6 ms·
2-10% for an already fast user space driver is nothing. State of the art for a lot of these use cases is still the kernel driver which is ~7 times slower. Sure
by emmericp 7y ago
2-10% for an already fast user space driver is nothing.
State of the art for a lot of these use cases is still the kernel driver which is ~7 times slower. Sure, all that stuff is moving to XDP/eBPF/AF_XDP, but that is still ~20-30% slower than a user-space driver.
Also, these 2-10% only show up when underclocking the CPU while running the unrealistic benchmark of forwarding packets bidirectionally on only one core (trivial to parallelize).
In the end it's about 6-12 cycles spent more in the driver. That's not a lot if you have a non-trivial application on top of it.