Re:JOP: A Tiny Java Processor Core for FPGA
by akeefer on Saturday October 17, 2009 @01:53AM
(#886687)
Azul systems (http://www.azulsystems.com/) has a special-purpose Java processor, and they do indeed get some impressive speed benefits from it. For example, having single-instruction write barriers in hardware allows them to have nearly-pauseless garbage collection on gigantic heaps.
Re:JOP: A Tiny Java Processor Core for FPGA
by bokchoi on Saturday October 17, 2009 @05:41AM
(#886814)
Here's an excellent google tech talk by Cliff Click of Azul talking about their hardware. Among the firehose of information in that talk, he mentions that the hardware guys wanted to add directly execute JVM bytecode. He told them not to bother since he could JIT it better and they wouldn't be able to support all the bytecodes anyway. I'd love to know more about why that is.
Re:JOP: A Tiny Java Processor Core for FPGA
by pmjordan on Saturday October 17, 2009 @06:03AM
(#886819)
JITing means you can do dynamic optimisations, e.g. predicting vtable dispatches. Doing this in hardware is tricky.
Re:JOP: A Tiny Java Processor Core for FPGA
by sketerpot on Saturday October 17, 2009 @03:40AM
(#886749)
Many of the ARM processor cores have hardware support for the most common Java bytecode operations; that's the "Jazelle" extension that others have been mentioning. (The less common operations are left to software, to keep silicon overhead low.)
Re:JOP: A Tiny Java Processor Core for FPGA
by richardw on Saturday October 17, 2009 @01:48AM
(#886684)
Also, Transmeta used to have a processor with configurable instructions so you could get it to mimic a given machine. Why, why why did they not make a JVM instead of trying to compete with Intel on their home ground?
See http://en.wikipedia.org/wiki/Java_processor