Intel Arc B370 vs NVIDIA H20 NVL16 Comparison

FAQ

Q: How does the Intel Arc B370 compare to the NVIDIA H20 NVL16 in raw graphics performance?

A: The database contains only one benchmark result for the Intel Arc B370: a 3DMark Steel Nomad DX12 score of 1184. The NVIDIA H20 NVL16 has no recorded benchmark entries in the database, so no direct head-to-head graphics performance comparison can be drawn from the measured data. The Arc B370 sits at the 5th percentile among all GPUs, while the H20 NVL16 is at the 50th percentile, a figure derived from its broader product positioning rather than a specific test.

Q: What are the nearest performance rivals to the Intel Arc B370?

A: The nearest rivals to the Arc B370, based on average benchmark scores, are the ATI Mobility Radeon HD 5570 (avg score 1186, 0.2% higher), the ATI Radeon HD 5770 (avg score 1190, 0.5% higher), the AMD Radeon HD 7650M (avg score 1192, 0.7% higher), and the AMD FirePro M2000 (avg score 1168, 1.4% lower). These deltas place the Arc B370 in a tightly packed performance cluster where the largest recorded gap is just 1.4%.

Q: What is the memory configuration of each product?

A: The Arc B370 uses system-shared memory, with its memory size, type, bus width, and bandwidth all listed as "System Shared" or "System Dependent". The H20 NVL16 uses 96 GB of HBM3 memory on a 6144-bit bus, delivering 4.03 TB/s of bandwidth with a memory clock of 1313 MHz (5.3 Gbps effective).

Q: What are the thermal and power specifications?

A: The Arc B370 has a 25 W TDP, uses no power connectors, and is an integrated graphics processor (IGP) with a slot width of "IGP". The H20 NVL16 has a 400 W TDP, is an SXM Module, and lists a suggested PSU of 800 W.

Q: Which APIs does each product support?

A: The Arc B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The H20 NVL16 lists "N/A" for DirectX, OpenGL, and Vulkan, which is consistent with its server-focused positioning and lack of display outputs.

Q: What are the release dates and production statuses?

A: The Intel Arc B370 was released on 2026-01-26 and is marked as Active. The NVIDIA H20 NVL16 was released on 2025-09-01 and is also marked as Active. The H20 NVL16 lists its predecessor as Server Ada and its successor as Server Blackwell.

The Verdict

The recorded data presents two products with fundamentally different purposes. The Intel Arc B370 is a 25 W integrated GPU built on Intel's 3 nm process with a 5th percentile standing among all GPUs, scoring 1184 in the sole recorded 3DMark Steel Nomad DX12 test. Its closest measured competitor, the ATI Radeon HD 5770, sits only 0.5% higher, indicating that the Arc B370's graphics output aligns with a specific low-power, portable-device segment.

The NVIDIA H20 NVL16 belongs to a different category entirely. It is a 400 W SXM module for servers, with no display outputs, no graphics API support, and no recorded graphics benchmarks. Its 50th percentile placement among all GPUs is a database-wide ranking that does not derive from the same test suite. The H20 NVL16 is defined by its 96 GB HBM3 memory, 4.03 TB/s bandwidth, 9984 shading units, and 312 tensor cores, all serving compute-oriented workloads.

The data indicates that a user selecting between these two would be choosing between an integrated graphics solution for portable systems and a server accelerator with no graphics output capability. The benchmark results do not place them in the same performance envelope; the Arc B370's 6.144 TFLOPS FP32 throughput and 48.00 GPixel/s pixel rate contrast with the H20 NVL16's 39.54 TFLOPS FP32 throughput and 47.52 GPixel/s pixel rate. The Arc B370 delivers slightly higher pixel throughput despite operating at a fraction of the power, while the H20 NVL16 dominates in texture rate, memory bandwidth, and raw compute.

Head-to-Head Benchmarks

The database contains no direct head-to-head benchmark entries between the Intel Arc B370 and the NVIDIA H20 NVL16. The only recorded performance data for either product is the Arc B370's 3DMark Steel Nomad DX12 score of 1184. The H20 NVL16 has no benchmark entries in the database, making a direct numerical comparison of graphics workloads impossible from the measured data.

The Arc B370's 1184 score can be evaluated against its nearest rivals instead. The ATI Mobility Radeon HD 5570 scores 1186, the ATI Radeon HD 5770 scores 1190, and the AMD Radeon HD 7650M scores 1192. The AMD FirePro M2000 trails at 1168. These results indicate that the Arc B370 performs within 0.7% of three older discrete GPUs and 1.4% ahead of the FirePro M2000, a narrow spread that suggests its integrated design achieves performance parity with a specific class of legacy mobile and entry-level desktop parts.

The H20 NVL16's compute capabilities are documented through its specification data rather than benchmark scores. Its FP32 throughput of 39.54 TFLOPS is roughly 6.4 times the Arc B370's 6.144 TFLOPS. Its texture rate of 617.8 GTexel/s is more than 6.4 times the Arc B370's 96.00 GTexel/s. The pixel rates are close: 47.52 GPixel/s for the H20 NVL16 versus 48.00 GPixel/s for the Arc B370, a 0.48 GPixel/s difference in favor of the Intel part. These figures show that in rasterization-bound pixel output, the two are nearly identical, while in shading, texturing, and memory throughput, the NVIDIA part is substantially ahead.

