<b>SPEC</b> JVM Client98 Help System Requirements


SPEC intends that this benchmark suite be applicable to networked and standalone Java client computers, either with disk (e.g., PC, workstation) or without disk (e.g., network computer) executing programs in an ordinary Java platform environment. Specific system requirements are:

Java Virtual Machine

A JVM compatible with Java version 1.1 or later is required on the client system under test. These benchmarks will not run on 1.0.2 based systems. These benchmarks stress the JVM and may expose bugs in a JVM which have not previously been noted. In such case some of the benchmarks may fail to run or may give incorrect results. SPECjvm98 metrics for such systems may not be reported until the bugs are corrected and a valid run is made.

Normally a JVM compatible with Java version 1.1 or later is also required on the web server for purposes of installation only. The InstallShield installation program is looking for a JVM with version number 1.1 (or 1.1.X). If you are using a JVM version 1.2, it may complain that you don't have an up-to-date JVM. You should be able to select it anyway and continue successfully with the installation. However, in case that fails, there are two work-arounds:

  1. install the benchmarks with version 1.1 and then proceed to run them with 1.2; or
  2. install the benchmarks from the install.tar.gz archive. In this case, no JVM is required on the server.

The benchmark suite is run as an applet from a web browser or applet viewer; see Using the Benchmarks. Some of the products that have been tested with this suite are:

Memory

The benchmarks generally require a system with about 32MB of memory. Different Java implementations may require more or less memory, and different operating systems may require different amounts of additional memory. Systems with less memory may be able to run the benchmarks, albeit with some performance penalty if additional garbage collection is required. Correspondingly, systems with a great deal more memory may be able to run the benchmarks faster.

Accordingly SPEC reports results in three categories of physical memory sizes:

Disk

No local disk is required.

Processor and Speed

No particular processor or speed is required. However, your patience to wait for results may impose some practical limit on processor speed. A typical (three executions of each benchmark) run of the suite on the reference machine, a 133MHz PowerPC 604, takes four to five hours elapsed time.

Operating System

No particular type of operating system is required. However, refer the the Common Problems section of the Users Guide for a bug in this version affecting MacOS and mainframe operating systems.

Web Server

For reportable results a web server is required to store the benchmark suite, and to serve class and data files to the benchmark applet. This web server may be on another machine networked to the client system under test, or the web server may be located on the client machine (http://localhost) in which case no network is required.

For research purposes the benchmarks may be run directly from local disk either as an applet or as an application.

SMTP Mail Server

Benchmark results are sent by email which requires an SMTP mail server (see http://info.internet.isi.edu:80/in-notes/rfc/files/rfc821.txt) located somewhere on your network accessible to the client system under test. While this is normally the case for any systems connected to the Internet, it may be the case that certain test labs may not have an accessible SMTP server. On a Windows-NT Intranet, for example, you may have installed only Microsoft Exchange's own mail protocol, and may have to install the SMTP option on one system. If you are unable to access an SMTP server, there is a workaround to save results.

Graphics

Reportable results require use of the SPEC tool harness which requires a graphics display. For research purposes the benchmarks may be run from the command line in which case the tool harness will not use an interactive GUI.




error-file:tidyout.logNext topic: Using the Benchmarks