Taiwanese edge AI firm Aetina has released embedded GPU modules using Nvidia architecture for use in smart healthcare, autonomous machines, smart manufacturing and commercial gaming.

The embedded MXM GPU series uses the Nvidia Ada Lovelace architecture. The MX2000A-VP, MX3500A-SP and MX5000A-WP are for real-time ray tracing and AI-based neural graphics. They enhance GPU performance, delivering gaming, creative and professional graphics, AI, and compute performance.

The global GPU (graphics processing unit) market is expected to achieve a 34.4% CAGR from 2023 to 2028, with advancements in the artificial intelligence (AI) industry being a key driver. As the trend of AI applications expands from the cloud to edge devices, many businesses are seeking to increase AI computing performance in small devices due to space constraints in deployment environments.

The MXM modules (www.aetina.com/products-list.php?t=23) use the Nvidia Ada Lovelace architecture.

The MX5000A-WP is equipped with the Nvidia RTX 5000 Ada GPU. It offers up to 9728 Cuda cores, 76 RT cores and 304 Tensor cores, supporting PCIe 4.0 interface and up to 16Gbyte of GDDR6 memory. With a power consumption of 115W, it meets the needs of edge AI applications requiring intensive computation within a compact, high-performance form factor.

Additionally, Aetina (www.aetina.com) offers various MXM options with different performance levels, including the MX3500A-SP with Nvidia RTX 3500 Ada GPU and MX2000A-VP with Nvidia RTX 2000 Ada GPU, to meet the requirements of different AI applications.

The modules have conformal coating for harsh environments, error correction code (ECC) for data integrity, and five-year support, fostering innovation in edge AI.