This is the paper: https://www.linuxplumbersconf.org/2013/ocw/system/presentati...
I'm also really curious why they require modifications to the Linux kernel. My first guess would be stronger integration with the IO model at the syscall boundary (similar to io_uring).
Edit: is this the talk your referring to? https://www.youtube.com/watch?v=KXuZi9aeGTw
With the fibers implementation you can just do that. It doesn't kill your performance, and you don't need to go to a painful async model just for performance reasons.
If by “nowadays” you mean ~2000 when I first learned socket programming (using select!)? ;-)
Boost also has fiber.
How does Google's Fibers differ from other C++ fiber implementations? What makes it so wonderful?