Intel Arc Graphics 1 Xe Mobile vs NVIDIA H20 NVL16 Comparison

Head-to-Head Benchmarks

The recorded database contains no direct benchmark scores for either the Intel Arc Graphics 1 Xe Mobile or the NVIDIA H20 NVL16. Both parts carry a percentile ranking of 50 against all GPUs, and both have an average benchmark score of 0. With no head-to-head results, wins and losses cannot be assigned. The absence of measured data means the comparison must rely entirely on architectural specifications and theoretical peak rates.

The Intel part delivers a pixel rate of 9.200 GPixel/s and a texture rate of 18.40 GTexel/s. The NVIDIA part delivers 47.52 GPixel/s and 617.8 GTexel/s. That places the NVIDIA accelerator at approximately 5.2 times the pixel throughput and 33.6 times the texture throughput of the Intel integrated graphics. FP32 compute tells a similar story: the Intel part is rated at 588.8 GFLOPS, while the NVIDIA part reaches 39.54 TFLOPS. The NVIDIA part also offers FP16 at 79.07 TFLOPS (2:1), compared to 1,177.6 GFLOPS (2:1) for the Intel part.

These are not comparable product classes. The Intel part is an integrated graphics processor with a 25 W TDP, while the NVIDIA part is a 400 W server module. The raw compute gap is expected given the power envelope and physical footprint. The data indicates a 67-fold difference in FP32, a 67-fold difference in FP16, a 5.2-fold difference in pixel rate, and a 33.6-fold difference in texture rate, all in favor of the NVIDIA H20 NVL16.

Architecture Differences

The Intel Arc Graphics 1 Xe Mobile uses the Wildcat Lake chip built on Intel's Xe3-LPG architecture, fabricated on a 3 nm process at Intel's foundry. It belongs to the Arc Graphics-M (Wildcat Lake) generation and lists its predecessor as HD Graphics-M. The NVIDIA H20 NVL16 uses the GH100 chip built on the Hopper architecture, fabricated on a 5 nm process at TSMC. It belongs to the Server Hopper (Hxx) generation, with its predecessor listed as Server Ada and its successor as Server Blackwell.

The Intel part has 128 shading units, 8 texture mapping units, 4 raster output units, and 1 ray tracing core. It has no listed tensor cores. The NVIDIA part has 9,984 shading units, 312 texture mapping units, 24 raster output units, and 312 tensor cores. The NVIDIA part does not list dedicated RT cores in the database, but its tensor core count is substantial. Transistor counts differ dramatically: the Intel part is listed as unknown, while the NVIDIA part carries 80,000 million transistors on an 814 mm² die, yielding a transistor density of 98.3M / mm².

Memory architecture is a fundamental split. The Intel part uses system shared memory, with a system dependent bandwidth and a memory clock listed as System Shared. The NVIDIA part uses 96 GB of HBM3 on a 6144 bit bus, with a memory clock of 1313 MHz (5.3 Gbps effective) and a bandwidth of 4.03 TB/s. The NVIDIA memory subsystem is a discrete, dedicated pool, while the Intel part depends entirely on the host system's memory.

API support also diverges. The Intel part supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA part lists N/A for DirectX, OpenGL, and Vulkan. This reflects their intended roles: the Intel part is a client-side integrated GPU for portable devices, while the NVIDIA part is a server accelerator with no display outputs. The Intel part's display outputs are listed as Portable Device Dependent, whereas the NVIDIA part has no outputs at all.

The bus interfaces differ as well. The Intel part uses an IGP bus interface and an IGP slot width, with no power connectors. The NVIDIA part uses PCIe 5.0 x16, an SXM Module slot width, and a suggested PSU of 800 W. The TDP figures are 25 W for Intel and 400 W for NVIDIA, a 16-fold difference in power draw.

The Verdict

The data indicates two devices built for entirely different environments. The Intel Arc Graphics 1 Xe Mobile is an integrated GPU with a 25 W TDP, system shared memory, and support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. Its production status is Active, and its release date is recorded as April 15, 2026. The NVIDIA H20 NVL16 is a 400 W server module with 96 GB of HBM3, 4.03 TB/s of bandwidth, 312 tensor cores, and no display outputs. Its production status is also Active, with a release date of September 1, 2025.

For client-side rendering workloads that require DirectX 12 Ultimate features, the Intel part is the only one of the two with API support in the database. The NVIDIA part lists N/A for all three major graphics APIs, which points to a compute-oriented server role rather than a rendering role. For compute-heavy server tasks, the NVIDIA part offers 39.54 TFLOPS FP32, 79.07 TFLOPS FP16, and a 4.03 TB/s memory bandwidth, all of which dwarf the Intel part's capabilities.

