On Saturday, 4 February 2017 at 06:54:01 UTC, Ali Çehreli wrote:
On 02/03/2017 08:47 PM, Profile Anaysis wrote:
What is the overhead of using a fiber?
The performance overhead of call() and yield() are comparable
to function calls because it's simply a few register
assignments in each case. (Change the stack pointer, etc.)
Memory overhead is memory for call stack, size of which can be
determined by the programmer.
Ali
Am I right understand that every yield(ed)/blocking function is
delegate to processing in system thread?
But what will be if our code would do a lot of blocking
functions? All of them will be delegate to system threads? If how
many system threads will be used? I am asking because it will
look like that in system threads we will get same problem that
was solved by fibers.
