We are getting veryclose to having true multiprocessor support for networking.
2 Replies to “Big Lock almost ready to go”
I do try to keep up, but some of the finer details elude me still… is this at least comparable to the state of multithreaded, giant-free networking that will be in FreeBSD 5.3, or does it go beyond?
Yes, I am aware that DragonFly’s networking code was made nearly completely MP safe in *much* less time that FreeBSD’s code was (congratulations on that BTW ;^), I’m just not sure if their recent activity has brought them up to speed so that both projects are now on “equal” footing or not.
Jeff Hsu can speak better on this, but I don’t think he reads here. “Equal” is not really a valid comparison here, since the methodology is completely different.
From what I understand, programming with the DragonFly version of this will be much easier; performance probably will be better, but I’m basing that on prior example and not testing. :)
I do try to keep up, but some of the finer details elude me still… is this at least comparable to the state of multithreaded, giant-free networking that will be in FreeBSD 5.3, or does it go beyond?
Yes, I am aware that DragonFly’s networking code was made nearly completely MP safe in *much* less time that FreeBSD’s code was (congratulations on that BTW ;^), I’m just not sure if their recent activity has brought them up to speed so that both projects are now on “equal” footing or not.
Jeff Hsu can speak better on this, but I don’t think he reads here. “Equal” is not really a valid comparison here, since the methodology is completely different.
From what I understand, programming with the DragonFly version of this will be much easier; performance probably will be better, but I’m basing that on prior example and not testing. :)