Intel Arc Pro B370 vs NVIDIA H100 PCIe 96 GB Comparison

Head-to-Head Benchmarks

The recorded database contains no direct benchmark scores for either the Intel Arc Pro B370 or the NVIDIA H100 PCIe 96 GB. Both products register an average benchmark score of zero, and the head-to-head comparison field is empty. Consequently, there are no measured wins for either part, and the wins tally sits at zero for both sides.

Without synthetic or real-world application scores, the performance relationship between these two accelerators must be inferred from their recorded hardware specifications rather than from direct measurement. The most striking contrast appears in raw compute throughput. The NVIDIA H100 PCIe 96 GB delivers 62.08 TFLOPS of FP32 performance, which is roughly ten times the 6.144 TFLOPS produced by the Intel Arc Pro B370. In FP16 workloads, the gap widens further: the H100 reaches 248.3 TFLOPS using a 4:1 ratio, while the Intel part achieves 12.29 TFLOPS with a 2:1 ratio.

Memory bandwidth tells a similarly lopsided story. The H100 PCIe 96 GB pairs 96 GB of HBM3 memory with a 5120-bit bus, yielding 3.36 TB/s of bandwidth. The Intel Arc Pro B370 relies on system shared memory, with bandwidth listed as system dependent. That dependency means the Intel GPU's effective memory throughput is bounded by the host platform's memory subsystem, which in practice falls far below the dedicated HBM3 stack on the NVIDIA card.

Texture and pixel throughput also favor the NVIDIA part, though less dramatically. The H100 records 969.9 GTexel/s and 44.09 GPixel/s. The Arc Pro B370 produces 96.00 GTexel/s and 48.00 GPixel/s. The pixel rate is the one specification where the Intel part leads, marginally ahead by roughly 9%. This suggests that for rasterization-bound tasks with simple shading, the Intel integrated graphics can keep pace with or slightly exceed the NVIDIA server accelerator. However, the H100's texture rate advantage of over ten times indicates a substantial lead in fill-rate-heavy workloads.

Both products sit at the 50th percentile in the database's all-GPU ranking, which reflects the absence of benchmark data rather than equivalent real-world performance. The architectural gulf between a 25 W integrated graphics processor and a 700 W dual-slot server accelerator is too wide for the missing benchmark scores to be interpreted as parity.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The NVIDIA H100 PCIe 96 GB delivers 62.08 TFLOPS of FP32 performance, compared to 6.144 TFLOPS for the Intel Arc Pro B370. The NVIDIA part is roughly ten times faster in this metric.

Q: How does memory capacity compare between the two?

A: The NVIDIA H100 PCIe 96 GB includes 96 GB of dedicated HBM3 memory with a 5120-bit bus and 3.36 TB/s bandwidth. The Intel Arc Pro B370 uses system shared memory, with capacity and bandwidth determined by the host system.

Q: What is the difference in power consumption?

A: The Intel Arc Pro B370 has a 25 W TDP and uses no external power connectors, functioning as an integrated graphics processor. The NVIDIA H100 PCIe 96 GB has a 700 W TDP and requires an 8-pin EPS power connector, with a suggested power supply of 1100 W.

Q: Do both GPUs support ray tracing?

A: The Intel Arc Pro B370 includes 10 ray tracing cores and supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA H100 PCIe 96 GB does not list ray tracing cores in the database and has no recorded API support entries.

Q: Which GPU has more shading units?

A: The NVIDIA H100 PCIe 96 GB contains 16,896 shading units, while the Intel Arc Pro B370 has 1,280 shading units. The NVIDIA part also features 528 texture mapping units and 528 tensor cores, versus 40 TMUs and no listed tensor cores for the Intel part.

Q: What are the physical form factor differences?

A: The Intel Arc Pro B370 is an integrated graphics processor (IGP) with a slot width of IGP and no dimensions listed. The NVIDIA H100 PCIe 96 GB is a dual-slot card measuring 268 mm in length and 111 mm in height, using a PCIe 5.0 x16 bus interface, and it has no display outputs.

Architecture Differences

The two processors come from different manufacturers and are built on different process nodes. Intel fabricates the Arc Pro B370 on a 3 nm process at its own foundry, while NVIDIA uses TSMC's 5 nm node for the GH100 chip that powers the H100 PCIe 96 GB. The Intel part belongs to the Xe3-LPG architecture, specifically the Arc Graphics-WM (Panther Lake) generation, and it replaces the HD Graphics-WM. The NVIDIA part uses the Hopper architecture, falls into the Server Hopper (Hxx) generation, and succeeds Server Ada while being succeeded by Server Blackwell.

Transistor counts differ enormously. The GH100 chip packs 80,000 million transistors on a die size of 814 mm², yielding a transistor density of 98.3 million per square millimeter. The Intel Arc Pro B370 does not have recorded transistor or die size data. This disparity reflects the fundamental design goals: one is a compact integrated GPU meant to share system resources, while the other is a massive dedicated accelerator built for server workloads.

