The concurrent GC is a great idea.&nbsp; A few interrogations.<br><br>- How &quot;stoppy&quot; would a stop-the-world parallel GC be in practice?&nbsp; The more parallelism<br>you have, the more work is done, the higher the frequency of a major collection.<br>
<br>- Would major allocations be serialized?&nbsp; What about other serialization points?<br><br>- I&#39;m afraid true concurrency will introduce an awful lot of bugs in native bindings.&nbsp; Thread-unsafe libraries will have to be replaced (Str, etc.)&nbsp; Also what would be the CPU<br>
and memory costs?&nbsp; Don&#39;t concurrent GCs require extra colors?<br><br>- In case of performance impacts, will the old single-threaded mode still be available?<br><br>The argument that &quot;you&#39;ll get the same old perfomance if you run it in single-threaded mode&quot;<br>
is not valid IMHO.&nbsp; Many people will use a thread here or there and then you won&#39;t realistically be able to run in single-threaded mode.<br><br>But then we can&#39;t pretend multi-core doesn&#39;t exist.&nbsp; A suggestion: making the parallel GC available only on 64-bit seems a reasonable restriction (if that&#39;s ever needed.)<br>
<br>Also Damien Doligez (in addition to Xavier Leroy) certainly have nice things to say about all this.<br>-- <br>Berke Durak<br><br>