There is a very good article that benchmarks a number of jvms
including the blackdown jdk1.1.7+tya at
http://www.javalobby.org/features/jpr
One interesting point is that TYA is on various benchmarks
2-8 times slower than the Symantec jit included with Sun's java
under Windows (which in turn is slower than the forthcoming hotspot).
Since TYA is roughly the same speed as the 'sunwjit'
jit supplied by sun for the linux java1.2, this means that Sun supplied
linux a much slower jit than the one it supplies with Windows.
Is this because they cannot legally supply the Windows jit? Where
did this jit come from?
-----
Somewhat related, some timings of two programs that I had
available in both C and Java versions:
The SPLINEFILL program rasterizes a closed spline contour and
then uses a seed fill algorithm to color its interior.
The java version is 1500 lines, not including some libraries.
The seed fill is likely to be the inner loop but I did not verify this.
The RANDOM program calls a custom random number generator
a number of times and prints the last-generated number.
This is a rather too simple (8-line) integer-heavy algorithm.
In both cases I verified that the output was identical between the
C and Java versions. The C timings on the SPLINEFILL were so short
that they are probably not reliable, but maybe they give a rough
guess about where we are. The RANDOM timings are encouraging,
but it seems that this performance is only obtained for small routines
with no floating point or objects.
The machines involved are an Intel 166mmx running Linux 2.0.36
and an SGI (not running linux) using a 195mhz R10K.
Both C,Java programs were compiled without optimization.
The program times are "user" times in seconds as shown by the 'time' command.
----------------
SPLINEFILL/intel
gcc .23
jdk1.2prev1 .638
jdk117v1a/tya13 .621
jdk117v1a/tya11v4 .676
jdk117v1a/shujit0.2.10 .998
jdk117v1a(interp) 1.139
SPLINEFILL/sgi
cc 0.086u+0.047s (not reliable)
jdk115+jit .525
RANDOM/intel
gcc 12.2
jdk1.2prev1 14.37
tya13 15.1
jdk117v1a interpreted 29.8
RANDOM/sgi
cc 3.83
jdk115+jit 7.2
-----------------------------------------------------------------------
j.p.lewis [EMAIL PROTECTED]
012 345 6789 //www.idiom.com/~zilla
----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
