Ranjith,
We have tested Simics 3.0.11 + GEMS 1.3 on both x86 and amd64-linux
hosts. Notably, in our Simics 3.x workspaces on 64-bit machines,
amd64-linux/lib is a diretory unto itself and NOT a link to
../../../x86-linux/lib or anything of the sort. I do not know if this
has changed Since Simics 3.0.11...
1) Have you set up your workspace in an unusual way?
2) Did you install and configure the amd64-linux version of Simics
3.0.X? One hackjob way of getting Simics to run on a 64-bit machine is
to make the link you are referring to.
Regards,
Dan Gibson
Ranjith Subramanian wrote:
Thanks Dan and Derin for the help.
I have another question though. Has anyone been able to run the
amd64-version of simics3.0 with gems 1.3? I notice that when ruby is
compiled with the MOSI_SMP_bcast protocol, the symlink lib in
$GEMS/simics/home/MOSI_SMP_bcast points to amd64-linux/lib which in turn
is a symlink to ../../../x86-linux/lib. But the x86-linux directory is
never created when the amd64-linux version is installed!
Ranjith
Dan Gibson wrote:
GEMS Users,
GEMS 1.3.1 is now available to address the frequent issues dealing of
Ruby's assumption about the name of Simics' physical memory object. GEMS
1.3.1 is a minor change (vs. GEMS 1.3) to Ruby only, but will prevent
some compatibility problems when using Simics 3.x, especially unexpected
segmentation faults at load-time or init-time.
The new version is available in the GEMS download directory,
http://www.cs.wisc.edu/gems/distribute/.
GEMS 1.3.1 is available as a complete tarball *OR* as a patch to GEMS
1.3. If you have made modifications to your GEMS 1.3 source, we suggest
you use the patch rather than the complete tarball.
Who should upgrade?
* Simics 3.x users should consider upgrading to GEMS 1.3.1.
* Users of both Simics 2.x and Simics 3.x would benefit from upgrading.
Who can/should ignore this upgrade?
* Users of only Simics 2.x can ignore this update.
* Users of GEMS 1.2 or earlier should ignore this update.
What files are affected?
* $GEMS/ruby/module/ruby.c: Now looks for both phys_mem0 and phys_mem
objects.
* $GEMS/ruby/simics/interface.C: Now looks for both phys_mem0 and
phys_mem objects.
Thanks to all those who helped to point out this bug, and possible
fixes. Happy simulating.
Regards,
Multifacet GEMS Devs
_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
--
http://www.cs.wisc.edu/~gibson [esc]:wq!
|