AMD Radeon Instinct MI308X vs Intel Arc Pro B370 Comparison

Head-to-Head Benchmarks

The recorded database contains no benchmark scores for either the AMD Radeon Instinct MI308X or the Intel Arc Pro B370. Both entries show an average benchmark score of zero, and the head-to-head benchmark table is empty. Consequently, there are no measured wins, losses, or performance deltas to report between these two accelerators. The percentile ranking for both parts sits at 50, which in this dataset reflects the absence of recorded test data rather than a meaningful performance comparison.

Without benchmark results, the only quantitative basis for comparison is the architectural specification data. That data shows an extreme disparity in compute resources. The AMD part delivers 81.72 TFLOPS of FP32 throughput, while the Intel part delivers 6.144 TFLOPS. That is a 13.3x difference in raw single-precision compute. In FP16, the AMD part reaches 653.7 TFLOPS with an 8:1 ratio, while the Intel part reaches 12.29 TFLOPS with a 2:1 ratio. The AMD part is 53.2x ahead in FP16 throughput.

Texture rate tells a similar story. The AMD Radeon Instinct MI308X produces 2,553.6 GTexel/s, while the Intel Arc Pro B370 produces 96.00 GTexel/s. The AMD part is 26.6x faster in texture fill. Pixel rate is a different matter: the AMD part lists 0 MPixel/s because it has zero ROPs, while the Intel part lists 48.00 GPixel/s with 20 ROPs. This is a case where the Intel part is technically ahead, but the AMD part is not designed for rasterization at all, so the comparison is structurally meaningless.

Memory capacity and bandwidth further separate the two. The AMD Radeon Instinct MI308X carries 192 GB of HBM3 across an 8192-bit bus, yielding 10.3 TB/s of bandwidth. The Intel Arc Pro B370 uses system shared memory with system dependent bandwidth, so no direct capacity or bandwidth figure can be compared. The AMD memory clock is 2525 MHz, translating to 10.1 Gbps effective. The Intel memory clock is listed as system shared.

The shading unit count reinforces the gap. AMD has 19,456 shading units, while Intel has 1,280. The AMD part also has 1,216 TMUs versus Intel's 40. The AMD part has no RT cores listed, while Intel has 10 RT cores. The AMD part has no tensor core count listed, and neither does Intel.

Clock speeds do not favor the larger part in every dimension. The AMD base clock is 1000 MHz with a boost of 2100 MHz. The Intel base clock is 300 MHz with a boost of 2400 MHz. Intel boosts 300 MHz higher, but that does little to close the raw throughput gap given the massive difference in execution resources.

Power consumption and physical format differ dramatically. The AMD Radeon Instinct MI308X has a TDP of 750 W, uses an OAM module slot, has no power connectors listed, and recommends a 1150 W PSU. The Intel Arc Pro B370 has a TDP of 25 W, is an IGP, has no power connectors, and lists no suggested PSU. The Intel part is an integrated GPU inside a Panther Lake processor, while the AMD part is a standalone accelerator module.

FAQ

Q: Which GPU has more raw FP32 compute?

A: The AMD Radeon Instinct MI308X delivers 81.72 TFLOPS of FP32, versus 6.144 TFLOPS for the Intel Arc Pro B370. That is a 13.3x advantage for the AMD part.

Q: Does the Intel Arc Pro B370 have a higher boost clock?

A: Yes. The Intel part boosts to 2400 MHz, while the AMD part boosts to 2100 MHz. The Intel base clock is 300 MHz, while the AMD base clock is 1000 MHz.

Q: How much memory does each GPU use?

A: The AMD Radeon Instinct MI308X has 192 GB of HBM3 with a 10.3 TB/s bandwidth. The Intel Arc Pro B370 uses system shared memory, so its capacity and bandwidth are system dependent and not fixed in the database.

Q: Which GPU supports ray tracing?

A: The Intel Arc Pro B370 lists 10 RT cores. The AMD Radeon Instinct MI308X lists no RT core count.

Q: What process node does each GPU use?

