3 ms·
Going by what I know of Azul's systems, their pauseless GC avoids a 1 second delay per GB of heap by focusing on solving the hardest case vs. deferring it. When
by hga 13y ago
Going by what I know of Azul's systems, their pauseless GC avoids a 1 second delay per GB of heap by focusing on solving the hardest case vs. deferring it. When running on their custom hardware (generic 64 bit RISC + their special sauce) it uses an instruction to cheaply implement a read barrier. According to a paper describing this system, which they've since improved, doing that on stock (x86_64) hardware would incur a ~20% performance penalty. If your application uses a lot of memory and can't afford pauses that's probably OK.
- nickik 13y agoThere Collector C4 is a soft-realtime they can not provide realtime but if they can be quite sure that they have very samll pauses. I truly belive in that vision. The problem is that the all the parts of they system have to be updated. At the moment they can run on x86_64 but its not as fast as it could be. The have a special build for every linux distribution. I am not sure how good this all works if you have very few cores and not a lot of threads. All in all however the C4 defently shows the way of the future. Hardware developer should learn from Azul, the OS guys should work more on supporting managed runtimes and the programming language community should do so to. Managed Memory is the future exept in very, very special cases.