High Performance
While the performance of interpreted bytecodes is usually more than adequate, there are situations where higher performance is required. The bytecodes can be translated on the fly (at runtime) into machine code for the particular CPU the application is running on. For those accustomed to the normal design of a compiler and dynamic loader, this is somewhat like putting the final machine code generator in the dynamic loader.
The bytecode format was designed with generating machine codes in mind, so the actual process of generating machine code is generally simple. Efficient code is produced: the compiler does automatic register allocation and some optimization when it produces the bytecodes.
In interpreted code we're getting about 300,000 method calls per second on an Sun Microsystems SPARCStation 10. The performance of bytecodes converted to machine code is almost indistinguishable from native C or C++.
No comments:
Post a Comment