search

dbeurle / neon

A finite element code
Other
10 stars 8 forks source link

Thread local stiffness matrices #119

Closed dbeurle closed 5 years ago

dbeurle commented 5 years ago

This work enables thread local storage for static function stiffness matrices. By using thread local storage we avoid continual allocation and deletion of memory, resulting in hopefully less heap fragmentation over longer simulations with multiple threads.