Intel Arc Pro B370 vs NVIDIA N1X 48SM Comparison

Head-to-Head Benchmarks

The database contains no recorded benchmark scores for either the Intel Arc Pro B370 or the NVIDIA N1X 48SM. Both entries show an average benchmark score of zero and no head-to-head benchmark comparisons are listed. This absence of measured performance data means the analysis must rely entirely on architectural specifications and theoretical throughput figures derived from the recorded specifications.

The Intel Arc Pro B370 delivers a peak FP32 throughput of 6.144 TFLOPS, while the NVIDIA N1X 48SM reaches 28.83 TFLOPS in FP32. That represents a 4.69x difference in raw single-precision compute. In FP16 operations, the Intel part achieves 12.29 TFLOPS using a 2:1 ratio, while the NVIDIA part sustains 28.83 TFLOPS at a 1:1 ratio, making the NVIDIA GPU 2.35x faster in half-precision workloads. These are the only direct numerical comparisons possible from the available data.

The pixel rate tells a similar story. The Arc Pro B370 renders 48.00 GPixel/s, while the N1X 48SM reaches 112.6 GPixel/s, a 2.35x advantage. Texture rate is even more lopsided: 96.00 GTexel/s versus 900.9 GTexel/s, a 9.38x gap. These figures reflect the vast difference in execution resources, with the NVIDIA part using 6144 shading units against 1280 for the Intel part.

The recorded data shows no wins for either product in the head-to-head section, as the field is empty. The only measurable comparison comes from the specification-driven throughput calculations, which consistently favor the NVIDIA part across every metric where numbers exist.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The Intel Arc Pro B370 uses the Panther Lake chip with Xe3-LPG architecture, belonging to the Arc Graphics-WM generation. The NVIDIA N1X 48SM uses the GB20B chip with Blackwell 2.0 architecture, part of the Blackwell IGP generation.

Process technology diverges sharply. Intel fabricates the Arc Pro B370 on a 3 nm node at its own foundry. NVIDIA uses a 5 nm process at TSMC. The die size is recorded only for the NVIDIA part at 382 mm²; the Intel die size is unknown. Transistor counts are unknown for both.

Memory architecture could not be more different. The Intel GPU uses system shared memory, with size, type, bus width, and bandwidth all listed as system dependent. The NVIDIA part has dedicated 128 GB of LPDDR5X memory on a 256 bit bus, delivering 273.2 GB/s of bandwidth. The NVIDIA memory clock is recorded as 1067 MHz with 8.5 Gbps effective data rate.

The compute resource allocation reveals the NVIDIA part's scale. It packs 6144 shading units, 384 texture mapping units, 48 ROPs, 48 ray tracing cores, and 192 tensor cores. The Intel part has 1280 shading units, 40 TMUs, 20 ROPs, and 10 ray tracing cores, with no tensor core count recorded.

Both parts use integrated graphics packaging with IGP slot width and no power connectors. The Intel GPU has a 25 W TDP, while the NVIDIA TDP is unknown. The Intel part uses an IGP bus interface, while the NVIDIA part connects via PCIe 5.0 x16.

API support shows a stark difference. The Intel Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA N1X 48SM lists N/A for DirectX, OpenGL, and Vulkan, suggesting it may target a different software ecosystem or rely on proprietary APIs.

Display outputs also differ: the Intel part is portable device dependent, while the NVIDIA part has a single HDMI output.

Where Each One Wins

The recorded data establishes the NVIDIA N1X 48SM as the dominant performer in every quantified metric. Its FP32 throughput of 28.83 TFLOPS exceeds the Intel part's 6.144 TFLOPS by more than fourfold. For FP16 workloads, the NVIDIA part's 28.83 TFLOPS at 1:1 ratio doubles the Intel part's 12.29 TFLOPS at 2:1, and critically, the NVIDIA part does not sacrifice half the throughput to achieve FP16, meaning it maintains full rate.

Rasterization throughput follows the same pattern. The NVIDIA part's pixel rate of 112.6 GPixel/s is 2.35x higher than the Intel part's 48.00 GPixel/s. Texture rate shows the largest proportional gap, with 900.9 GTexel/s versus 96.00 GTexel/s, a 9.38x difference. This suggests the NVIDIA part is heavily optimized for texture-heavy workloads.

Memory bandwidth gives the NVIDIA part another decisive advantage. Its dedicated 273.2 GB/s of bandwidth from 128 GB of LPDDR5X on a 256 bit bus stands in contrast to the Intel part's system shared memory with bandwidth dependent on the host platform. For workloads that scale with memory bandwidth, such as large dataset processing or high-resolution texture streaming, the NVIDIA part has a clear edge.

