GSI Technology, Inc. (Nasdaq: GSIT), a pioneer in Associative Processing Units (APUs) and compute-in-memory (CIM) technology, announced significant findings from a Cornell University-led research paper. The study reveals that GSI’s Gemini-I APU achieves GPU-equivalent performance on large-scale AI workloads while reducing energy consumption by over 98% compared to traditional GPUs, highlighting its efficiency and sustainability. Additionally, the APU outperforms standard CPUs in data retrieval tasks, drastically reducing processing time by up to 80%. This validates GSI Technology’s claim that CIM can disrupt the $100 billion AI inference market. The Cornell research introduces an analytical framework enhancing the APU’s scalability for developers. GSI’s second-generation Gemini-II APU promises even faster throughput and improved energy efficiency. As industries increasingly prioritize performance-per-watt, GSI is poised to capitalize on markets such as Edge AI, data centers, and aerospace applications. For more, visit GSI’s official site.
Source link
