.

Q: What are the transistor counts and die sizes?

A: The H20's GH100 chip has 80,000 million transistors on a 814 mm² die. The RTX 4090 Mobile's AD103 chip has 45,900 million transistors on a 379 mm² die. The H20 has nearly twice the transistor count and a die more than twice as large.

Specification Differences

| Specification | NVIDIA GeForce RTX 4090 Mobile | NVIDIA H20 |

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

| Architecture | Ada Lovelace | Hopper |

| Generation | GeForce 40 Mobile | Server Hopper (Hxx) |

| Chip | AD103 | GH100 |

| Transistors | 45,900 million | 80,000 million |

| Die Size | 379 mm² | 814 mm² |

| Transistor Density | 121.1M / mm² | 98.3M / mm² |

| Base Clock | 1335 MHz | 1830 MHz |

| Boost Clock | 1695 MHz | 1980 MHz |

| Memory Size | 16 GB | 96 GB |

| Memory Type | GDDR6 | HBM3 |

| Memory Bus Width | 256 bit | 6144 bit |

| Memory Bandwidth | 576.0 GB/s | 4.03 TB/s |

| Memory Clock | 2250 MHz (18 Gbps effective) | 1313 MHz (5.3 Gbps effective) |

| Shading Units | 9728 | 9984 |

| TMUs | 304 | 312 |

| ROPs | 112 | 24 |

| RT Cores | 76 | None recorded |

| Tensor Cores | 304 | 312 |

| Pixel Rate | 189.8 GPixel/s | 47.52 GPixel/s |

| Texture Rate | 515.3 GTexel/s | 617.8 GTexel/s |

| FP32 Performance | 32.98 TFLOPS | 39.54 TFLOPS |

| FP16 Performance | 32.98 TFLOPS (1:1) | 79.07 TFLOPS (2:1) |

| TDP | 120 W | 500 W |

| Slot Width | IGP | SXM Module |

| Power Connectors | None | Not recorded |

| Suggested PSU | Not recorded | 900 W |

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

| Display Outputs | Portable Device Dependent | No outputs |

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

| OpenGL Support | 4.6 | N/A |

| Vulkan Support | 1.4 | N/A |

| Release Date | 2023-01-02 | 2024-01-31 |

| Predecessor | GeForce 30 Mobile | Server Ada |

| Successor | GeForce 50 Mobile | Server Blackwell |

| Average Benchmark Score | 43,667 | 0 |

| Percentile vs All GPUs | 84 | 50 |