A: The AMD Radeon Instinct MI308X uses a 5 nm process at TSMC. The Intel Arc Pro B370 uses a 3 nm process at Intel.

Q: What is the TDP of each part?

A: The AMD Radeon Instinct MI308X has a TDP of 750 W and recommends a 1150 W PSU. The Intel Arc Pro B370 has a TDP of 25 W and lists no PSU requirement.

Architecture Differences

The AMD Radeon Instinct MI308X is built on CDNA 3.0 architecture with the Aqua Vanjaram chip. It uses a 5 nm process at TSMC and packs 153,000 million transistors on a 1017 mm² die. The transistor density is 150.4 million per square millimeter. This is a data center compute accelerator, part of the Radeon Instinct (MIx) generation, and it has no display outputs. Its predecessor is FirePro Data Center.

The Intel Arc Pro B370 is built on Xe3-LPG architecture with the Panther Lake chip. It uses a 3 nm process at Intel, with transistor count and die size listed as unknown. It is part of the Arc Graphics-WM (Panther Lake) generation and functions as an IGP, meaning it is integrated into a processor package. It supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Its predecessor is HD Graphics-WM, and its production status is listed as active.

The AMD part has zero ROPs and zero pixel rate, which reflects a design with no rasterization pipeline. The Intel part has 20 ROPs and a pixel rate of 48.00 GPixel/s, indicating a conventional graphics pipeline. The AMD part has no API support listed for DirectX, OpenGL, or Vulkan, consistent with a compute-only accelerator. The Intel part lists full graphics API support.

The AMD part uses HBM3 memory with an 8192-bit bus. The Intel part uses system shared memory, meaning the memory subsystem depends entirely on the host system. The AMD memory clock is 2525 MHz with 10.1 Gbps effective. The Intel memory clock is system shared. The AMD part connects via PCIe 5.0 x16, while the Intel part uses an IGP bus interface.

The AMD part is an OAM module, a board form factor for data center racks. The Intel part is an IGP, occupying no expansion slot. The AMD part has no power connectors because the OAM module draws power from the baseboard. The Intel part also has no power connectors because it draws power through the processor socket. The AMD part has no dimensions listed, and neither does the Intel part.

The release dates are widely separated. The AMD Radeon Instinct MI308X launched on December 5, 2023. The Intel Arc Pro B370 launched on January 26, 2026. Neither part has a successor listed in the database.

The Verdict

The data shows two parts built for entirely different purposes. The AMD Radeon Instinct MI308X is a high-power data center compute accelerator with 750 W TDP, 192 GB of HBM3, and 81.72 TFLOPS of FP32. The Intel Arc Pro B370 is a low-power integrated graphics solution at 25 W TDP, using system shared memory, with 6.144 TFLOPS of FP32 and full graphics API support.

For raw compute workloads, the AMD part is the only choice based on the recorded specifications. It has 13.3x the FP32 throughput, 53.2x the FP16 throughput, and 26.6x the texture rate. It also has dedicated memory with 10.3 TB/s bandwidth, while the Intel part depends on system memory. No benchmark scores exist to validate real-world performance, but the specification gap is so large that any measured result would need to be extraordinary to change the hierarchy.

For graphics and display workloads, the AMD part has no outputs and no raster pipeline. The Intel part has display outputs that are portable device dependent, supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, and has 10 RT cores. The Intel part is the only one of the two that can drive a display or run conventional graphics applications.

The Intel part also wins on integration. It is an IGP with a 25 W TDP, fits inside a Panther Lake processor, and requires no external power connectors or PSU recommendation. The AMD part requires an OAM slot, an 1150 W PSU, and has no power connectors because the module is powered by the baseboard.

Neither part has recorded benchmark scores, so the verdict rests on the specification sheets. For compute density and memory bandwidth, the AMD Radeon Instinct MI308X is the clear leader. For integrated graphics, display output, and low power, the Intel Arc Pro B370 is the clear leader. The two parts do not compete in the same market segment.

