Intel Arc Pro B370 vs NVIDIA H800 PCIe 80 GB Comparison

Head-to-Head Benchmarks

The database contains no recorded benchmark scores for either the Intel Arc Pro B370 or the NVIDIA H800 PCIe 80 GB. Both entries show an average benchmark score of zero, and the head-to-head comparison table is empty. Consequently, the win count for each part is zero, and no direct performance deltas can be computed from the recorded data. The percentile ranking for both products sits at exactly the 50th mark relative to all GPUs in the database, which in the absence of measured scores reflects their positional midpoint rather than any verified performance equivalence.

What can be established from the recorded specifications is a stark contrast in raw compute ceilings. The NVIDIA H800 PCIe 80 GB delivers a FP32 throughput of 51.22 TFLOPS, which is roughly 8.3 times the 6.144 TFLOPS of the Intel Arc Pro B370. In FP16 workloads, the divergence widens further: the H800 reaches 204.9 TFLOPS using a 4:1 ratio, while the Arc Pro B370 manages 12.29 TFLOPS via a 2:1 ratio. That places the NVIDIA part at approximately 16.7 times the FP16 throughput of the Intel part, assuming both operate at their rated formats. The texture rate tells a similar story, with the H800 posting 800.3 GTexel/s against the Arc Pro B370's 96.00 GTexel/s, an 8.3-fold advantage. Pixel rate is the one metric where the Intel part leads: 48.00 GPixel/s versus 42.12 GPixel/s, a modest 14% edge for the Arc Pro B370.

Memory capacity and bandwidth are equally lopsided. The H800 carries 80 GB of HBM2e across a 5120-bit bus, delivering 2.04 TB/s of bandwidth. The Arc Pro B370 uses system shared memory, with its bandwidth listed as system dependent and its bus width as system shared, meaning no fixed memory bandwidth figure exists in the database. The H800's memory clock is 1593 MHz, translating to 3.2 Gbps effective, while the Intel part's memory clock is likewise system shared. These figures indicate that the H800 is engineered for data-intensive server workloads, whereas the Arc Pro B370 relies entirely on the host platform's memory subsystem.

Clock speeds also differ substantially. The Arc Pro B370 idles conceptually at a 300 MHz base and boosts to 2400 MHz, while the H800 runs at a 1095 MHz base and a 1755 MHz boost. The Intel part's higher boost clock does not compensate for its far smaller execution resource pool: 1280 shading units, 40 texture mapping units, and 20 ROPs versus the H800's 14592 shading units, 456 TMUs, and 24 ROPs. The H800 also fields 456 tensor cores, while the Arc Pro B370 lists no tensor core count. Ray tracing cores differ in kind: the Intel part has 10 dedicated RT cores, while the H800 entry records none.

Power consumption and physical format reinforce the intended use cases. The Arc Pro B370 carries a 25 W TDP, has no power connectors, and is integrated directly as an IGP. The H800 draws 350 W, requires a single 16-pin power connector, suggests a 750 W power supply, and occupies a dual-slot PCIe 5.0 x16 card measuring 268 mm in length and 111 mm in height. The H800 has no display outputs, while the Arc Pro B370's display outputs are portable device dependent.

The Verdict

Based strictly on the recorded data, the NVIDIA H800 PCIe 80 GB is the clear choice for compute-bound server tasks. Its FP32 throughput of 51.22 TFLOPS, FP16 throughput of 204.9 TFLOPS, 80 GB HBM2e memory, and 2.04 TB/s bandwidth make it suitable for large-scale numerical workloads, machine learning training, and high-bandwidth data processing. The 456 tensor cores provide dedicated hardware for matrix operations, a feature the Intel part lacks entirely. The H800's 350 W TDP and dual-slot footprint indicate it belongs in a workstation or data center chassis with adequate power delivery and cooling.

The Intel Arc Pro B370 serves a different purpose entirely. Its 25 W TDP, integrated form factor, and system shared memory mean it operates within a portable device, likely a laptop or compact system, where power efficiency and space are paramount. Its higher boost clock of 2400 MHz and its 10 ray tracing cores give it capabilities in graphics rendering and real-time ray tracing workloads that the H800 does not list. The Arc Pro B370 also supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, whereas the H800 entry lists no API support at all in the database, further indicating that the NVIDIA part is not oriented toward conventional graphics rendering.

The pixel rate advantage of the Intel part, 48.00 GPixel/s versus 42.12 GPixel/s, suggests it can drive display outputs with comparable fill rates to the H800 despite its much smaller compute core. However, the H800 has no display outputs, so this metric is irrelevant for that product. The production status of both parts is active, and the H800 was released in March 2023, while the Arc Pro B370 was released in January 2026. The H800's predecessor is Server Ada and its successor is Server Blackwell, while the Arc Pro B370's predecessor is HD Graphics-WM, with no successor listed.

