AMD Instinct MI350P vs Intel Arc Pro B370 Comparison

Where Each One Wins

The recorded data shows no benchmark wins for either part. The AMD Instinct MI350P and Intel Arc Pro B370 both register zero wins in the head-to-head comparison, and both sit at the 50th percentile versus all GPUs in the database. With an average benchmark score of zero for each, the database does not yet contain performance measurements that separate them.

The absence of recorded wins does not mean the two are equivalent in capability. The AMD Instinct MI350P is built for compute acceleration in data center environments, while the Intel Arc Pro B370 is an integrated graphics processor designed for mobile or compact systems. Their intended workloads diverge sharply, even though the current benchmark data does not quantify that split.

The AMD part carries 8,192 shading units, 512 texture mapping units, and no raster operation units. Its pixel rate is listed as 0 MPixel/s and texture rate reaches 1,126.4 GTexel/s. The Intel part has 1,280 shading units, 40 texture mapping units, 20 raster operation units, and 10 ray tracing cores. Its pixel rate is 48.00 GPixel/s and texture rate is 96.00 GTexel/s.

These figures indicate opposite design philosophies. The AMD accelerator prioritizes raw compute throughput with enormous texture processing capacity, while the Intel integrated GPU provides conventional rendering pipelines with ray tracing support. The Intel part supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, whereas the AMD part lists N/A for all three APIs.

Architecture Differences

The AMD Instinct MI350P uses the MI350 128CU chip built on CDNA 4.0 architecture. The manufacturing process is 3 nm at TSMC, with 73,000 million transistors on a 1,190 mm² die. Transistor density measures 61.3M per mm². The Intel Arc Pro B370 uses the Panther Lake chip with Xe3-LPG architecture, also on a 3 nm process but at Intel's foundry. Transistor count and die size are listed as unknown in the database.

Memory configurations differ completely. The AMD part has 144 GB of HBM3e memory on an 8,192-bit bus, delivering 8.19 TB/s of bandwidth. Memory clock is 2000 MHz with 8 Gbps effective speed. The Intel part uses system shared memory with system dependent bandwidth and no dedicated memory clock. This reflects the integrated nature of the Intel GPU, which borrows memory from the host system rather than carrying its own pool.

Clock behavior also diverges. The AMD part runs at a 1000 MHz base clock and boosts to 2200 MHz. The Intel part starts at 300 MHz base and boosts to 2400 MHz, a higher ceiling but a much lower floor. The AMD accelerator draws 600 W through a single 16-pin connector with a suggested 1000 W power supply. The Intel integrated GPU consumes 25 W and requires no power connectors or suggested PSU.

Physical design separates them further. The AMD card is dual-slot, 267 mm long, 111 mm tall, and 40 mm wide, with no display outputs. The Intel part is an IGP with portable device dependent display outputs and no listed dimensions. Bus interface for AMD is PCIe 5.0 x16, while Intel uses IGP.

Compute throughput numbers reveal the scale gap. The AMD part delivers 36.04 TFLOPS FP32 and 36.04 TFLOPS FP16 at a 1:1 ratio. The Intel part delivers 6.144 TFLOPS FP32 and 12.29 TFLOPS FP16 at a 2:1 ratio. The AMD accelerator has nearly six times the FP32 throughput and roughly three times the FP16 throughput, though Intel's FP16 advantage over its own FP32 is larger in relative terms.

Head-to-Head Benchmarks

The head-to-head benchmark array is empty, so exact performance deltas between these two parts cannot be drawn from the database. However, the specification data provides a basis for comparing their respective strengths in different workload categories.

The AMD Instinct MI350P shows its largest advantage in memory bandwidth. With 8.19 TB/s available, it exceeds any integrated graphics solution by an enormous margin, since the Intel Arc Pro B370's bandwidth is system dependent and therefore limited by the host platform. For memory-bound compute tasks such as large matrix operations or data center inference workloads, the AMD part's HBM3e stack provides the bandwidth necessary to feed its 8,192 shading units.

Texture throughput favors AMD by a factor of roughly 11.7 based on the recorded figures. The AMD part processes 1,126.4 GTexel/s versus 96.00 GTexel/s for Intel. This suggests the AMD accelerator can sustain much higher texture fetch rates, which matters for certain compute kernels that rely on texture sampling as a data access pattern.

The Intel Arc Pro B370 counters in pixel throughput. Its 48.00 GPixel/s rasterization rate exceeds the AMD part's 0 MPixel/s, since the AMD accelerator has no ROPs and is not designed for display output or traditional rasterization. The Intel part also includes 10 ray tracing cores, a feature absent from the AMD specification.

FP16 compute shows a different ratio. The Intel part achieves 12.29 TFLOPS FP16, which is double its 6.144 TFLOPS FP32 figure. The AMD part achieves identical 36.04 TFLOPS for both FP32 and FP16. Applications that rely on reduced precision may find the Intel part's 2:1 ratio useful for certain neural network inference tasks, though the absolute throughput remains far below the AMD part.

