
Latest benchmarks prove outstanding No.1 Java performance
The combination of PRIMEPOWER and Solaris™ OS has again proved to be the best platform for Java server applications
February 02, 2006
PRIMEPOWER has achieved No.1 Java application performance in total throughput and throughput per JVM(Java Virtual Machine) instance. A PRIMEPOWER 2500 achieved world No.1 performance with 1,157,619 bops tested in 16 JVM instances environment in the SPECjbb® 2005, Java application benchmark. The PRIMEPOWER also achieved No.1 performance per JVM instance, with 811,607 bops in a single JVM instances environment. This means PRIMEPOWER is a triple crown winner in Java performance benchmarking. PRIMEPOWER 2500 maintains its world No.1 record in SPECjbb® with a recorded 2,586,698 ops/sec. Plus PRIMEPOWER also keeps its No.1 position in SPECjAppServer® 2002 (Multiple Nodes) gained in Aug. 1st 2004.
SPECjbb2005, simulating a distributor's ordering application, evaluates total Java server performance related to both hardware and operating system.
Note: All benchmark results are as of 2th Feb. 2006.
For the detail please refer to "SPEC - Standard Performance Evaluation Corporation"
PRIMEPOWER’s world No.1 records in Java benchmark tests
| Bechmark | Results | Environment |
|---|---|---|
| SPECjbb2005 (New) | 1,157,619bops, 72,351bops/JVM |
PRIMEPOWER 2500 (SPARC64™ V, 2.08GHz, 128CPUs[128cores]) OS:Solaris 10 Number of JVM instances : 16 |
| 811,607bops, 811,607bops/JVM |
PRIMEPOWER 2500 (SPARC64™ V, 2.08GHz, 128CPUs[128cores]) OS:Solaris 10 Number of JVM instances : 1 |
|
| SPECjbb2000 (New) | 2,586,698ops/sec | PRIMEPOWER 2500 (SPARC64™ V, 2.08GHz, 128CPUs[128cores]) OS:Solaris 10 |
| SPECjAppServer2002 (Multiple Node) |
5,991TOPS | Application Server PRIMEPOWER 450 (SPARC64™ V, 1.32GHz, 4CPUs[4cores]) x 9 servers Oracle Application Servers 10g DB server PRIMEPOWER 2500 (SPARC64™ V, 1.3GHz, 56CPUs[56cores]) Oracle Database 10g OS : Solaris 9 |
What is SPECjbb2005 ?
SPECjbb2005 is a benchmark measure developed by SPEC (Standard Performance Evaluation Corporation) to evaluate server performance in Java-based business application environments. This benchmark simulates a distributor’s ordering application, and evaluates system performance, not limited to Java performance. Hardware performance related to CPUs and memory is evaluated through processing data in memory. JVM performance and related JIT (Just-In Time) compiler and garbage collector contained in JVM are also evaluated.
SPECjbb2005 has two performance indices: total throughput ( by unit name, bops) and throughput per JVM instance (by unit name, bops/JVM). Where an application uses the majority of CPUs and memory resources, you should refer to the throughput per JVM index. When multiple applications run in parallel, you should refer to the total throughput index.
There are many differences between SPECjbb2005 and SPECjbb2000, so you can’t compare both benchmark results with each other. SPECjbb2005 is the successor of SPECjbb2000 and was developed by SPEC, who prepared new features including database sorting, changes to called procedure entry and new feature additions meeting the Java 5.0 standard. Moreover, SPECjbb2005 take regard of either multiple or single JVM instance, but SPECjbb2000 only runs in a single JVM instance.
PRIMEPOWER's high performance is proven by benchmark testing

PRIMEPOWER has recorded world highest performance in six major benchmark categories including SPECjbb2005 announced here.
SPARC64™ V, the heart of PRIMEPOWER high performance incorporates high performance features including super scalar
construction, Out-of-Order execution and non-blocking cache control.
PRIMEPOWER containing up to 128 SPARC64™ V processors, is one of the most scalable server families in the world.
It allows you to meet changes flexibly as the business environment changes and/or data volumes grow.
Fujitsu is continuing its fine tradition of developing servers and processors to deliver the highest possible performance
in UNIX server.
| PRIMEPOWER holds six world No.1 benchmark records (as of 2nd Feb. 2006) | ||
|---|---|---|
| SPECjbb2005 | SPECjbb2000 | SPECjAppServer2002 |
| SAP® SD two tier | SAP® ATO two tier | SAP® ISU/CCS |
Notes
- The above article is correct at time of issue.
- Benchmark records are on web page published by SPEC, SAP AG and Ideas International (3rd party organization). For the detail,
please refer to the following pages.
