True but on a large amount of cores you can just move the code to new
processes  instead of threads meaning the pauses become a non issue ( due to
the smaller memory) ... 

 

That simply isn't true. An arbitrarily large number of threads can become
blocked on the GC of a large object. More cores doesn't help.

 

 That depends on the architecture  eg a web server serving pages with a
separate GC for each engine will incur very small pauses if the # requests
to # process ratio is low.  I did mention some scientific  apps/3D  which
could be an issue  but I would be hesitant to design a GP GC initially for
these cases in most cases large data structures can be served in a number of
ways 

-          Isolate and use a ref counting GC 

-          as a large static object ( and hence play little roll in the GC
collect) 

-          GC don't move large objects.

 

 

Ben

_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to