3 ms·
Anyone that thought otherwise doesn't understand how complicated context switches are in CPUs That’s not true. I understand how context switches work down to t
by wruza 2y ago
Anyone that thought otherwise doesn't understand how complicated context switches are in CPUs
That’s not true. I understand how context switches work down to tss records, but can’t immediately see why nonblock should be less efficient. Is it due to for-rw vs for-poll-rw? Doesn’t kqueue/iocp ought to solve that?
- nullindividual 2y agokqueue, like all non-io_uring implementations on Linux, are synchronous under the hood.