Intel Arc Pro B370 vs NVIDIA H200 NVL Comparison

Head-to-Head Benchmarks

The benchmark comparison between these two parts is stark because the recorded data is asymmetrical. The Intel Arc Pro B370 has no individual benchmark entries in the database, meaning its measured performance pool is empty. The NVIDIA H200 NVL, by contrast, holds a single recorded benchmark result: a Geekbench OpenCL score of 334,891. This places the H200 NVL at the 100th percentile of all GPUs in the database, a perfect placement. The Intel Arc Pro B370 sits at the 50th percentile, but that percentile is derived from its specification standing rather than from direct measured workloads.

The nearest rivals for the H200 NVL further contextualize its score. The NVIDIA B200 posts an average score of 345,482, which is 3.1% higher than the H200 NVL. The NVIDIA B300 SXM6 AC reaches 369,831, a 9.4% advantage. On the other side, the AMD Instinct MI300X scores 317,994, meaning the H200 NVL is 5.3% ahead of it. The NVIDIA L40S scores 295,763, and the H200 NVL leads that part by 13.2%. These deltas show the H200 NVL operating in the upper tier of accelerator-class hardware, trading blows with the newest B-series parts while comfortably outpacing the previous generation and AMD's flagship Instinct.

Because the head-to-head benchmark array is empty, there are no direct score comparisons between the Arc Pro B370 and the H200 NVL. The database records no wins for either part in direct competition. The gap in recorded test data is itself informative: the H200 NVL is a server accelerator with a validated OpenCL workload, while the Arc Pro B370 is an integrated graphics processor with no benchmark submissions on file. The raw compute figures confirm the scale of the difference. The H200 NVL delivers 60.32 TFLOPS of FP32 performance, while the Arc Pro B370 delivers 6.144 TFLOPS. That is a factor of roughly 9.8 based on the recorded numbers. In FP16, the H200 NVL posts 120.6 TFLOPS (2:1) against the Arc Pro B370's 12.29 TFLOPS (2:1), again a near tenfold difference.

Architecture Differences

The two chips come from entirely different design philosophies and manufacturing ecosystems. The Intel Arc Pro B370 uses the Panther Lake chip, built on Intel's Xe3-LPG architecture, and belongs to the Arc Graphics-WM (Panther Lake) generation. Its process node is 3 nm, fabricated by Intel. The NVIDIA H200 NVL uses the GH100 chip, built on the Hopper architecture, and belongs to the Server Hopper (Hxx) generation. Its process node is 5 nm, fabricated by TSMC. Intel's process advantage in node size is real, but the H200 NVL's transistor budget is enormous: the die size is 814 mm² with 80,000 million transistors, giving a transistor density of 98.3 million per mm². The Intel part's transistor count and die size are listed as unknown in the database.

The functional block counts reflect their roles. The Arc Pro B370 has 1280 shading units, 40 texture mapping units, 20 ROPs, and 10 ray tracing cores. It has no tensor core entry. The H200 NVL has 16,896 shading units, 528 TMUs, 24 ROPs, and 528 tensor cores. The H200 NVL carries no ray tracing core count. The pixel rates are close on paper: the Arc Pro B370 posts 48.00 GPixel/s, while the H200 NVL posts 42.84 GPixel/s. That is a narrow advantage for the Intel part in raw pixel throughput, though the H200 NVL's texture rate of 942.5 GTexel/s dwarfs the Arc Pro B370's 96.00 GTexel/s. The H200 NVL's tensor core count of 528 highlights its compute-oriented design; the Arc Pro B370 has no tensor core figure recorded at all.

Memory architecture is another fundamental split. The Arc Pro B370 uses system shared memory, with system shared type, bus width, and system dependent bandwidth. The H200 NVL uses 141 GB of HBM3e memory on a 6144-bit bus, delivering 4.89 TB/s of bandwidth. Clock behavior also diverges. The Intel part runs a 300 MHz base clock and a 2400 MHz boost clock, while the H200 NVL runs a 1365 MHz base and 1785 MHz boost. The H200 NVL memory clock is 1593 MHz, translating to 6.4 Gbps effective. The Arc Pro B370's memory clock is listed as system shared, meaning it is not a fixed figure.

Power and physical design are opposites. The Arc Pro B370 has a 25 W TDP, is an integrated graphics processor (IGP) with no power connectors, and uses the IGP bus interface. The H200 NVL has a 600 W TDP, is a dual-slot card, requires an 8-pin EPS power connector, and the database suggests a 1000 W power supply. The H200 NVL is 267 mm long and 111 mm tall. The Arc Pro B370 has no recorded dimensions. API support also differs meaningfully. The Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The H200 NVL lists N/A for DirectX, OpenGL, and Vulkan, which aligns with its role as a compute accelerator without display outputs. The H200 NVL has no display outputs, while the Arc Pro B370's display outputs are portable device dependent.

The Verdict

The recorded data points to two separate products for two separate purposes. The NVIDIA H200 NVL is a server-class accelerator with a 600 W TDP, 141 GB of HBM3e, 4.89 TB/s of memory bandwidth, and a validated Geekbench OpenCL score of 334,891 that places it at the 100th percentile. It sits within 3.1% of the NVIDIA B200 and 9.4% of the B300 SXM6 AC, while leading the AMD Instinct MI300X by 5.3% and the NVIDIA L40S by 13.2%. The Intel Arc Pro B370 is an integrated GPU with a 25 W TDP, system shared memory, and no recorded benchmark scores. It posts a 50th percentile standing based on its specification profile.