The 50th percentile ranking for both parts against all GPUs is a neutral data point. With no benchmark scores recorded, the percentile cannot be interpreted as a performance tier. The database shows zero average benchmark scores for both, so the percentile values may reflect the absence of data rather than measured performance.

Specification Differences

| Specification | Intel Arc Graphics 1 Xe Mobile | NVIDIA H20 NVL16 |

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

| Chip | Wildcat Lake | GH100 |

| Architecture | Xe3-LPG | Hopper |

| Process Node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Transistors | unknown | 80,000 million |

| Die Size | unknown | 814 mm² |

| Transistor Density | null | 98.3M / mm² |

| Base Clock | 300 MHz | 1830 MHz |

| Boost Clock | 2300 MHz | 1980 MHz |

| Memory Clock | System Shared | 1313 MHz 5.3 Gbps effective |

| Memory Size | System Shared | 96 GB |

| Memory Type | System Shared | HBM3 |

| Memory Bus Width | System Shared | 6144 bit |

| Memory Bandwidth | System Dependent | 4.03 TB/s |

| Shading Units | 128 | 9984 |

| TMUs | 8 | 312 |

| ROPs | 4 | 24 |

| RT Cores | 1 | null |

| Tensor Cores | null | 312 |

| Pixel Rate | 9.200 GPixel/s | 47.52 GPixel/s |

| Texture Rate | 18.40 GTexel/s | 617.8 GTexel/s |

| FP32 | 588.8 GFLOPS | 39.54 TFLOPS |

| FP16 | 1,177.6 GFLOPS (2:1) | 79.07 TFLOPS (2:1) |

| TDP | 25 W | 400 W |

| Slot Width | IGP | SXM Module |

| Power Connectors | None | null |

| Suggested PSU | null | 800 W |

| Bus Interface | IGP | PCIe 5.0 x16 |

| Display Outputs | Portable Device Dependent | No outputs |

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

| OpenGL | 4.6 | N/A |

| Vulkan | 1.4 | N/A |

| Release Date | 2026-04-15 | 2025-09-01 |

| Predecessor | HD Graphics-M | Server Ada |

| Successor | null | Server Blackwell |

FAQ

Q: Which GPU has more shading units?

A: The NVIDIA H20 NVL16 has 9,984 shading units, while the Intel Arc Graphics 1 Xe Mobile has 128 shading units.

Q: What memory configurations do the two GPUs use?

A: The Intel part uses system shared memory with system dependent bandwidth. The NVIDIA part uses 96 GB of HBM3 on a 6144 bit bus with 4.03 TB/s of bandwidth.

Q: Do both GPUs support DirectX?

A: No. The Intel part supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA part lists N/A for DirectX, OpenGL, and Vulkan.

Q: What are the TDP figures?

A: The Intel Arc Graphics 1 Xe Mobile has a 25 W TDP. The NVIDIA H20 NVL16 has a 400 W TDP and a suggested PSU of 800 W.

Q: Which GPU has tensor cores?

A: The NVIDIA H20 NVL16 has 312 tensor cores. The Intel Arc Graphics 1 Xe Mobile has no listed tensor cores.

Q: What is the release date for each GPU?

A: The Intel part has a release date of April 15, 2026. The NVIDIA part has a release date of September 1, 2025. Both are listed as Active in production status.

Where Each One Wins

The Intel Arc Graphics 1 Xe Mobile wins in client-side graphics API support. It is the only one of the two with DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 support. It also carries display outputs, listed as Portable Device Dependent, which makes it suitable for integrated portable systems. Its 25 W TDP and IGP slot width with no power connectors indicate a low-power integrated solution. The 3 nm process node gives it a manufacturing advantage in density per watt, though no transistor density figure is recorded for it.

The NVIDIA H20 NVL16 wins in raw compute, memory capacity, and bandwidth. Its 39.54 TFLOPS FP32 and 79.07 TFLOPS FP16 figures are orders of magnitude above the Intel part. The 96 GB HBM3 pool with 4.03 TB/s bandwidth is a server-class memory subsystem. The 312 tensor cores give it a dedicated path for tensor workloads, which the Intel part lacks entirely. The 80,000 million transistor count on an 814 mm² die with 98.3M / mm² density reflects a large, specialized compute die.

For rendering workloads on portable devices, the Intel part has the API support and display outputs required. For server-side compute tasks that rely on FP16 throughput and tensor operations, the NVIDIA part is the only one with the necessary hardware. The two parts do not compete in the same market segment. The database records no overlapping benchmarks, and the specification sheets confirm separate design goals: one is a low-power integrated GPU for mobile systems, the other is a high-power accelerator for server deployments.