论文部分内容阅读
相信各位在观看计算机类产品广告中,一定会发现有许多产品广告中都提供了一连串的性能测试数据和相应的性能图表,产品推销商在向你介绍该款产品时也会列举这些数据作为参考,以证明该产品的性能有多强大!那么事实是这样的吗? Winstone等专业性能测试程序是权威检测机构用来对产品进行测试的主要依据,应该说这些程序所反映出的数据是真实可靠的,但事实真是这样吗? SUN的Java事件 SUN公司就给了我们一个很好的答案,去年SUN公司曾参加了一个名为“CaffeineMark”的产品评测活动,他们的参评产品为Java虚拟器(JVM)。SUN公司为了在评测中取得好的成绩,所以就对他们的Java虚拟器的即时编译器(JITCompiler)进行了一些小调整,即让编译器针对评测用的程序去刻意地避开那些会参加运算,但不会被用到的“Dead Code”的变数例行程序,这样SUN公司的Java虚拟器在“CaffeineMark”所进行的九项测试中竟一路领先,成绩比以往好上几十倍。一般对Java有所了解的人都会对Java虚拟器如
I believe you are watching computer products advertising, we will find a lot of product ads are provided in a series of performance test data and the corresponding performance charts, product vendors introduce you to the product will also list these data as a reference , To prove that the performance of the product is how powerful! Then the fact is it? Winstone and other professional performance testing program is the main basis for the authority of testing agencies to test the product, it should be said that these programs reflect the data is true and reliable Sun’s Java event SUN company gave us a good answer last year, SUN company participated in a product called “CaffeineMark” evaluation of their products for the evaluation of the Java Virtual Machine ( JVM). SUN company in order to get good results in the evaluation, so on their JITCompiler Java JITCompiler made some minor adjustments, that is, the compiler for the evaluation program to deliberately avoid those who will participate in the operation , But will not be used in the “Dead Code” variable routine, so that SUN’s Java Virtual Machine went one notch ahead of nine tests conducted by CaffeineMark, scores several times better than before. People who are familiar with Java generally use Java virtual machines