The NVIDIA H100 PCIe 96 GB carries 16,896 shading units, 528 texture mapping units, 24 ROPs, and 528 tensor cores, with no ray tracing cores listed. The Intel Arc Pro B370 has 1,280 shading units, 40 TMUs, 20 ROPs, and 10 ray tracing cores, with no tensor cores listed. The presence of ray tracing cores on the Intel part and tensor cores on the NVIDIA part indicates different workload priorities, with the former targeting graphics features and the latter focused on matrix operations.

Clock speeds favor the NVIDIA part in absolute terms. The H100 PCIe 96 GB runs at a base clock of 1665 MHz with a boost clock of 1837 MHz. The Arc Pro B370 has a 300 MHz base clock and a 2400 MHz boost clock. Despite the lower base clock, the Intel part's higher boost clock suggests it can scale up significantly under load, though its power envelope is far more constrained.

Memory architecture is entirely different. The NVIDIA H100 PCIe 96 GB uses 96 GB of HBM3 with a 5120-bit bus and 3.36 TB/s bandwidth, clocked at 1313 MHz with 5.3 Gbps effective speed. The Intel Arc Pro B370 uses system shared memory with no dedicated VRAM, a system shared bus width, and system dependent bandwidth. This means the Intel GPU competes with the CPU for memory access, while the NVIDIA card has exclusive access to its own high-bandwidth stack.

API support also diverges. The Intel Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA H100 PCIe 96 GB has no recorded API support in the database, consistent with its role as a compute-oriented server accelerator with no display outputs. The Intel part's display outputs are described as portable device dependent, while the NVIDIA card has no outputs at all.

The Verdict

The data clearly separates these two products into distinct categories. The Intel Arc Pro B370 is an integrated graphics solution with a 25 W TDP, no external power connectors, and system shared memory. The NVIDIA H100 PCIe 96 GB is a 700 W dual-slot server accelerator with 96 GB of dedicated HBM3, 528 tensor cores, and a PCIe 5.0 x16 interface.

For FP32 and FP16 compute workloads, the NVIDIA H100 PCIe 96 GB is the superior choice by a wide margin, delivering 62.08 TFLOPS and 248.3 TFLOPS respectively. Its 3.36 TB/s memory bandwidth and 16,896 shading units provide the resources needed for large-scale parallel processing. The Intel Arc Pro B370's 6.144 TFLOPS FP32 and 12.29 TFLOPS FP16 performance, combined with system dependent memory bandwidth, place it in a fundamentally different performance class.

For graphics-oriented tasks that leverage ray tracing, the Intel Arc Pro B370 has a structural advantage with its 10 ray tracing cores and support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The NVIDIA H100 PCIe 96 GB lacks listed ray tracing cores and display outputs, making it unsuitable for conventional graphics rendering. The Intel part's 48.00 GPixel/s pixel rate also edges out the NVIDIA card's 44.09 GPixel/s.

The choice between these two depends strictly on workload type. Server-side compute, AI inference, and training workloads align with the NVIDIA H100 PCIe 96 GB's tensor core count, massive memory pool, and high FP16 throughput. Integrated graphics duties in a portable device, where low power consumption and ray tracing support matter, point toward the Intel Arc Pro B370. The 25 W TDP versus 700 W TDP difference further reinforces that these parts serve mutually exclusive use cases.

Specification Differences

| Specification | Intel Arc Pro B370 | NVIDIA H100 PCIe 96 GB |

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

| Process node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Transistors | Unknown | 80,000 million |

| Die size | Unknown | 814 mm² |

| Transistor density | Not listed | 98.3M / mm² |

| Base clock | 300 MHz | 1665 MHz |

| Boost clock | 2400 MHz | 1837 MHz |

| Memory size | System Shared | 96 GB |

| Memory type | System Shared | HBM3 |

| Memory bus width | System Shared | 5120 bit |

| Memory bandwidth | System Dependent | 3.36 TB/s |

| Shading units | 1280 | 16896 |

| TMUs | 40 | 528 |

| ROPs | 20 | 24 |

| Ray tracing cores | 10 | Not listed |

| Tensor cores | Not listed | 528 |

| FP32 performance | 6.144 TFLOPS | 62.08 TFLOPS |

| FP16 performance | 12.29 TFLOPS (2:1) | 248.3 TFLOPS (4:1) |

| Pixel rate | 48.00 GPixel/s | 44.09 GPixel/s |

| Texture rate | 96.00 GTexel/s | 969.9 GTexel/s |

| TDP | 25 W | 700 W |

| Slot width | IGP | Dual-slot |

| Power connectors | None | 8-pin EPS |

| Suggested PSU | Not listed | 1100 W |

| Bus interface | IGP | PCIe 5.0 x16 |

| Display outputs | Portable Device Dependent | No outputs |

| DirectX support | 12 Ultimate (12_2) | Not listed |

| OpenGL support | 4.6 | Not listed |

| Vulkan support | 1.4 | Not listed |

| Length | Not listed | 268 mm (10.6 inches) |

| Height | Not listed | 111 mm (4.4 inches) |

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

| Production status | Active | Active |