All About Sun Microsystems Message Board

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : All About Sun Microsystems

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: Krishna A. Ubrani who wrote (5380)	11/7/1997 6:55:00 PM
From: Chung Yang	Read Replies (1) of 64865

I thought they did. The representative from Javasoft said that they identified patterns from the Caffinemark and optmized around it. The idea is that if Caffinemark is indeed representative of the real world application, then real world application should benefit from such optimizations. Which is true. The problem with benchmark is that it has to be well understood and everyone has to know what it is supposed to do. And how to interpret the result. SpecInt95 and SpecFP95 is the closest thing to a good benchmark for microprocessors. BTW, this is a pretty common practice. For example, the SPEC performance number for Intel chips are from "special compilers" which you can't get anywhere else. The problem with benchmark for a virtual machine is that the performance can be tweaked to no end. The virtual machine can be as fast or as slow as you wanted it. The point is, you can make application that runs on virtual machine ungodly fast. If you know what application you are optimizing for. Or dog slow. My question is, why would you want to lift out a code to purposefully make the JVM slow? My feeling, you don't benchmark JVMs. It is useless. Caffinemark give you some kind of an idea of how fast your Java applets will run, but it doesn't tell you the whole story. I think they should benchmark all Java applications (inc JVM) against the MicroJava chip when it comes out later next year. There you will have no chance of tweaking anything except the compiler or to write better code. - Chung >>> I agree, Sun's response would have been more credible if they had owned up to the trick, (as can be understood from the available responses) but also said "It is standard industry practice and we are no different". I wouldn't want Solaris to look bad just because they did not do the same tricks that compitetors do. A better solution is to let a third (nuetral) party do the tests (which cannot be influenced by people up in NW). Because it does seem like Pendragon is right about the JVM not performing so well when it's code is lifted from it. Krishna <<<<

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next