Specification Differences

| Field | AMD Radeon Instinct MI308X | Intel Arc Pro B370 |

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

| Chip | Aqua Vanjaram | Panther Lake |

| Architecture | CDNA 3.0 | Xe3-LPG |

| Generation | Radeon Instinct (MIx) | Arc Graphics-WM (Panther Lake) |

| Process Node | 5 nm | 3 nm |

| Foundry | TSMC | Intel |

| Transistors | 153,000 million | unknown |

| Die Size | 1017 mm² | unknown |

| Base Clock | 1000 MHz | 300 MHz |

| Boost Clock | 2100 MHz | 2400 MHz |

| Memory Clock | 2525 MHz, 10.1 Gbps effective | System Shared |

| Memory Size | 192 GB | System Shared |

| Memory Type | HBM3 | System Shared |

| Memory Bus Width | 8192 bit | System Shared |

| Memory Bandwidth | 10.3 TB/s | System Dependent |

| Shading Units | 19456 | 1280 |

| TMUs | 1216 | 40 |

| ROPs | 0 | 20 |

| RT Cores | null | 10 |

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

| Texture Rate | 2,553.6 GTexel/s | 96.00 GTexel/s |

| FP32 | 81.72 TFLOPS | 6.144 TFLOPS |

| FP16 | 653.7 TFLOPS (8:1) | 12.29 TFLOPS (2:1) |

| TDP | 750 W | 25 W |

| Slot Width | OAM Module | IGP |

| Power Connectors | None | None |

| Suggested PSU | 1150 W | null |

| Bus Interface | PCIe 5.0 x16 | IGP |

| Display Outputs | No outputs | Portable Device Dependent |

| DirectX | null | 12 Ultimate (12_2) |

| OpenGL | null | 4.6 |

| Vulkan | null | 1.4 |

| Production Status | null | Active |

| Release Date | 2023-12-05 | 2026-01-26 |

| Predecessor | FirePro Data Center | HD Graphics-WM |

Where Each One Wins

The AMD Radeon Instinct MI308X wins in every compute-heavy specification. It has 81.72 TFLOPS of FP32, 653.7 TFLOPS of FP16, 2,553.6 GTexel/s of texture rate, and 10.3 TB/s of memory bandwidth. It carries 192 GB of dedicated HBM3 memory, which removes any dependency on host system memory. Its 19,456 shading units and 1,216 TMUs dwarf the Intel part. It uses a PCIe 5.0 x16 interface, which supports high-bandwidth data transfer to the host. It also has a larger die at 1017 mm² and a transistor count of 153,000 million, both of which indicate a much more complex and capable silicon.

The Intel Arc Pro B370 wins in integration and conventional graphics features. It has 10 RT cores, which the AMD part does not list. It has 20 ROPs and a 48.00 GPixel/s pixel rate, while the AMD part has zero ROPs and zero pixel rate. It supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, while the AMD part lists no API support. It has display outputs, albeit portable device dependent, while the AMD part has none. Its 25 W TDP is 30x lower than the AMD part's 750 W. Its 3 nm process at Intel is a smaller node than the AMD part's 5 nm at TSMC. It is an IGP, meaning it needs no expansion slot, no power connectors, and no PSU recommendation.

The boost clock is the only clock specification where the Intel part leads, at 2400 MHz versus 2100 MHz. The base clock favors AMD at 1000 MHz versus 300 MHz. The Intel part also leads in production status, listed as active, while the AMD part has no production status listed.

Use-case separation is clear from the specifications. For AI training, HPC simulation, or any FP16-heavy workload, the AMD Radeon Instinct MI308X provides 653.7 TFLOPS of FP16 throughput, which is 53.2x the Intel part's 12.29 TFLOPS. For graphics rendering with ray tracing, the Intel Arc Pro B370 has RT cores and full graphics API support. For a compact, low-power integrated system, the Intel part fits in a processor package with no additional cooling or power infrastructure. For a rack-mounted accelerator, the AMD part provides the memory and compute density required for large-scale workloads.