problem with BEAGLE

330 views
Skip to first unread message

Bena

unread,
Feb 25, 2010, 3:22:29 AM2/25/10
to beast-users
Hi beast users,
I have tried to run beast with BEAGLE on my linux computer but I
failed.
I downloaded (with svn) BEAGLE rev 700 and I compiled it for my
architecture (linux ubuntu 9.10 64 bit).
When I tested the Beagle installation with BEAST1.5.3 using java
64bit, the result was:

BEAGLE resources available:
0 : CPU
Flags: DOUBLE SINGLE ASYNCH COMPLEX LSCALE


1 : GeForce GTX 260
Global memory (MB): 1791
Clock speed (Ghz): 1.24
Number of cores: 216
Flags: DOUBLE ASYNCH COMPLEX LSCALE CPU

But actually, when I run any xml file with Beagle (benchmark.xml for
example), Beast crashes with this report:

"Using BEAGLE TreeLikelihood
Branch rate model used: strictClockBranchRates
#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f9d0cc290f2, pid=2500,
tid=140312522012944
#
# JRE version: 6.0-b16
# Java VM: OpenJDK 64-Bit Server VM (14.0-b16 mixed mode linux-amd64 )
# Distribution: Ubuntu 9.10, package 6b16-1.6.1-3ubuntu1
# Problematic frame:
# V [libjvm.so+0x3cb0f2]
#
# An error report file with more information is saved as:
# /media/dati/BEASTv1.5.3/examples/hs_err_pid2500.log
#
# If you would like to submit a bug report, please include
# instructions how to reproduce the bug and visit:
# https://bugs.launchpad.net/ubuntu/+source/openjdk-6/
#
Aborted"

I also tried to run Beast-Beagle with java 32bit but with the
following error:

"Failed to load BEAGLE library: /usr/local/lib/libhmsbeagle-jni.so: /
usr/local/lib/libhmsbeagle-jni.so: wrong ELF class: ELFCLASS64
(Possible cause: architecture word width mismatch)"

I belive that the cause of all problems is the 64 bit architecture of
my system.
Anyone can help me?

I have installed the last cuda driver and toolkit.

ramonf

unread,
Feb 25, 2010, 1:18:00 PM2/25/10
to beast-users
Well Beagle recognises your GPU card so you're quite close. You must
have had java reasonably well configured to get Beagle compiled in the
first place.

Your problems do seem very java oriented. Stick to the 64bit versions,
I'd say.

On my 64bit linux, I get beast to use beagle using this command line:

java -Dbeagle.resource.order=1 -Djava.library.path=/usr/local/lib -
cp ../lib/beast.jar:../lib/beast-beagle.jar dr.app.beast.BeastMain -
beagle benchmark.xml

this is done from within the examples directory of BEAST

Your java.library.path may differ, you need to find it. I have found
that most, but not all, xml files in this directory work with beagle
in my config.

Bena

unread,
Feb 26, 2010, 6:01:50 AM2/26/10
to beast-users
When I try to run beast with that command the problem persist.
My java.library.path is also /usr/local/lib.
I tried to run Beast/Beagle with all example files in the Beast folder
but anyone works.
Can the version of Beagle (rev 700) be the problem?

jcs

unread,
Mar 11, 2010, 2:17:59 PM3/11/10
to beast-users
It looks like you are using OpenJDK - is that true? Is there a
proprietary java distribution for Ubuntu? I'd try that out.

Bena

unread,
Mar 15, 2010, 5:28:47 AM3/15/10
to beast-users
yes,
I wrongly used OpenJDK instead proprietary java distribution.
But now, with the Sun java 1.6, I have another error:

Random number seed: 1268644577918

Parsing XML file: benchmark.xml
File encoding: UTF8
looking for plugins in/media/dati/BEASTv1.5.3/examples/plugins
Read alignment, ':alignment
Sequences = 17
Sites = 1485
Datatype = nucleotide
Site patterns 'patterns' created from positions 1-1485 of alignment
'alignment'
pattern count = 138
Creating the tree model, 'treeModel'
initial tree topology = (((((((D4Brazi82,D4Indon76),
(D4ElSal94,D4Tahiti85)),
((D4Philip84,D4Thai84),D4NewCal81)),D4Tahiti79),
(D4ElSal83,D4Thai78)),D4PRico86),((((D4Indon77,D4Philip64),D4SLanka78),
(D4Philip56,D4Thai63)),D4Mexico84))
tree height = 1104.8884441566972
Using strict molecular clock model.
Creating state frequencies model: Using empirical frequencies from
data = {0,30751, 0,18768, 0,27245, 0,23236}
Creating HKY substitution model. Initial kappa = 1.0
diff = 1.0E-10 freq = 1.0E-10
Creating site model:
4 category discrete gamma with initial shape = 0.5


