ADDITIONAL SYSTEM INFORMATION :
Windows 10 / Server 2019

A DESCRIPTION OF THE PROBLEM :
Currently sending large arrays to native code is "copied" which is painfully slow.
Using a Direct ByteBuffer is not possible because is doesn't support array() and converting all code to use a direct ByteBuffer is painful as well.
Need either a GC that doesn't do compacting (64bit systems have lots of memory) that allows unlimited pinning or some other way to pass large arrays to JNI without copying it.
Creating video editing apps in Java without this is too slow.

Thanks,