Users who need maximum FP32 or FP16 throughput, large memory capacity, and tensor core acceleration should select the H800. Users who need an integrated graphics solution with ray tracing support, modern graphics APIs, and minimal power draw should select the Arc Pro B370. The data does not support any middle-ground recommendation, as the two products occupy entirely separate performance and physical domains.

FAQ

Q: Which GPU has higher FP32 performance?

A: The NVIDIA H800 PCIe 80 GB delivers 51.22 TFLOPS FP32, which is approximately 8.3 times the 6.144 TFLOPS of the Intel Arc Pro B370.

Q: Does the Intel Arc Pro B370 have more memory than the NVIDIA H800?

A: No. The H800 has 80 GB of HBM2e memory, while the Arc Pro B370 uses system shared memory, meaning it has no dedicated VRAM of its own.

Q: Which product supports ray tracing?

A: The Intel Arc Pro B370 lists 10 dedicated ray tracing cores. The NVIDIA H800 PCIe 80 GB entry records no ray tracing cores.

Q: What is the power consumption difference?

A: The Intel Arc Pro B370 has a 25 W TDP and no power connectors, while the NVIDIA H800 PCIe 80 GB has a 350 W TDP and requires a single 16-pin power connector.

Q: Which GPU has higher texture fill rate?

A: The NVIDIA H800 PCIe 80 GB reaches 800.3 GTexel/s, which is about 8.3 times the 96.00 GTexel/s of the Intel Arc Pro B370.

Q: What are the bus interfaces for each product?

A: The Intel Arc Pro B370 uses an integrated graphics processor (IGP) bus interface, while the NVIDIA H800 PCIe 80 GB uses PCIe 5.0 x16.

Specification Differences

| Specification | Intel Arc Pro B370 | NVIDIA H800 PCIe 80 GB |

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

| 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 | 1095 MHz |

| Boost clock | 2400 MHz | 1755 MHz |

| Memory size | System Shared | 80 GB |

| Memory type | System Shared | HBM2e |

| Memory bus width | System Shared | 5120 bit |

| Memory bandwidth | System Dependent | 2.04 TB/s |

| Shading units | 1280 | 14592 |

| TMUs | 40 | 456 |

| ROPs | 20 | 24 |

| RT cores | 10 | null |

| Tensor cores | null | 456 |

| Pixel rate | 48.00 GPixel/s | 42.12 GPixel/s |

| Texture rate | 96.00 GTexel/s | 800.3 GTexel/s |

| FP32 | 6.144 TFLOPS | 51.22 TFLOPS |

| FP16 | 12.29 TFLOPS (2:1) | 204.9 TFLOPS (4:1) |

| TDP | 25 W | 350 W |

| Slot width | IGP | Dual-slot |

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

| Suggested PSU | null | 750 W |

| Bus interface | IGP | PCIe 5.0 x16 |

| Display outputs | Portable Device Dependent | No outputs |

| DirectX | 12 Ultimate (12_2) | null |

| OpenGL | 4.6 | null |

| Vulkan | 1.4 | null |

| Length | null | 268 mm (10.6 inches) |

| Height | null | 111 mm (4.4 inches) |

| Release date | 2026-01-26 | 2023-03-20 |

| Predecessor | HD Graphics-WM | Server Ada |

| Successor | null | Server Blackwell |

Architecture Differences

The Intel Arc Pro B370 is built on the Panther Lake chip using the Xe3-LPG architecture, fabricated on a 3 nm process at Intel's own foundry. It belongs to the Arc Graphics-WM (Panther Lake) generation. The NVIDIA H800 PCIe 80 GB uses the GH100 chip with the Hopper architecture, fabricated on a 5 nm process at TSMC, and belongs to the Server Hopper (Hxx) generation. The process node difference gives the Intel part a density advantage, though its transistor count and die size are listed as unknown. The H800 contains 80,000 million transistors on an 814 mm² die, yielding a transistor density of 98.3M per mm².

The Intel part's Xe3-LPG architecture provides 1280 shading units, 40 TMUs, and 20 ROPs, with 10 dedicated ray tracing cores. It does not list tensor cores. Its FP16 throughput is achieved at a 2:1 ratio relative to FP32, indicating a simpler execution path. The NVIDIA part's Hopper architecture provides 14592 shading units, 456 TMUs, and 24 ROPs, with 456 tensor cores and no ray tracing cores. Its FP16 throughput is achieved at a 4:1 ratio, meaning it can pack more FP16 operations per cycle. The H800's memory subsystem uses HBM2e across a 5120-bit bus, whereas the Intel part has no dedicated memory interface and relies on the host system's shared memory.

The API support also differs. The Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, making it suitable for modern graphics applications. The H800 lists no DirectX, OpenGL, or Vulkan support in the database, consistent with its role as a compute accelerator without display outputs. The Intel part's integrated nature means it draws power from the host platform with no additional connectors, while the H800 requires a 16-pin connector and a 750 W suggested power supply. The H800's physical dimensions of 268 mm by 111 mm place it as a standard dual-slot PCIe card, whereas the Arc Pro B370 is an IGP with no physical dimensions recorded.