Add a LoadUB (unsigned byte) and LoadUI (unsigned int) opcode class so we have these load optimizations in the first place and do not need to handle them in the matcher.

These classes will handle code like, e.g.:

bytearray[i] & 0xFF
intarray[i] & 0xFFFFFFFF
Posted Date : 2009-01-23 18:19:27.0

N/A

Some micro benchmarks show a big performance win.
Posted Date : 2009-02-24 15:45:02.0

http://hg.openjdk.java.net/jdk7/hotspot-comp/hotspot/rev/337400e7a5dd
Posted Date : 2009-03-09 12:43:57.0

@brackeen: Well the only thing I can think of (for x86) is to use partial register moves (using MOVZX instructions instead of SHR-AND sequences), e.g.:

  int g = (anInt >> 8) & 0xff;

  MOV    EBX,ECX
  SHR    EBX,#8
  AND    EBX,#255

vs.

  MOVZX8 EBX,CH

or:

  int b = anInt & 0xff;

  MOV    EBX,ECX
  AND    EBX,#255

vs.

  MOVZX8 EBX,CL

While the latter one is likely to have the same performance, as the resulting micro instructions should be the same.  I will run some tests.  Do you have a use-case you can send me?
Posted Date : 2009-09-21 11:53:23.0

Sorry about my duplicate:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6799997

... but maybe note my testcase code!

What's about:
    char c = (char)(b & 0xFF)
    char c = (char)(i & 0xFF)
After optimization, it should be fast as or faster as
    char c = (char)b
Goal: determination, if b is positive, becomes superfluous, so we could use { char c = (char)(i & 0xFF) } in any case without speed loss.

Also consider:
    char c = (char)(i & 0xFFFF)
    char c = (char)(intarray[i] & 0xFFFF)

Could a similar optimization be applied to extracting unsigned bytes from an integer? For example, getting the color components out of a packed ARGB:
int a = (anInt >>> 24);
int r = (anInt >> 16) & 0xff;
int g = (anInt >> 8) & 0xff;
int b = anInt & 0xff;