Home / Taiwan Excellence Award / Award Winning Products / Product Details
TRUSTA’s AI Scaler Toolkit integrates hardware and software to dynamically orchestrate GPU, DRAM, and SSD resources, improving Large Language Model (LLM) performance in resource-limited environments. The toolkit optimizes workload distribution across GPU, DRAM, and SSD to reduce reliance on GPU HBM capacity. It automatically adjusts GPU memory usage, DRAM cache allocation, and SSD offloading based on workload requirements, enhancing AI training and inference efficiency. As LLMs continue to grow, high-capacity HBM GPUs face challenges including high cost, limited supply, and scalability constraints. TRUSTA AI Scaler Toolkit enables customers to expand AI computing capability without replacing existing GPUs or adding larger HBM capacity, delivering a more flexible and cost-effective AI infrastructure solution.
Award winning products for the year (7)
Award winning products in the past (50)
