xom-interest AT lists.ibiblio.org

Subject: XOM API for Processing XML with Java

List archive

[XOM-interest] Verifier performance patches

From: Wolfgang Hoschek <whoschek AT lbl.gov>
To: xom-interest AT lists.ibiblio.org
Subject: [XOM-interest] Verifier performance patches
Date: Fri, 19 Nov 2004 16:00:41 -0800

Here are some performance patches for Verifier.checkAbsoluteURIReference and Verifier.checkPCData, along the lines previously discussed: http://dsd.lbl.gov/~hoschek/xom-patches
These retain the same original behaviour, but significantly improve performance, according to profiling.

But for bnux deserialization there's still quite a difference when completely disabling both of those checks.
bnux with both Verifier checks disabled and with checks enabled:

75 MB/s vs. 55 MB/s (factbook.xml)
25 MB/s vs. 20 MB/s (soap.xml)

This is with the just released nux-1.0beta3 version.

###############
FACTBOOK.XML:
###############
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] export JAVA_OPTS='-server -agentlib:hprof=cpu=samples,depth=10 -Dnux.xom.tests.BinaryXMLTest.enablePatches=true'
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] time bin/fire-java nux.xom.tests.BinaryXMLTest deser bnux 0 1000 data/factbook.xml
patchesEnabled=true
now processing data/factbook.xml
compression factor = 1.6840668

****** SUMMARY ******
files = 1000
secs = 54.967
mean throughput MB/s = 73.26267
mean compression factor = 1.6840669
files/sec = 18.192734

[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] export JAVA_OPTS='-server -agentlib:hprof=cpu=samples,depth=10 -Dnux.xom.tests.BinaryXMLTest.enablePatches=false'
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] time bin/fire-java nux.xom.tests.BinaryXMLTest deser bnux 0 1000 data/factbook.xml
patchesEnabled=false
now processing data/factbook.xml
compression factor = 1.6840668

****** SUMMARY ******
files = 1000
secs = 72.227
mean throughput MB/s = 55.755177
mean compression factor = 1.6840669
files/sec = 13.845239
checksum = 11

###############
SOAP.XML:
###############
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] export JAVA_OPTS='-server -agentlib:hprof=cpu=samples,depth=10 -Dnux.xom.tests.BinaryXMLTest.enablePatches=true'
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] time bin/fire-java nux.xom.tests.BinaryXMLTest deser bnux 0 5000 data/soap.xml patchesEnabled=true
now processing data/soap.xml
compression factor = 3.2463272

****** SUMMARY ******
files = 5000
secs = 31.101
mean throughput MB/s = 24.663187
mean compression factor = 3.2463272
files/sec = 160.76654

[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] export JAVA_OPTS='-server -agentlib:hprof=cpu=samples,depth=10 -Dnux.xom.tests.BinaryXMLTest.enablePatches=false'
[grolsch /home/portnoy/u5/hoschek/tmp/tmp/firefish] time bin/fire-java nux.xom.tests.BinaryXMLTest deser bnux 0 5000 data/soap.xml
patchesEnabled=false
now processing data/soap.xml
compression factor = 3.2463272

****** SUMMARY ******
files = 5000
secs = 38.377
mean throughput MB/s = 19.987226
mean compression factor = 3.2463272
files/sec = 130.28638

[XOM-interest] Verifier performance patches, Wolfgang Hoschek, 11/19/2004