> I actually really like RCU for shared access: https://en.m.wikipedia.org/wiki/Read-copy-update Deallocation is always done in a background thread and there is no synchronization when reading. I'm not sure what public libraries are available, unfortunately (I've only used our internal rcu library).
I do like "eventually consistent" data structures like this where you either see the old one or the new one consistently.
However, at that point, you've basically created garbage collection as you've lost your deterministic behavior. (It will get collected--sometime, maybe).
They mention the Linux kernel and reference counting specifically, so I'll look at this more in depth.
> If you want to make efficient use of your zillion cores, you need to make sure single threaded performance is excellent ;) See Amdahl's law.
Erm, that's NOT what Amdahl's law says.
"The overall performance improvement gained by optimizing a single part of a system is limited by the fraction of time that the improved part is actually used"
Cores mostly spend their time waiting--so improving single core performance isn't a great benefit. In fact, going backwards to lot of simpler cores but lots more cache per core is probably a better benefit.
Apple got this. They made the memory system a giant cache and got a huge performance boost.
I do like "eventually consistent" data structures like this where you either see the old one or the new one consistently.
However, at that point, you've basically created garbage collection as you've lost your deterministic behavior. (It will get collected--sometime, maybe).
They mention the Linux kernel and reference counting specifically, so I'll look at this more in depth.
> If you want to make efficient use of your zillion cores, you need to make sure single threaded performance is excellent ;) See Amdahl's law.
Erm, that's NOT what Amdahl's law says.
"The overall performance improvement gained by optimizing a single part of a system is limited by the fraction of time that the improved part is actually used"
Cores mostly spend their time waiting--so improving single core performance isn't a great benefit. In fact, going backwards to lot of simpler cores but lots more cache per core is probably a better benefit.
Apple got this. They made the memory system a giant cache and got a huge performance boost.