Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

"I don't see why programming for multicore processors has to be so complex. Why not just run four copies of the program on all four cores?"

This is actually a good idea.



That only works if your program is a one-way pipe for data, like a graphics card. Most programs need to communicate within themselves.


How about:

Message passing is more robust than the currently prevailing shared memory + thread spaghetti practice. Many/most problems being attacked with the latter approach can be reformulated to use message passing and looser coupling.


Tanenbaum, is that you?




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

Search: