Hacker News new | past | comments | ask | show | jobs | submit login

Another reason has to do with memory access and cache invalidation, when working with sparse matrices the entire graph structure is laid out in a contiguous block of memory which is ideal for CPU caching, unlike adjacency list where you have lots for small allocations scattered all over the memory space, this makes a huge difference when traversing the graph.



Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: