Talk:Microprocessor Design/Cache

Response to a question on the page

Open question: To speed up running Java applications in a JVM (and similar interpreters and CPU emulators), would it help to have 3 separate caches -- a machine instruction cache indexed by the program counter PC, a byte code cache indexed by the VM's instruction pointer IP, and a data cache ?

I suppose not. On current CPUs, the bytecode is treated as data (accessed by the same instructions) or compiled just-in-time (transformed into instructions). No CPU, except perhaps some ARMs with Jazelle, handles bytecode in a specific way: there is no need to waste area and add *a lot* of complexity to handle bytecode differently, when it already fits perfectly in the split D/I cache hierarchy. I suggest someone removes the question from the page, since it hasn't much theoretical interest IMHO. Topeil (discuss • contribs) 13:23, 9 August 2013 (UTC)