Clock speed behavior matters for burst workloads. The Intel part boosts to 2400 MHz, which is 200 MHz higher than the AMD part's 2200 MHz boost. For short, latency-sensitive tasks that fit within the integrated GPU's thermal envelope, the higher boost clock could provide a per-core advantage. The AMD part's 1000 MHz base clock versus Intel's 300 MHz base clock indicates different sustained operating characteristics.

The Verdict

The data supports a clear separation of use cases. The AMD Instinct MI350P targets compute-heavy environments where 144 GB of HBM3e, 8.19 TB/s bandwidth, and 36.04 TFLOPS FP32 are the priority. Its 600 W power draw, dual-slot form factor, and absence of display outputs confirm it is a server accelerator, not a client graphics solution.

The Intel Arc Pro B370 targets integrated graphics scenarios where 25 W power consumption, ray tracing support, and DirectX 12 Ultimate compatibility matter. Its system shared memory and portable device dependent outputs place it in laptops or compact devices where discrete graphics is not an option.

The database currently shows no benchmark results for either part, so percentile rankings remain neutral at 50 for both. Buyers should consult the specification data to match workload requirements. Compute acceleration with massive memory capacity points to AMD. Conventional graphics rendering with low power consumption points to Intel.

The AMD part's release date is recorded as May 6, 2026, while the Intel part's release date is January 26, 2026. The Intel part is listed as active in production status, while the AMD part has no production status recorded. The AMD predecessor is Radeon Instinct, and the Intel predecessor is HD Graphics-WM. Neither part has a successor listed.

For organizations running AI training, scientific simulation, or large-scale data processing, the AMD accelerator's memory capacity and bandwidth are the determining factors. For client devices needing graphics output, media acceleration, or light compute with minimal power draw, the Intel integrated GPU is the appropriate choice.

FAQ

Q: Which GPU has higher FP32 compute throughput?

A: The AMD Instinct MI350P delivers 36.04 TFLOPS FP32, compared to 6.144 TFLOPS for the Intel Arc Pro B370. The AMD part leads by a factor of roughly 5.9.

Q: Does the Intel Arc Pro B370 support ray tracing?

A: Yes, the Intel part includes 10 ray tracing cores and supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The AMD Instinct MI350P lists no ray tracing cores and N/A for all three APIs.

Q: What memory configuration does each GPU use?

A: The AMD Instinct MI350P has 144 GB of HBM3e memory on an 8,192-bit bus with 8.19 TB/s bandwidth. The Intel Arc Pro B370 uses system shared memory with system dependent bandwidth.

Q: How much power does each GPU consume?

A: The AMD Instinct MI350P has a 600 W TDP and requires a 1x 16-pin power connector with a suggested 1000 W power supply. The Intel Arc Pro B370 has a 25 W TDP and requires no power connectors.

Q: What is the memory clock for each GPU?

A: The AMD part runs at 2000 MHz with 8 Gbps effective memory speed. The Intel part lists memory clock as system shared, meaning it depends on the host system's memory configuration.

Q: Which GPU has a higher boost clock?

A: The Intel Arc Pro B370 boosts to 2400 MHz, while the AMD Instinct MI350P boosts to 2200 MHz. The Intel part also has a lower base clock at 300 MHz versus the AMD part's 1000 MHz.

Specification Differences

| Specification | AMD Instinct MI350P | Intel Arc Pro B370 |

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

| Chip | MI350 128CU | Panther Lake |

| Architecture | CDNA 4.0 | Xe3-LPG |

| Process Node | 3 nm (TSMC) | 3 nm (Intel) |

| Transistors | 73,000 million | Unknown |

| Die Size | 1190 mm² | Unknown |

| Base Clock | 1000 MHz | 300 MHz |

| Boost Clock | 2200 MHz | 2400 MHz |

| Memory Size | 144 GB | System Shared |

| Memory Type | HBM3e | System Shared |

| Memory Bus Width | 8192 bit | System Shared |

| Memory Bandwidth | 8.19 TB/s | System Dependent |

| Shading Units | 8192 | 1280 |

| TMUs | 512 | 40 |

| ROPs | 0 | 20 |

| Ray Tracing Cores | None listed | 10 |

| Pixel Rate | 0 MPixel/s | 48.00 GPixel/s |

| Texture Rate | 1,126.4 GTexel/s | 96.00 GTexel/s |

| FP32 | 36.04 TFLOPS | 6.144 TFLOPS |

| FP16 | 36.04 TFLOPS (1:1) | 12.29 TFLOPS (2:1) |

| TDP | 600 W | 25 W |

| Slot Width | Dual-slot | IGP |

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

| Suggested PSU | 1000 W | Not listed |

| Bus Interface | PCIe 5.0 x16 | IGP |

| Display Outputs | No outputs | Portable Device Dependent |

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

| OpenGL | N/A | 4.6 |

| Vulkan | N/A | 1.4 |

| Dimensions | 267 mm x 111 mm x 40 mm | Not listed |

| Release Date | 2026-05-06 | 2026-01-26 |

| Predecessor | Radeon Instinct | HD Graphics-WM |

| Production Status | Not listed | Active |