Fujitsu's new middleware optimizes GPU usage by allocating resources intelligently, distinguishing between GPU-requiring applications and those manageable by CPU, enhancing computational efficiency.
The middleware demonstrated a significant 2.25x increase in efficiency during trials with leading companies, showcasing its capacity for real-time resource allocation and priority processing.
Fujitsu's innovation not only improves resource allocation but also integrates memory management capabilities, allowing GPUs to tackle tasks exceeding their memory limits.
While currently limited to single GPU setups, Fujitsu plans to develop the middleware for multi-GPU, multi-server environments, enhancing overall computing performance.
Collection
[
|
...
]