There is no scenario in the data where the Arc Pro B370 competes with the H200 NVL on compute throughput. The H200 NVL has roughly 9.8 times the FP32 throughput and roughly 9.8 times the FP16 throughput. The Arc Pro B370's advantages are limited to pixel rate, where it leads 48.00 GPixel/s versus 42.84 GPixel/s, and its lower power envelope of 25 W. For a builder assembling a compute server, the H200 NVL is the only viable option among these two based on the recorded figures. For a portable device requiring integrated graphics with modern API support, the Arc Pro B370 is the only option that fits, since the H200 NVL has no display outputs and no graphics API support.

FAQ

Q: Which GPU has a higher benchmark score in the database?

A: The NVIDIA H200 NVL has a Geekbench OpenCL score of 334,891. The Intel Arc Pro B370 has no recorded benchmark scores in the database.

Q: How does the H200 NVL compare to its nearest rivals?

A: The NVIDIA B200 scores 345,482, which is 3.1% higher. The NVIDIA B300 SXM6 AC scores 369,831, which is 9.4% higher. The AMD Instinct MI300X scores 317,994, which is 5.3% lower. The NVIDIA L40S scores 295,763, which is 13.2% lower.

Q: What are the memory capacities of these two parts?

A: The NVIDIA H200 NVL has 141 GB of HBM3e memory on a 6144-bit bus with 4.89 TB/s of bandwidth. The Intel Arc Pro B370 uses system shared memory with system dependent bandwidth.

Q: Which part supports DirectX 12 Ultimate?

A: The Intel Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA H200 NVL lists N/A for DirectX, OpenGL, and Vulkan.

Q: What is the power consumption difference?

A: The Intel Arc Pro B370 has a 25 W TDP and uses no power connectors. The NVIDIA H200 NVL has a 600 W TDP, uses an 8-pin EPS power connector, and the database suggests a 1000 W power supply.

Q: Do both parts have tensor cores?

A: The NVIDIA H200 NVL has 528 tensor cores. The Intel Arc Pro B370 has no tensor core figure recorded in the database.

Where Each One Wins

The Intel Arc Pro B370 wins in pixel throughput. Its pixel rate of 48.00 GPixel/s edges out the H200 NVL's 42.84 GPixel/s. It also wins decisively on power draw, with a 25 W TDP against the H200 NVL's 600 W. The Arc Pro B370 is the only one of the two with graphics API support, covering DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. It is an integrated processor with no power connectors and no slot width, making it suitable for compact portable devices. Its display outputs are portable device dependent, which means it can drive screens in the right host system, something the H200 NVL cannot do at all.

The NVIDIA H200 NVL wins in every compute-heavy metric. Its FP32 throughput of 60.32 TFLOPS is nearly ten times the Arc Pro B370's 6.144 TFLOPS. Its FP16 throughput of 120.6 TFLOPS (2:1) dwarfs the 12.29 TFLOPS (2:1) of the Intel part. The H200 NVL has 141 GB of dedicated HBM3e memory versus system shared memory, and its 4.89 TB/s memory bandwidth is a fixed specification rather than system dependent. The H200 NVL has 528 tensor cores, 16,896 shading units, and 528 TMUs, all far above the Arc Pro B370's counts. Its texture rate of 942.5 GTexel/s is nearly ten times the Arc Pro B370's 96.00 GTexel/s. The H200 NVL also has the recorded benchmark evidence: its OpenCL score of 334,891 places it at the 100th percentile, while the Arc Pro B370 has no measured score.

Specification Differences

The two parts differ across nearly every recorded specification field. The process node is 3 nm for the Intel Arc Pro B370 versus 5 nm for the NVIDIA H200 NVL. The foundry is Intel for the former and TSMC for the latter. Transistor count is unknown for the Arc Pro B370 and 80,000 million for the H200 NVL. Die size is unknown for the Arc Pro B370 and 814 mm² for the H200 NVL. Transistor density is not recorded for the Arc Pro B370 and is 98.3 million per mm² for the H200 NVL.

Clock speeds differ substantially. The Arc Pro B370 runs a 300 MHz base and 2400 MHz boost. The H200 NVL runs a 1365 MHz base and 1785 MHz boost. Memory clock is system shared for the Arc Pro B370 and 1593 MHz (6.4 Gbps effective) for the H200 NVL. Memory size, type, bus width, and bandwidth all differ, with the Arc Pro B370 using system shared values and the H200 NVL using 141 GB HBM3e, 6144 bit, and 4.89 TB/s.

Compute unit counts diverge sharply. The Arc Pro B370 has 1280 shading units, 40 TMUs, 20 ROPs, 10 ray tracing cores, and no tensor cores. The H200 NVL has 16,896 shading units, 528 TMUs, 24 ROPs, no ray tracing cores, and 528 tensor cores. Pixel rate favors the Arc Pro B370 at 48.00 GPixel/s versus 42.84 GPixel/s. Texture rate favors the H200 NVL at 942.5 GTexel/s versus 96.00 GTexel/s. FP32 and FP16 both favor the H200 NVL by a wide margin.

Power and physical specifications are opposites. The Arc Pro B370 has a 25 W TDP, IGP slot width, no power connectors, and an IGP bus interface. The H200 NVL has a 600 W TDP, dual-slot width, an 8-pin EPS power connector, a suggested 1000 W PSU, and a PCIe 5.0 x16 bus interface. The H200 NVL is 267 mm long and 111 mm tall; the Arc Pro B370 has no recorded dimensions. Display outputs are portable device dependent on the Arc Pro B370 and absent on the H200 NVL. API support is present on the Arc Pro B370 and listed as N/A on the H200 NVL. Release dates differ as well: the Arc Pro B370 released on January 26, 2026, while the H200 NVL released on November 17, 2024. The Arc Pro B370's predecessor is HD Graphics-WM, and the H200 NVL's predecessor is Server Ada with a successor of Server Blackwell. Neither part has a launch MSRP recorded in the database.