AMD Radeon Instinct MI300A vs Intel Arc Pro B370 Comparison

Head-to-Head Benchmarks

The database contains no direct benchmark comparisons between the AMD Radeon Instinct MI300A and the Intel Arc Pro B370. Both products have an average benchmark score of 0, and both sit at the 50th percentile among all GPUs in the database, indicating that neither has accumulated a measurable performance dataset. The head-to-head benchmark array is empty, and the recorded wins for each product are zero. Consequently, any direct performance comparison must be derived from the architectural specifications listed in the database, not from empirical test results.

The AMD Radeon Instinct MI300A delivers 81.72 TFLOPS of FP32 compute, while the Intel Arc Pro B370 delivers 6.144 TFLOPS. This represents a 13.3x advantage for the AMD part in raw single-precision floating-point throughput. In FP16, the MI300A reaches 653.7 TFLOPS (at an 8:1 ratio), whereas the Arc Pro B370 reaches 12.29 TFLOPS (at a 2:1 ratio). The FP16 gap is even larger, with the MI300A offering roughly 53x the FP16 throughput of the Intel part.

Texture throughput also favors the AMD accelerator decisively. The MI300A sustains 2,553.6 GTexel/s, compared to 96.00 GTexel/s for the Arc Pro B370, a 26.6x margin. The pixel rate tells a different story: the MI300A records 0 MPixel/s because it has no ROPs and no display outputs, while the Arc Pro B370 delivers 48.00 GPixel/s from its 20 ROPs. This difference reflects fundamentally different design goals: the MI300A is a compute-oriented accelerator without a rasterization pipeline, while the Arc Pro B370 is an integrated graphics processor with full display and pixel output capability.

Memory capacity and bandwidth further separate the two. The MI300A carries 192 GB of HBM3 memory across an 8192-bit bus, achieving 10.3 TB/s of bandwidth. The Arc Pro B370 uses System Shared memory, with its bandwidth described as System Dependent, meaning its memory performance is tied to the host platform rather than a dedicated pool. The MI300A's memory clock is 2525 MHz (10.1 Gbps effective), while the Arc Pro B370 has no dedicated memory clock recorded.

Where Each One Wins

The AMD Radeon Instinct MI300A wins in every compute-heavy workload category represented in the database. Its FP32 throughput of 81.72 TFLOPS and FP16 throughput of 653.7 TFLOPS position it for large-scale scientific computing, AI training, and simulation workloads that demand massive parallel floating-point execution. The 192 GB HBM3 memory pool with 10.3 TB/s bandwidth supports working sets far larger than any integrated GPU could accommodate, making it suitable for models and datasets that exceed the memory capacity of conventional systems. The 5 nm process node from TSMC, with 153,000 million transistors on a 1017 mm² die, provides the physical substrate for this compute density, achieving a transistor density of 150.4M per mm².

The Intel Arc Pro B370 wins in scenarios where power efficiency and system integration matter. Its 25 W TDP is 30x lower than the MI300A's 750 W TDP, allowing deployment in portable devices and compact platforms without dedicated power delivery infrastructure. The Arc Pro B370 is an IGP (integrated graphics processor) with no power connectors and no suggested PSU, whereas the MI300A is an OAM Module with a suggested PSU of 1150 W. The Intel part also provides display outputs (Portable Device Dependent), while the MI300A has no outputs whatsoever. For rasterization and pixel processing, the Arc Pro B370's 48.00 GPixel/s and 96.00 GTexel/s are functional, whereas the MI300A cannot produce pixels at all due to its 0 ROPs and 0 MPixel/s pixel rate.

The Arc Pro B370 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, making it compatible with modern graphics APIs. The MI300A lists no API support in the database, reinforcing its role as a compute accelerator rather than a graphics renderer. The Intel part also has 10 ray tracing cores, a feature entirely absent from the MI300A's specification sheet.

Architecture Differences

The two products belong to different architectural lineages. The AMD Radeon Instinct MI300A uses the CDNA 3.0 architecture, built on the Aqua Vanjaram chip, and belongs to the Radeon Instinct (MIx) generation. The Intel Arc Pro B370 uses the Xe3-LPG architecture, built on the Panther Lake chip, and belongs to the Arc Graphics-WM (Panther Lake) generation. Neither product lists a codename in the database.

Process technology separates them clearly. The MI300A is fabricated by TSMC on a 5 nm process node, with 153,000 million transistors on a 1017 mm² die. The Arc Pro B370 is fabricated by Intel on a 3 nm process node, with transistor count and die size recorded as unknown. The MI300A's transistor density works out to 150.4M transistors per mm², a figure not available for the Intel part.

