>We're not here to persuade each other, but to put our ideas to the test of careful examination and to try to persuade others.
Young Jedi, you betray your roots. When Tom says MSFT's JVM is faster, you nitpick, you obfuscate, and you ridicule. Just like you would if you were cross-examining an expert witness. I, however, am a scientist. I question, I test, and I report my findings (more later).
>Assuming the Microsoft VM is faster, I guess that makes it better, I don't know.
Would it be better if it was slower? The mantra of technology is faster, smaller, cheaper. Or have you forgotten?
>So, is the Microsoft marketing strategy that users and developers will flock to Windows because (a) the Java VM on Windows is faster, and (b) developers can write Java programs which use native Windows features not available on other platforms?
a) You can have this car with 100hp. Or this one with 300hp. Which would you prefer?
b) You can have this car without air conditioning. Or this one with air conditioning. If you choose the car with the A/C, you are under no obligation to use it. Which would you prefer?
Anyway, enough of this so-called debate. Here's why I really posted.
JAVA VM COMPARISON
MOTIVATION: To test Tom's contention that Microsoft has a superior Java solution. To quantify what heretofore has been discussed in vague and qualitative terms. To teach Jerry a thing or two about scientific process.
METHODOLOGY: Tests were performed on my system running NT 4.0 SP3 with the latest versions of Netscape and Microsoft browsers available to me. Benchmark used was Caffeinemark 2.5, a Java test suite that contains *NO* implementation-specific optimizations. In other words, this is a benchmark of 100% pure Java.
RESULTS:
Netscape Communicator Pro 4.01, 1772.
Internet Explorer 4.0 post-Platform Preview beta, 2301.
Internet Explorer 4.0 latest beta, >47,000.
DISCUSSION: The latest Microsoft implementation is more than twenty times faster than the one before it, which was already a mainstream JIT compiler that is significantly faster than the Borland and Symantec JVMs that Netscape uses. As a point of reference, an Intel Pentium II 266Mhz processor is significantly less than twice as fast as an Intel Pentium 166Mhz MMX.
CONCLUSION: Smokin! |