The NVIDIA part also holds advantages in ray tracing resources with 48 RT cores against 10, and in tensor operations with 192 tensor cores versus none recorded for the Intel part. The Intel part does have one significant advantage: its 25 W TDP. The NVIDIA TDP is unknown, but the Intel part's power draw is explicitly recorded as 25 W, making it a low-power integrated solution.

The Intel part also wins on API compatibility. Its support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 means it can run standard graphics APIs, while the NVIDIA part lists N/A for all of these. For any workload requiring these standard APIs, the Intel part is the only option between the two.

The Verdict

The data divides these two products into entirely separate categories. The NVIDIA N1X 48SM delivers 4.69x the FP32 throughput, 2.35x the pixel rate, and 9.38x the texture rate compared to the Intel Arc Pro B370. It also provides 128 GB of dedicated memory with 273.2 GB/s bandwidth, 48 RT cores, and 192 tensor cores. For any compute-intensive or graphics-intensive workload where these resources matter, the NVIDIA part is the clear choice based on the recorded specifications.

The Intel Arc Pro B370 offers a different value proposition. Its 25 W TDP makes it suitable for power-constrained environments. Its support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 gives it broad software compatibility that the NVIDIA part lacks entirely. The Intel part also uses system shared memory, which could simplify system design by eliminating dedicated memory allocation.

The choice between these two depends entirely on the use case. The NVIDIA part wins on raw performance, memory capacity, bandwidth, and specialized compute resources. The Intel part wins on power efficiency, API compatibility, and integrated simplicity. Neither product has recorded benchmark scores, so these conclusions derive strictly from the architectural specifications.

The release dates differ, with the Intel part released on January 26, 2026, and the NVIDIA part on May 31, 2026. Both are listed as active production status.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The NVIDIA N1X 48SM delivers 28.83 TFLOPS in FP32, compared to 6.144 TFLOPS for the Intel Arc Pro B370, making the NVIDIA part 4.69x faster.

Q: What is the memory configuration of each GPU?

A: The Intel Arc Pro B370 uses system shared memory with type, bus width, and bandwidth all dependent on the host system. The NVIDIA N1X 48SM has 128 GB of LPDDR5X memory on a 256 bit bus with 273.2 GB/s bandwidth.

Q: Does the NVIDIA N1X 48SM support DirectX?

A: The database records DirectX as N/A for the NVIDIA N1X 48SM. The Intel Arc Pro B370 supports DirectX 12 Ultimate (12_2).

Q: What is the process node for each GPU?

A: The Intel Arc Pro B370 is fabricated on a 3 nm process at Intel's foundry. The NVIDIA N1X 48SM uses a 5 nm process at TSMC.

Q: How many ray tracing cores does each GPU have?

A: The NVIDIA N1X 48SM has 48 ray tracing cores, while the Intel Arc Pro B370 has 10 ray tracing cores.

Q: What is the power consumption of the Intel Arc Pro B370?

A: The Intel Arc Pro B370 has a recorded TDP of 25 W. The TDP of the NVIDIA N1X 48SM is unknown.

Specification Differences

| Specification | Intel Arc Pro B370 | NVIDIA N1X 48SM |

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

| Chip | Panther Lake | GB20B |

| Architecture | Xe3-LPG | Blackwell 2.0 |

| Generation | Arc Graphics-WM (Panther Lake) | Blackwell IGP (N1x) |

| Process Node | 3 nm | 5 nm |

| Foundry | Intel | TSMC |

| Die Size | unknown | 382 mm² |

| Base Clock | 300 MHz | 741 MHz |

| Boost Clock | 2400 MHz | 2346 MHz |

| Memory Clock | System Shared | 1067 MHz 8.5 Gbps effective |

| Memory Size | System Shared | 128 GB |

| Memory Type | System Shared | LPDDR5X |

| Memory Bus Width | System Shared | 256 bit |

| Memory Bandwidth | System Dependent | 273.2 GB/s |

| Shading Units | 1280 | 6144 |

| TMUs | 40 | 384 |

| ROPs | 20 | 48 |

| RT Cores | 10 | 48 |

| Tensor Cores | null | 192 |

| Pixel Rate | 48.00 GPixel/s | 112.6 GPixel/s |

| Texture Rate | 96.00 GTexel/s | 900.9 GTexel/s |

| FP32 Performance | 6.144 TFLOPS | 28.83 TFLOPS |

| FP16 Performance | 12.29 TFLOPS (2:1) | 28.83 TFLOPS (1:1) |

| TDP | 25 W | unknown |

| Bus Interface | IGP | PCIe 5.0 x16 |

| Display Outputs | Portable Device Dependent | 1x HDMI |

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

| OpenGL Support | 4.6 | N/A |

| Vulkan Support | 1.4 | N/A |

| Release Date | 2026-01-26 | 2026-05-31 |

| Predecessor | HD Graphics-WM | null |