Specification Differences

The Intel Arc B370 and NVIDIA H20 NVL16 differ across nearly every recorded specification field. The Arc B370 uses the Panther Lake chip with Xe3-LPG architecture, built on Intel's 3 nm process. The H20 NVL16 uses the GH100 chip with Hopper architecture, built on TSMC's 5 nm process. The H20 NVL16's die size is 814 mm² with 80,000 million transistors and a transistor density of 98.3M per mm²; the Arc B370's transistor count and die size are listed as unknown.

Clock speeds differ sharply. The Arc B370 has a 300 MHz base clock and a 2400 MHz boost clock. The H20 NVL16 has a 1830 MHz base clock and a 1980 MHz boost clock. Despite a lower boost frequency, the H20 NVL16 achieves far higher throughput due to its larger execution resource pool.

The Arc B370 has 1280 shading units, 40 TMUs, 20 ROPs, and 10 RT cores, with no tensor core count listed. The H20 NVL16 has 9984 shading units, 312 TMUs, 24 ROPs, no RT core count listed, and 312 tensor cores. The FP32 and FP16 figures scale accordingly: the Arc B370 delivers 6.144 TFLOPS FP32 and 12.29 TFLOPS FP16 (2:1), while the H20 NVL16 delivers 39.54 TFLOPS FP32 and 79.07 TFLOPS FP16 (2:1).

Memory configurations are entirely different. The Arc B370 uses system-shared memory with system-dependent bandwidth. The H20 NVL16 uses 96 GB of HBM3 with a 6144-bit bus and 4.03 TB/s bandwidth. Power requirements span from 25 W for the Arc B370 to 400 W for the H20 NVL16, with the latter recommending an 800 W PSU. The Arc B370 is an IGP with no power connectors and portable-device-dependent display outputs; the H20 NVL16 is an SXM module with no display outputs and a PCIe 5.0 x16 bus interface.

Architecture Differences

The two products represent divergent architectural lineages. The Arc B370 is built on Intel's Xe3-LPG architecture, part of the Arc Graphics-M generation for Panther Lake, fabricated on Intel's 3 nm process. The H20 NVL16 uses NVIDIA's Hopper architecture, fabricated by TSMC on a 5 nm process, and belongs to the Server Hopper generation.

The Arc B370 integrates graphics, rendering, and ray tracing into a 25 W package. It includes 10 RT cores for hardware-accelerated ray tracing and supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, making it a full-featured graphics solution for portable devices. Its memory subsystem relies entirely on the host system, with capacity, bus width, and bandwidth all dependent on the platform.

The H20 NVL16 is a server accelerator built around a different priority set. It has 312 tensor cores and no recorded RT core count, and its API support is listed as N/A for graphics APIs. Its 96 GB HBM3 stack with 6144-bit bus width and 4.03 TB/s bandwidth serves large-scale compute workloads. The 814 mm² die with 80,000 million transistors reflects a design focused on parallel throughput rather than graphics output, consistent with its lack of display connectors.

The transistor density figures highlight the process difference: 98.3M transistors per mm² for the H20 NVL16 on TSMC's 5 nm node. The Arc B370's density is not recorded, but its 3 nm process node from Intel indicates a newer manufacturing generation. The H20 NVL16's predecessor is listed as Server Ada and its successor as Server Blackwell, placing it within a defined server product progression. The Arc B370 lists no predecessor or successor, indicating it is a standalone entry in the Arc Graphics-M line.

Where Each One Wins

The Intel Arc B370 wins in scenarios defined by low power and integrated graphics. Its 25 W TDP, lack of power connectors, and IGP form factor make it suitable for portable devices where discrete power delivery is unavailable. Its pixel rate of 48.00 GPixel/s edges out the H20 NVL16's 47.52 GPixel/s, and its API support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 means it can run standard graphics workloads. Its 3DMark Steel Nomad DX12 score of 1184 places it among a cluster of legacy GPUs, with a 1.4% lead over the AMD FirePro M2000 and a 0.7% deficit against the AMD Radeon HD 7650M.

The NVIDIA H20 NVL16 wins in compute-heavy, server-oriented scenarios. Its 39.54 TFLOPS FP32 throughput, 79.07 TFLOPS FP16 throughput, 617.8 GTexel/s texture rate, and 4.03 TB/s memory bandwidth are decisive advantages for workloads that scale across 9984 shading units and 312 tensor cores. The 96 GB HBM3 memory capacity and 6144-bit bus width support large data sets. Its 400 W TDP and 800 W suggested PSU reflect a design that assumes dedicated server power delivery. Its 50th percentile standing among all GPUs, while not derived from a specific benchmark in the database, places it in the middle of the overall GPU distribution, a far higher rank than the Arc B370's 5th percentile.

The data does not record any benchmark where both products participated, so the win split is defined by specification analysis rather than measured head-to-head results. The Arc B370's advantage lies in pixel throughput, API compatibility, and power efficiency. The H20 NVL16's advantage lies in compute throughput, memory capacity, bandwidth, and tensor performance. Each product wins in the domain its design targets.