Congratulations! You will be nominated for this years life-achievement award!
Can you save scenes and reopen them?

cheers,
thomas



On 09/29/2013 06:32 AM, Jon Swindells wrote:
just some notes from the past week or so on getting 2014 installed and working
well in Mint 15 Cinnamon.

Most of this info, if not all, is available on the webs (most of it on Stephen's blog) but, as i used the list extensively when looking for answers, i though this would be the appropriate place for my notes.

##########################################
NOTES:
update everything BEFORE proprietry driver: // wasn't obvious to my win-centric mindset
sudo apt-get update && sudo apt-get upgrade && sudo apt-get dist-upgrade
//reboot into new kernel if tehre is one
// headers
sudo apt-get install -y linux-headers-$(uname -r)

i needed cuda but if you don't, go with the os nvidia drivers.
xsi/maya worked best with 310's for me, later drivers gave me segfaults in xsi - quite
possibly due to conflicting ogl linkage.

############################  CUDA  ############################
only needed for cuda dev (ie, you only need this install if you plan on compiling your own cuda apps)

https://developer.nvidia.com/cuda-downloads
i used the 5.5 ubuntu 12.10 .run version. now i know a bit more, i think the deb package would be easier
all round. ah well. live and learn

the installer always crapped out if i didn't split/unpack it:
sh cuda*.run -extract=/path/to/extract/dir/

drivers: NVIDIA-Linux-x86_64-319.37
cuda: cuda-linux64-rel-5.5.22-16488124
samples: cuda-samples-linux-5.5.22-16488124
!!! 5.5 toolkit installs and compiles fine with gcc 4.7. no need to build or altenate 4.6!!!

//install Driver, cuda then samples

drop to a cmd shell (ctr+alt + F2)
stop x server:
sudo service mdm stop ##cinnamon/mate
## obviously ## sudo cp /etc/default/grub /etc/default/grub__bak
sudo nano /etc/default/grub
add or change GRUB_CMDLINE_LINUX_DEFAULT :
GRUB_CMDLINE_LINUX_DEFAULT="quiet nouveau.modeset=0"
Save, close nano
sudo update-grub
sudo reboot ## i blacklisted the other drivers before rebooting

at log in prompt drop to a shell again,stop the mdm and:
sudo nano /etc/modprobe.d/blacklist.conf
add to the end:
blacklist amd76x_edac
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist NVidiafb
blacklist rivatv
save and quit nano
if you had previous nvidia drivers: ## i didn't
sudo apt-get remove --purge NVidia*

install driver:
sudo sh NVIDIA-Linux-x64-319.37.run
registered with DKMS
NO to 32 bit libs ## if you want 32bit support,
 ## you'll need to do some shennanigans with xsi to
 ## stop it segfaulting (see below)
yes to run the xconfig util ## always failed for me but they still went in fine every time - permissions ??
it once asked me to remove nouveau - it failed. did it above anyway

Reboot and you should be good to go.

http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/index.html#runfile-installation
!!! pts 5 & 7 !!!
libGL.so <-- this puppy is the cause of most startup segfaults, xsi typically can't find it

cuda would only install from cmd shell with mdm stopped.

samples went in without issue every time (from desktop even)

i used ~/.bashrc to export required paths:
export PATH=$PATH:/usr/local/cuda-5.5/bin
export LD_LIBRARY_PATH=/usr/local/cuda-5.5/lib64:/lib:$LD_LIBRARY_PATH