Using BEAGLE TreeLikelihood
Branch rate model used: strictClockBranchRates

Error running file: benchmark.xml
Fatal exception: null
java.lang.NullPointerException
at beagle.BeagleJNIWrapper.createInstance(Native Method)
at beagle.BeagleJNIImpl.<init>(Unknown Source)
at beagle.BeagleFactory.loadBeagleInstance(Unknown Source)
at
dr.app.beagle.evomodel.treelikelihood.BeagleTreeLikelihood.<init>(Unknown
Source)
at
dr.app.beagle.evomodel.parsers.TreeLikelihoodParser.parseXMLObject(Unknown
Source)
at dr.xml.AbstractXMLObjectParser.parseXMLObject(Unknown Source)
at dr.xml.XMLParser.convert(Unknown Source)
at dr.xml.XMLParser.convert(Unknown Source)
at dr.xml.XMLParser.parse(Unknown Source)
at dr.app.beast.BeastMain.<init>(Unknown Source)
at dr.app.beast.BeastMain.main(Unknown Source)

I launch BEAST with /usr/lib/jvm/java-6-sun-1.6.0.15/bin/java -


Dbeagle.resource.order=1 -Djava.library.path=/usr/local/lib -cp ../lib/
beast.jar:../lib/beast-beagle.jar dr.app.beast.BeastMain -beagle

benchmark.xml .
I also tried to set beagle_GPU or beagle_CPU but Beast doesn't work...
Any help is appreciated...

Thanks!


On 11 Mar, 20:17, jcs <conrad.st...@gmail.com> wrote:
> It looks like you are using OpenJDK - is that true?   Is there a
> proprietary java distribution for Ubuntu?  I'd try that out.
>
> On Feb 26, 6:01 am, Bena <bnz...@unife.it> wrote:
>
> > When I try to run beast with that command theproblempersist.
> > My java.library.path is also /usr/local/lib.

> > I tried to run Beast/Beaglewith all example files in the Beast folder


> > but anyone works.
> > Can the version ofBeagle(rev 700) be theproblem?
>
> > On Feb 25, 7:18 pm, ramonf <ramonfal...@gmail.com> wrote:
>

> > > WellBeaglerecognises your GPU card so you're quite close. You must
> > > have had java reasonably well configured to getBeaglecompiled in the


> > > first place.
>
> > > Your problems do seem very java oriented. Stick to the 64bit versions,
> > > I'd say.
>

> > > On my 64bit linux, I get beast to usebeagleusing this command line:

jcs

unread,
Mar 15, 2010, 5:39:59 PM3/15/10
to beast-users
Have you tried re-building BEAST using the command below?

ant -f build.xml dist_all_BEAST

and make sure JAVA_HOME is pointing to the correct 64-bit and
propriety version of the JDK.

Bena

unread,
Mar 16, 2010, 6:32:37 AM3/16/10
to beast-users
I have downloaded the last version of Beast (1.6.0) with svn and built
with ant.
Now I can use Beast with beagle! I tested all beagle options without
problems.

The last question, my graphic board is a nvidia gtx 260 and the nvidia
site indicates that
my board can compute in double precision but Beast allows only single
precision with my GPU.
Is it normal?

Thanks a lot.

> >         atbeagle.BeagleFactory.loadBeagleInstance(Unknown Source)

Andrew Rambaut

unread,
Mar 16, 2010, 11:51:14 PM3/16/10
to bnz...@unife.it, beast-users
BEAGLE is work in progress so, as you have discovered, the latest version of BEAGLE will likely only work with the latest version of BEAST. By latest version, I mean the current source code in the repository. The BEAGLE repository has a version tagged 'BeforeCzech2010' which will work with the current released binary BEAST (v1.5.3).

Finally, to answer your question, although many NVidia boards support double precision, they are very much slower in this mode and at present we haven't implemented double precision for these boards. The next generation of board (Fermi) promises considerable improvements in the performance of double precision so we will review this when these are available. On the other hand, single precision is always likely to be faster so we are concentrating on maintaining numerical stability in single precision.

Andrew

> --
> You received this message because you are subscribed to the Google Groups "beast-users" group.
> To post to this group, send email to beast...@googlegroups.com.
> To unsubscribe from this group, send email to beast-users...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/beast-users?hl=en.
>
>

___________________________________________________________________
Andrew Rambaut
Institute of Evolutionary Biology University of Edinburgh
Ashworth Laboratories Edinburgh EH9 3JT
EMAIL - a.ra...@ed.ac.uk TEL - +44 131 6508624

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Reply all
Reply to author
Forward
0 new messages