[fh973]

Do multi-threaded workloads, which a lot of high performance workloads are, Java has also the potential to be faster than C++ due to garbage collection and threading aware optimizations it can make. Memory allocation, especially when memory lifecycle spans threads can be expensive in C++.

"A JVM does that???" by Cliff Click is a good introduction on what can go in the background.

[jackmott]

so write a specialized GC for the task in C or C++.

C++ can always win!

JVM has hotspot? use profile guided optmization in C!

There is no escape!

[sgift]

If you are one of the three people in the world that can do it. ;-)

[fauigerzigerk]

Just use an arena allocator (or similar) where appropriate. It's not difficult and it's how C, C++ and Rust win all micro benchmarks that involve memory allocation: http://benchmarksgame.alioth.debian.org/u64q/binarytrees.htm...