News for hackers. Stuff that matters.
Re:Perceus: Garbage Free Reference Counting with Reuse [pdf] by pstch on Saturday December 19, 2020 @06:55AM (#25476862)
Java being faster on nqueens shows an interesting result : GCs can actually make code "faster", by moving some work to another core.

It's possible this could still be done with the described approach, but it looks much more difficult.

Re:Perceus: Garbage Free Reference Counting with Reuse [pdf] by djwatson24 on Saturday December 19, 2020 @11:45AM (#25478644)
Unless I am misreading their graph, Java beats neither c++ nor koka in time or rss.
Re:Perceus: Garbage Free Reference Counting with Reuse [pdf] by pstch on Monday December 21, 2020 @06:29PM (#25501009)
Oops, it was not on nqueens, but on deriv.

> Finally,Java performs best on this benchmark; we can see whilerunning the benchmark that it can run the G1 collectorfully concurrent on another core.

Re:Perceus: Garbage Free Reference Counting with Reuse [pdf] by Bootvis on Saturday December 19, 2020 @01:41PM (#25479880)
I read them the same. Now it is interesting that C++ is so much slower on cfold.