Friday February 29, 2008
Friday February 29, 2008 The SPECjbb2005 World Record results on Sun Intel systems are now live on the SPEC website.
World Record Performance on 2-chip x86 Systems running 4-JVMS: 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM
World Record Performance on x86 Systems with a Single JVM: 277,585 SPECjbb2005 bops
The Sun JDK powered by the HotSpot JVM is the the most widely deployed, scalable, and reliable JVM on the planet, but we need your help to continue to improve. Please participate at OpenJDK and the performance forum at Java.net, let us know what your application needs for reliable performance. You can even dive in and help. Competitive benchmarks are fun, but at the end of the day its our customer's application that really matter, so let us know what we can do to help!
SPEC Disclosure Statement
SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Sun Fire X4150 results submitted to SPEC. Other results as of 02/15/08 on www.spec.org. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM. Dell 2950 III (2 chips, 8 cores, BEA JRockit 6.0 P27.4.0) 303,130 SPECjbb2005 bops , 75,783 SPECjbb2005 bops/JVM.
World Record Performance on 2-chip x86 Systems running 4-JVMS: 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM
World Record Performance on x86 Systems with a Single JVM: 277,585 SPECjbb2005 bops
The Sun JDK powered by the HotSpot JVM is the the most widely deployed, scalable, and reliable JVM on the planet, but we need your help to continue to improve. Please participate at OpenJDK and the performance forum at Java.net, let us know what your application needs for reliable performance. You can even dive in and help. Competitive benchmarks are fun, but at the end of the day its our customer's application that really matter, so let us know what we can do to help!
SPEC Disclosure Statement
SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Sun Fire X4150 results submitted to SPEC. Other results as of 02/15/08 on www.spec.org. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM. Dell 2950 III (2 chips, 8 cores, BEA JRockit 6.0 P27.4.0) 303,130 SPECjbb2005 bops , 75,783 SPECjbb2005 bops/JVM.
Powered by ScribeFire.
Friday February 22, 2008 Some rather ridiculous comments from "Rick Jones" on the Bmseer's blog prompted me to write this. Rick stated:
"If that four JVM figure wasn't run over and over again,
one is left wondering just how bad it was before. Given
the same CPUs at the same frequency, one would expect
more from "major impact" performance changes."
We have made huge performance improvements over the last year. Yes, our previous capabilities on the Intel platform were less than what we can do now. I thought that would be obvious.
But you know what the biggest difference with our SPECjbb2005 World Record? Sun Hardware + Sun Solaris + Sun HotSpot JVM. Its not a result from any old box vendor who runs with the latest and greatest OS and JVM and claims the result as their own. This is a Sun World Record, through and through. Who else can deliver the entire stack, its a short list isn't it?
Hmm, Think of the possibilities for customers. Who do you call when you have a JVM issue? Sun. OS issue? Sun. Hardware issue? Sun. What if you need a specific feature which requires a tailored approach from hardware to OS to JVM? Sun. Thats the difference.
"If that four JVM figure wasn't run over and over again,
one is left wondering just how bad it was before. Given
the same CPUs at the same frequency, one would expect
more from "major impact" performance changes."
We have made huge performance improvements over the last year. Yes, our previous capabilities on the Intel platform were less than what we can do now. I thought that would be obvious.
But you know what the biggest difference with our SPECjbb2005 World Record? Sun Hardware + Sun Solaris + Sun HotSpot JVM. Its not a result from any old box vendor who runs with the latest and greatest OS and JVM and claims the result as their own. This is a Sun World Record, through and through. Who else can deliver the entire stack, its a short list isn't it?
Hmm, Think of the possibilities for customers. Who do you call when you have a JVM issue? Sun. OS issue? Sun. Hardware issue? Sun. What if you need a specific feature which requires a tailored approach from hardware to OS to JVM? Sun. Thats the difference.
Powered by ScribeFire.
Thursday February 21, 2008 Our latest SPECjbb2005 World Record on the Sun Fire X4150 is getting some coverage on Sun web pages and blogs.
As always, the bmseer has something to say, check out his post on the SPECjbb2005 World Record here.
The Sun marketing folks posted a page as well, check it out here.
We're very excited about these results, and hope our customers are too. Let us know if you have any questions!
As always, the bmseer has something to say, check out his post on the SPECjbb2005 World Record here.
The Sun marketing folks posted a page as well, check it out here.
We're very excited about these results, and hope our customers are too. Let us know if you have any questions!
Powered by ScribeFire.
Friday February 15, 2008 It was one year ago that Sun and Intel kicked off a collaboration to optimize Solaris and the HotSpot JVM for Intel Xeon systems. The last year has been a fun ride, working together we've made our performance steadily increase, and finally today is the day.
