Shulou(Shulou.com)06/03 Report--

What is the performance gap between JVM1.6 and JVM1.5? many novices are not very clear about this. In order to help you solve this problem, the following editor will explain it in detail. People with this need can come and learn. I hope you can get something.

The performance gap between JVM1.6 and JVM1.5, through the testing tool VolanoBenchMark, we can clearly see the gap between the two, as well as the progress of the new software. In fact, this is originally a WebChat software. Based on the function of ChatApp, the author developed this small software to test the performance of JVM, which can be used to test the performance of different JVM on machines with different hardware configurations or on different Kernel operating systems. It measures the time required for the client program to broadcast batch information by establishing multiple client connections to the server, so as to achieve the results of the JVM performance test.

Horizontal comparison of JVM1.6 and JVM1.5 performance

In fact, this software can evaluate the performance of the server in the following aspects: scheduler, network throughput and response of TCP/IP and EthernetDriver, Pthreads threads, SMP (SymmetricMultiProcessing symmetric multiprocessing) performance measurement.

Configuration of Benchmark

Number of CHAT rooms: 10

Number of users per room: 20

Amount of information per user: 1000

The server and client are on the same computer.

JVM used in the test

SunClientJVM1.6.0-beta-b59g:

Jdk-6-beta-windows-i586

Optionsused:-server-Xmx128m (default)

SunClientJVM1.5.0_01-b08:

Jdk-1_5_0_01-windows-i586-p

Optionsused:-server-Xmx128m (default)

I would like to add here that I did not configure the startup Option of JVM, because what is done here is only a comparison between Mustang (1.6m) and Tiger (1.5m), so we will not discuss whether to choose server or Client/Hotspot and whether the pre-allocated Heap size is 128m or 512m. If you need more in-depth discussion, welcome to discuss, here I personally feel-server is faster than-Client (that is, Hotspot in the Java option), starting will be a little slower, but running faster. At the same time, Net mode is not used for testing here, only Loop is used.

Hardware configuration

OS:MicrosoftWindowsXP [version 5.1.2600] (ServicePack2)

CPU:AMDAthlon (tm) XP2200+Thoroughbred-BProcessor (0.13 μ m)

Memory:512MB

Cache:64+64KBL1Cache+258KBL2Cache

Test results:

The test results are measured by the amount of information per second, and the test results are taken 5 times.

The average of the test excluding * * and * *:

Test results:

MustangJVM1.6 β: 8969.667

TigerJVM1.5:9394

It didn't feel right, so I restarted the computer and tried again:

JVM1.6 and JVM1.5 test results:

MustangJVM1.6 β: 9011.333

TigerJVM1.5:9335.333

The performance of JVM1.5 is better than JVM1.6 β. Personally, I don't quite understand, of course, I don't think about using Server or Client and other factors, but as an ordinary application developer, the performance of beta 1.6JVM doesn't surprise us.

After comparing the performance of JVM1.6 and JVM1.5 horizontally:

A foreign Blog columnist once said that version 1.5 is a great improvement of Java, but 1.6 makes us see Java in its twilight years, but firmly believes that Java will not Dieout, because we can see from languages such as COBOL or C++, even in this era, we can still see the widespread use of COBOL and C++, who have lived for decades, while Java is only 10 years old, and Java has the largest user base ever. So I believe that Java will be stronger. But it is also believed that there will soon be another young "Java" to replace the old Java, just as Java was born.

There may be a lot of mistakes in my evaluation results. You are welcome to point out that at the same time, the JVM of Mustang I use here is relatively old in the Demo version, which may be an important factor affecting the results. Sun's Java website also has evaluation results on two Demo versions of MustangJVM and TigerJVM, and the results are the same. The old version of MustangJVM is lower than Tiger, but the new version of MustangJVM can exceed Tiger by 3040%. But I'm not sure if it's the old version.

If you want to use VolanoBenchmark to test JVM, network throughput and server performance, I can provide a brief tutorial. In addition to Volano, there are actually several software to test the performance of Java servers, all three of which you can take a look at:

One of RUBiSbenchmarkObjectWeb's JMOB projects, not to mention that you don't know what ObjectWeb is, it is the website of the international open source middleware consortium management community. The function of this testing tool is large in three. I don't have time to get in touch with her, so I don't use her. Recommended!

VolanoMark, needless to say, free and easy to use, requires cygwin to run in a Windows environment (actually just to run shell files, and you can transform it without cygwin if you are confident about your DosBatch level). Ability to test server performance, network performance, and application performance. If you need help, you can leave a message or write to chenpengyi#gmail.com.

* one is SPECjbb2000, which used to be called 2000, but now its name has been changed to 2005. It seems that there is a charge, which is not cheap, so ahem, say no more. Please let me know if you know this is free.

Is it helpful for you to read the above content? If you want to know more about the relevant knowledge or read more related articles, please follow the industry information channel, thank you for your support.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.