Hi,
We have no single problem in running SPM2 and SPM99 in MatlabR13 on
SuSE9.2Pro machine,
while SPM2/99 GUIs disappear in R14SP2. Our DICOM program (not SPM2
DICOM toolbox)
crash badly in R14SP2. :(
Cheers,
Witaya Sungkarat
> Hello,
>
>> We have not run SPM2/SPM99 on this machine yet, but we have R13 and
>> assuming
>> older kernel as fallback plans.
>
>
> Generally speaking kernels are irrelevant for end-user programs.
>
> More important is the version of the glibc library - and from time to
> time some other libraries. For example the thread library changed at
> some point (on Red Hat systems this change appeared in Red Hat 9) and
> caused problems to lots of programs. Setting LD_ASSUME_KERNEL switches
> back to the old thread library:
> http://www.mathworks.com/support/solutions/data/1-1ATCE.html
>
> Also if your current locale uses and UTF-8 encoding you may want to
> change it to a non-UTF-8 encoding. Specifically if the LANG or LC_ALL
> environment variables are set for example to
> en_US.UTF-8
> de_DE.UTF-8
> fr_FR.UTF-8
> you may want to reset them to
> en_US
> de_DE
> fr_FR
> in your Matlab startup script. At least the version of Java bundled with
> Matlab R13 seems to have problems under UTF-8 locales.
>
> Finally I would suggest upgrading from Matlab 6.5 (R13) to Matlab 6.5.2
> (R13SP2) if you haven't done so already. For example some bugs related
> to 'reshape' (but not all) seem to have been fixed in this version:
> http://www.mathworks.com/support/solutions/data/1-1A2NC.html
>
> We now run Matlab 6.1 (R12), Matlab 6.5.2 (R13SP2) and Matlab 7.0.4
> (R14SP2) on Fedora Core 2 workstations without too many problems; they
> are quite similar to SUSE 9.2 workstations. We experience some GUI
> crashes from time to time, but it's not clear whether they are specific
> to our Linux distribution. Also we see very different results between
> the same caclulations run on Windows and Linux but the issue is still
> under investigation, so again this is maybe not specific to a given
> Linux distribution. Maybe it's just a corner case that happens for
> degenerate data.
|