Without further ado,
Sun HotSpot JVM, Solaris, and the Sun Fire X4150 with 2 Intel X5460 quad-core processors now hold 2 SPECjbb2005 World Records!
World Record Performance on 2-chip x86 Systems
303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM running 4 JVMs!
World Record Performance on x86 Systems with a Single JVM
277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM
Thats right, Sun is fastest on Intel DP, inching past BEA to take the lead. Game on JRockit.
We submitted two results, one with 4 JVMs, and another with 1 JVM. Why? When using SPECjbb2005 to compare software and hardware configurations, it's critical to run and compare both single-JVM and multi-JVM configurations. With both single-JVM and multi-JVM results a more complete picture can be drawn of the performance capabilities a software/hardware stack. If your application is horizontally distributed, but is configured to run 1 JVM per system, then look at the single JVM results. If you plan to configure each system with processor sets and/or core affinity and will run multiple JVMs, look at the multi-JVM results. I encourage my colleagues from other JVM vendors to submit both single JVM results and multi-JVM results. We, and our competitors owe it to the Java community to show the whole truth about JVM performance. It's the Java community which will benefit from the active competition and drive JVM development forward. I've written about this before, take a look here for more background.
The JRE we used to achieve the world record results is Java 6 Update 5 Performance Release (Java 6 Update 5-p). This is our second performance release, and will be available mid-May 2008. Our first performance release Java 6 Update 4-p, is available on our performance page at java.sun.com/performance. Java 6 Update 4-p is a step along the way to our current high scores, and is only ~10% behind Java 6 Update 5-p in the general case. Go ahead, give it a try and let us know what you think.
Congratulations to the Sun and Intel Java performance collaboration team. For the last year we've worked together to deliver some grand performance improvements. Great work everyone, let's keep up the pace!
SPEC Disclosure Statement
SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Sun Fire X4150 results submitted to SPEC. Other results as of 02/15/08 on www.spec.org. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM. Dell 2950 III (2 chips, 8 cores, BEA JRockit 6.0 P27.4.0) 303,130 SPECjbb2005 bops , 75,783 SPECjbb2005 bops/JVM.
Without further ado,
Sun HotSpot JVM, Solaris, and the Sun Fire X4150 with 2 Intel X5460 quad-core processors now hold 2 SPECjbb2005 World Records!
World Record Performance on 2-chip x86 Systems
303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM running 4 JVMs!
World Record Performance on x86 Systems with a Single JVM
277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM
Thats right, Sun is fastest on Intel DP, inching past BEA to take the lead. Game on JRockit.
We submitted two results, one with 4 JVMs, and another with 1 JVM. Why? When using SPECjbb2005 to compare software and hardware configurations, it's critical to run and compare both single-JVM and multi-JVM configurations. With both single-JVM and multi-JVM results a more complete picture can be drawn of the performance capabilities a software/hardware stack. If your application is horizontally distributed, but is configured to run 1 JVM per system, then look at the single JVM results. If you plan to configure each system with processor sets and/or core affinity and will run multiple JVMs, look at the multi-JVM results. I encourage my colleagues from other JVM vendors to submit both single JVM results and multi-JVM results. We, and our competitors owe it to the Java community to show the whole truth about JVM performance. It's the Java community which will benefit from the active competition and drive JVM development forward. I've written about this before, take a look here for more background.
The JRE we used to achieve the world record results is Java 6 Update 5 Performance Release (Java 6 Update 5-p). This is our second performance release, and will be available mid-May 2008. Our first performance release Java 6 Update 4-p, is available on our performance page at java.sun.com/performance. Java 6 Update 4-p is a step along the way to our current high scores, and is only ~10% behind Java 6 Update 5-p in the general case. Go ahead, give it a try and let us know what you think.
Congratulations to the Sun and Intel Java performance collaboration team. For the last year we've worked together to deliver some grand performance improvements. Great work everyone, let's keep up the pace!
SPEC Disclosure Statement
SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Sun Fire X4150 results submitted to SPEC. Other results as of 02/15/08 on www.spec.org. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 303,297 SPECjbb2005 bops, 75,824 SPECjbb2005 bops/JVM. Sun Fire X4150 (2 chips, 8 cores, Sun JDK 6u5-p) 277,585 SPECjbb2005 bops, 277,585 SPECjbb2005 bops/JVM. Dell 2950 III (2 chips, 8 cores, BEA JRockit 6.0 P27.4.0) 303,130 SPECjbb2005 bops , 75,783 SPECjbb2005 bops/JVM.