The fact that newer CPUs are not going up in clock speed but in number of cores 
means that performance can only come from effective multi-threading. Given Nim 
plays in the performant side of programming languages, having a competitive 
strategy for multi-threading is crucial to longterm success.

@Araq knows this and the fact that current GCs don't play well multi-threading 
is also well known. I don't see newruntime as a distraction but valuable 
research that will benefit everyone longterm.

What I will say though is that I hope the implementation can coexist with the 
GC so that mostly single threaded apps can continue to have the same user 
experience as today. The newruntime capabilities can be reserved for the 
standard library and good third party libs and for anyone who wants to leverage 
all the power of modern CPUs.

Over time, that use case will become more prevalent and the ecosystem that 
provides the best support and safety will go a long way. 

Reply via email to