AMD Instinct MI355X vs Intel Arc Pro B390 Comparison

Head-to-Head Benchmarks

The database contains no recorded benchmark scores for either the AMD Instinct MI355X or the Intel Arc Pro B390. Both products show an average benchmark score of 0, and the head-to-head benchmark table is empty. This is not unusual for products at opposite ends of the compute spectrum: the MI355X is a dedicated accelerator module with no display outputs, while the Arc Pro B390 is an integrated GPU designed for mobile and embedded platforms.

Without direct performance measurements, the nearest rival comparisons are also absent. Neither product has a percentile ranking above the 50th percentile mark, which indicates that neither has been tested against the broader GPU population in the database. The data cannot support any claim about which part is faster in real workloads, only what the specifications suggest.

What the specifications do show is a massive gap in raw compute resources. The MI355X delivers 78.64 TFLOPS of FP32 performance, while the Arc Pro B390 delivers 7.680 TFLOPS. That is an order of magnitude difference. In FP16, the MI355X again delivers 78.64 TFLOPS with a 1:1 ratio, whereas the Arc Pro B390 delivers 15.36 TFLOPS with a 2:1 ratio. The MI355X has 16384 shading units versus 1536 for the Arc Pro B390, and 1024 texture mapping units versus 48. These are not competing products in any practical sense.

The pixel rate comparison is inverted. The Arc Pro B390 has a pixel rate of 60.00 GPixel/s, while the MI355X is recorded at 0 MPixel/s. The MI355X has no ROPs, which is consistent with its role as a compute accelerator rather than a graphics renderer. The Arc Pro B390 has 24 ROPs and 12 ray tracing cores, making it a fully functional graphics processor.

Architecture Differences

The AMD Instinct MI355X uses the CDNA 4.0 architecture, built on a 3 nm process at TSMC. The chip is designated MI350 256CU, which indicates 256 compute units. The transistor count is 185,000 million, and the die size is 2380 mm². The transistor density works out to 77.7 million transistors per square millimeter. This is a massive, purpose-built accelerator die.

The Intel Arc Pro B390 uses the Xe3-LPG architecture, built on a 3 nm process at Intel's own foundry. The chip is designated Panther Lake, and the generation is listed as Arc Graphics-WM. Intel does not disclose transistor count or die size for this part. The architecture is labeled Xe3-LPG, which is a low-power graphics variant designed for integration.

The memory subsystems could not be more different. The MI355X has 288 GB of HBM3e memory on an 8192-bit bus, delivering 8.19 TB/s of bandwidth. The memory clock is 2000 MHz with 8 Gbps effective data rate. The Arc Pro B390 uses system shared memory, with a system dependent bandwidth. Its memory type, bus width, and size are all listed as system shared or system dependent. The MI355X has fixed, dedicated memory; the Arc Pro B390 relies on the host system's memory pool.

The MI355X has no display outputs and no API support for DirectX, OpenGL, or Vulkan. The Arc Pro B390 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI355X is a compute-only device with a PCIe 5.0 x16 bus interface. The Arc Pro B390 uses an IGP bus interface, meaning it is integrated into the processor package.

Power and physical form factor follow the same divergence. The MI355X has a TDP of 1400 W and requires a suggested PSU of 1800 W. It is an OAM module, measuring 102 mm in length and 165 mm in width. It has no power connectors because it receives power through the OAM socket. The Arc Pro B390 has a TDP of 80 W, is an IGP (integrated graphics processor), and has no power connectors or dimensions listed because it is soldered into a host chip.

The MI355X base clock is 1000 MHz with a boost of 2400 MHz. The Arc Pro B390 has a base clock of 300 MHz and a boost of 2500 MHz. The higher boost clock on the Intel part is notable, but it operates with far fewer execution resources.

Where Each One Wins

The AMD Instinct MI355X wins in every metric that matters for compute acceleration. It has 10.7 times the shading units, 21.3 times the texture mapping units, and 10.2 times the FP32 throughput. The FP16 throughput is 5.1 times higher on the AMD part. The memory bandwidth advantage is even more pronounced: 8.19 TB/s versus system dependent, which in practice will be a fraction of that figure. The 288 GB of dedicated HBM3e memory versus system shared memory means the MI355X can hold far larger datasets and model weights in fast memory.