The compute resources differ by an order of magnitude. The MI300A has 19,456 shading units, 1,216 texture mapping units, and no ROPs, ray tracing cores, or tensor cores listed. The Arc Pro B370 has 1,280 shading units, 40 TMUs, 20 ROPs, and 10 ray tracing cores, with tensor cores not listed. Clock behavior also diverges: the MI300A has a base clock of 1000 MHz and a boost clock of 2100 MHz, while the Arc Pro B370 has a base clock of 300 MHz and a boost clock of 2400 MHz. The Intel part boosts to a higher absolute frequency, but its much smaller shader array limits aggregate throughput.

Memory architecture highlights the different deployment contexts. The MI300A uses dedicated HBM3 memory: 192 GB, 8192-bit bus, 10.3 TB/s bandwidth, with a memory clock of 2525 MHz (10.1 Gbps effective). The Arc Pro B370 uses System Shared memory, with System Shared type, bus width, and bandwidth that is System Dependent. This means the Intel part has no independent memory resources; its performance scales with the host system's memory capabilities.

The bus interface and form factor also differ fundamentally. The MI300A connects via PCIe 5.0 x16 and occupies an OAM Module slot. The Arc Pro B370 is an IGP with an IGP bus interface, directly integrated into a processor package. Power delivery reflects this: the MI300A has a 750 W TDP and no power connectors (it draws power through the OAM module interface), with a suggested PSU of 1150 W. The Arc Pro B370 has a 25 W TDP, no power connectors, and no suggested PSU requirement.

Release timing and lineage also differ. The MI300A was released on 2023-12-05 and its predecessor is recorded as FirePro Data Center. The Arc Pro B370 was released on 2026-01-26, has a production status of Active, and its predecessor is HD Graphics-WM. Neither product has a successor listed in the database.

The Verdict

The data shows two products with no overlapping use cases. The AMD Radeon Instinct MI300A is a data-center compute accelerator with 81.72 TFLOPS FP32, 653.7 TFLOPS FP16, 192 GB of HBM3, and 10.3 TB/s of memory bandwidth. It draws 750 W, requires a 1150 W suggested PSU, and provides no display outputs. The Intel Arc Pro B370 is an integrated graphics processor with 6.144 TFLOPS FP32, 12.29 TFLOPS FP16, 48.00 GPixel/s pixel throughput, and 10 ray tracing cores. It draws 25 W, uses system shared memory, and supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

For workloads requiring massive parallel compute, large memory capacity, and high memory bandwidth, the MI300A is the only option with the recorded specifications to handle them. Its 13.3x FP32 advantage and 53x FP16 advantage over the Arc Pro B370, combined with 192 GB of dedicated HBM3, make it suitable for large-scale AI training and scientific simulation. The Arc Pro B370 cannot approach these figures, and its System Shared memory architecture places a hard ceiling on memory-bound workloads.

For systems requiring graphics output, rasterization, ray tracing, and low power consumption, the Arc Pro B370 is the only functional choice. The MI300A has no ROPs, no pixel rate, no display outputs, and no graphics API support, making it incapable of driving a display or rendering graphics. The Arc Pro B370's 25 W TDP allows integration into portable devices, and its IGP form factor eliminates the need for discrete power connectors or a separate PSU.

Neither product has benchmark scores in the database, so the percentile ranking of 50th for both reflects a lack of measured data rather than actual performance parity. The architectural records provide the only basis for comparison, and they indicate that these parts target entirely different market segments. A user requiring compute acceleration should select the MI300A; a user requiring integrated graphics should select the Arc Pro B370. There is no scenario in the recorded data where either product could substitute for the other.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The AMD Radeon Instinct MI300A delivers 81.72 TFLOPS, which is 13.3x the 6.144 TFLOPS of the Intel Arc Pro B370.

Q: Can the AMD MI300A output video to a display?

A: No. The MI300A has no display outputs and a pixel rate of 0 MPixel/s, with 0 ROPs, so it cannot produce or output any graphics frames.

Q: What memory does each product use?

A: The MI300A uses 192 GB of HBM3 memory on an 8192-bit bus with 10.3 TB/s bandwidth. The Arc Pro B370 uses System Shared memory with System Dependent bandwidth.

Q: What is the power consumption difference?

A: The MI300A has a 750 W TDP and requires a suggested PSU of 1150 W. The Arc Pro B370 has a 25 W TDP and lists no suggested PSU.

Q: Does the Intel Arc Pro B370 support modern graphics APIs?

A: Yes. The database lists DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 support for the Arc Pro B370. The MI300A has no API support listed.

Q: What are the release dates for these products?

A: The MI300A was released on 2023-12-05. The Arc Pro B370 was released on 2026-01-26 and has an Active production status.