InitialCodeCacheSize sizing. . .
dmdabbs at gmail.com
Sun Sep 12 00:45:24 PDT 2010
In comments in c2_globals_x86.hpp indicate that
// InitialCodeCacheSize [was] derived from specjbb2000 run.
and the flag value is set to
// Integral multiple of CodeCacheExpansionSize
define_pd_global(intx, InitialCodeCacheSize, 2496*K);
define_pd_global(intx, CodeCacheExpansionSize, 64*K);
As of January 2006, according to the spec.org website,
"SPECjbb2000 has been retired and replaced by SPECjbb2005,
results submissions are no longer being accepted and
support is no longer provided."
Would the InitialCodeCacheSize default setting have changed significantly in the
(presumably) four-five years since SPECjbb2000 was used to size the CodeCache initial size?
Or, put another way, what impact on performance would an undersized CodeCache have?
Do you think there are other system or product_pd defaults that might be "dusty"?
FWIW, here are the spec.org blurbs regarding Java-related benchmarks. See http://www.spec.org/benchmarks.html#java.
SPECjAppServer2004 is designed to measure the performance of J2EE 1.3 application servers.
Includes an enhanced workload by adding a web tier, JMS, and other changes to SPECjAppServer2002.
Measures performance for Java EE 5 or later application servers, databases and supporting infrastructure.
It expands the scope of the SPECjAppServer2004 benchmark.
Evaluates the performance of servers running typical Java business applications.
JBB2005 represents an order processing application for a wholesale supplier.
Can be used to evaluate performance of hardware and software aspects of Java Virtual Machine (JVM) servers.
SPECjvm2008 is a benchmark suite for measuring performance of a Java Runtime Environment (JRE).
It contains several real life applications and benchmarks focusing on core java functionality.
The SPECjvm2008 workload mimics a variety of common general purpose application computations.
If one were to re-evaluate the setting, seems like SPECjEnterprise2010 would be the one to use.
More information about the hotspot-dev