Search
A research team from Shenzhen Beili Moscow University has pioneered a high-performance algorithm that dramatically enhances the computational prowess of NVIDIA's consumer-grade GPUs in scientific applications, particularly in the realm of peridynamics calculations, where it achieves a staggering 800-fold increase in speed. Leveraging CUDA programming technology, this algorithm refines both the algorithmic framework and memory management, effectively harnessing the GPU's parallel processing capabilities. In large-scale simulations, the algorithm excels, finding utility across diverse domains such as aviation, engineering, and military applications. Notably, it does not necessitate high-end GPU chips and can operate seamlessly on standard consumer-grade GPUs, rendering it impervious to US sanctions.
-
C114 Communication Network -
Communication Home