XSI PRE-INSTALL NOTES:
deps: tcsh, alien lsb gamin ## i didn't make a note of these but you'll know instantly if you don't have the right dependencies for install
extra downloads:
##the adlm that comes with 2014 sp2 will not install for me no matter what i do
http://images.autodesk.com/adsk/files/nlm11.11_ipv4_linux64.tar.gz
tar xzvf nlm11.11_ipv4_linux64.tar.gz
use alien on the rpms
sudo alien -ci "rpmsFrom_nlm11.11_ipv4_linux64" ## include scripts and install (dkpg -i)


the setup script references /usr/lib64, doesn't exist on mint so -

ln -s /usr/lib /usr/lib64 ## or change the setup script

mental ray:
sudo mkdir /usr/tmp/
sudo chmod a+wr /usr/tmp/
## or simlink to /tmp, up to you

make sure _setup has the right permissions (note the underscore) (chmod +x )
tcsh
sudo sh setup
i used network for my server type but this:
http://wikihelp.autodesk.com/Installation_Help/enu/2014/Help/0161-Product_161/0538-Softimag538/0539-Softimag539/0540-Softimag540/0541-Installi541
leads me to believe that you could probably use a standalone edu license in 2014 ala maya

setup usually goes ok but with errors:
Error: Application/bin/cmdreg -f "Application/bin/XSICOMDLLs.lst"
Error: Registering python
the rpm errors can be ignored if you set it up previously.

you'll need to register the stuff the installer missed: (from SB's blog)
tcsh
source /usr/Softimage/Softimage_2014_SP2/.xsi_2014_SP2
cd /usr/Softimage/Softimage_2014_SP2/Application/bin/
cmdreg libdsprsr.so
ldd for any missing libs and install them
register com dlls
tcsh
cd /usr/Softimage/Softimage_2014_SP2/Application/bin/
cmdreg -f XSICOMDLLs.lst

register python
mwpython win32com/axscript/client/pyscript.py

Deleting the X11 folder (and all it's contents) gets rid of the main segfault problem
/usr/Softimage/Softimage_2014_SP2/Application/mainwin/mw/lib-amd64_linux/X11

run xsi from a c-shell
tcsh
source /usr/Softimage/Softimage_2014_SP2/.xsi_2014_SP2
/usr/Softimage/Softimage_2014_SP2/Application/bin/xsi

if you get a segfault or _some.so error here, it's usually opengl linkage
you can either set the path in your .bashrc (LD_LIBRARY_PATH=/usr/lib/nvidia:$LD_LIBRARY_PATH),
using setenv in the .xsi_2014_SP2
or symlink it.
since i changed the /lib path on the install and, in my case, the cuda exports fixed
the libGL error

bizzarely, i needed to use mayas adlm to stick my details in the .pit
/usr/autodesk/maya2014-x64/bin/adlmreg -i N 590F1 590E1 2014.0.0.F 000-0000000 /usr/Softimage/Softimage_2014_SP2/Adlm/PITConfig/SoftimageConfig.pit where 000- is your actual serial. i used my edu one, didn't seem to make a difference what you put in there as long as it was there. this irks me, confirmed that it's actually needed on my machine but i seem to be a majority of one in this instance. i'd suggest trying to run xsi
before you try registering the .pit

PROBLEMS on cinnamon:
xsi pays absoloutly no attention to the fact it's in a desktop session (even with the MWM env set).
turning off keyboard repeat (in the os) will sort the supra-key issue,
when ran maximised, it has all sorts of refresh and camera reset annoyances. regardless, when you first boot xsi up you'll have to click a few menus, create a few objects, open a few ppgs, resize the window etc until it settles down and starts behaving. !! not so much in mate (but still there to some extent) and not at all in fluxbox :)

once xsi is comfy, it's rock solid in a flakey kind of way :)

moral of the story: run xsi windowed always or in fluxbox if that annoys you
##############################################################################
--
Jon Swindells
squi...@gmail.com <mailto:squi...@gmail.com>


--------------------------
To unsubscribe: mail softimage-requ...@listproc.autodesk.com with subject 
"unsubscribe" and reply to the confirmation email.

--------------------------
To unsubscribe: mail softimage-requ...@listproc.autodesk.com with subject 
"unsubscribe" and reply to the confirmation email.

Reply via email to