The Intel Arc Pro B390 wins in the areas that matter for graphics output. It has 24 ROPs and a pixel rate of 60.00 GPixel/s, whereas the MI355X has zero ROPs and zero pixel throughput. The Arc Pro B390 supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, making it usable for gaming, rendering, and general display workloads. The MI355X supports none of these APIs and has no display outputs. The Arc Pro B390 also has 12 ray tracing cores, which provide hardware-accelerated ray tracing, a feature entirely absent from the MI355X.

The power envelope favors Intel decisively. The Arc Pro B390 has a TDP of 80 W, which is 17.5 times lower than the MI355X's 1400 W. The Arc Pro B390 requires no separate power connectors and no suggested PSU. It is designed to run within the thermal budget of a mobile or embedded processor. The MI355X needs an 1800 W PSU recommendation and an OAM module slot.

The release dates are also telling. The MI355X was released on June 11, 2025. The Arc Pro B390 has a release date of January 26, 2026. The Intel part is newer, and its production status is listed as active, while the MI355X has no production status listed.

FAQ

Q: Which GPU has more raw compute power?

A: The AMD Instinct MI355X delivers 78.64 TFLOPS in both FP32 and FP16, while the Intel Arc Pro B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16. The AMD part has 16384 shading units versus 1536 on the Intel part.

Q: Can the AMD Instinct MI355X output video to a display?

A: No. The MI355X has no display outputs and supports no graphics APIs. It is a compute accelerator only.

Q: Does the Intel Arc Pro B390 support ray tracing?

A: Yes. The Arc Pro B390 has 12 ray tracing cores and supports DirectX 12 Ultimate (12_2), which includes DirectX Raytracing.

Q: What type of memory does each GPU use?

A: The MI355X uses 288 GB of HBM3e memory on an 8192-bit bus with 8.19 TB/s bandwidth. The Arc Pro B390 uses system shared memory with system dependent bandwidth.

Q: How much power does each GPU require?

A: The MI355X has a TDP of 1400 W and a suggested PSU of 1800 W. The Arc Pro B390 has a TDP of 80 W and no suggested PSU requirement.

Q: Which GPU was released more recently?

A: The Intel Arc Pro B390 has a release date of January 26, 2026. The AMD Instinct MI355X has a release date of June 11, 2025.

The Verdict

The AMD Instinct MI355X is for compute workloads that need massive memory capacity and bandwidth. The 288 GB of HBM3e memory and 8.19 TB/s bandwidth are the defining characteristics. Any workload that fits within that memory pool and needs matrix or vector math at scale will favor the MI355X. Its 78.64 TFLOPS of FP32 throughput and 1024 texture mapping units indicate a device built for dense computation, not graphics.

The Intel Arc Pro B390 is for systems that need graphics output with modest compute capability. Its 60.00 GPixel/s pixel rate, 24 ROPs, and 12 ray tracing cores make it a functional graphics solution. The support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 means it can run standard graphics applications. The 80 W TDP and IGP form factor make it suitable for integration into a mobile or embedded processor package.

Neither product is a substitute for the other. The MI355X cannot render frames, and the Arc Pro B390 cannot approach the memory bandwidth or compute density of the AMD part. The choice between them is determined by the workload: compute acceleration versus graphics output. The data shows no overlap in their intended functions.

Specification Differences

| Specification | AMD Instinct MI355X | Intel Arc Pro B390 |

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

| Architecture | CDNA 4.0 | Xe3-LPG |

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

| Transistors | 185,000 million | unknown |

| Die Size | 2380 mm² | unknown |

| Base Clock | 1000 MHz | 300 MHz |

| Boost Clock | 2400 MHz | 2500 MHz |

| Memory Size | 288 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 | 16384 | 1536 |

| TMUs | 1024 | 48 |

| ROPs | 0 | 24 |

| Ray Tracing Cores | None | 12 |

| Pixel Rate | 0 MPixel/s | 60.00 GPixel/s |

| Texture Rate | 2,457.6 GTexel/s | 120.0 GTexel/s |

| FP32 Performance | 78.64 TFLOPS | 7.680 TFLOPS |

| FP16 Performance | 78.64 TFLOPS (1:1) | 15.36 TFLOPS (2:1) |

| TDP | 1400 W | 80 W |

| Slot Width | OAM Module | IGP |

| Bus Interface | PCIe 5.0 x16 | IGP |

| Display Outputs | No outputs | Portable Device Dependent |

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

| OpenGL Support | N/A | 4.6 |

| Vulkan Support | N/A | 1.4 |

| Suggested PSU | 1800 W | None |

| Release Date | 2025-06-11 | 2026-01-26 |

| Production Status | Not listed | Active |