Fluent launching from Workbench 19.2

[Joshua]

Hello,

I am having the same issues as were reported here for Fluent 16.2

https://sourceforge.net/p/virtualgl/mailman/message/34859603/

I can launch the standalone version of Fluent 19.2 and get the VGL_LOGO to appear with VGL_LOGO, but when launching from within workbench, I am not able to get it to work.  In general you get "Warning: Cannot find accelerated drivers.  Performance might be severely impacted."  warning, but then you can set "setenv FLUENT_WB_OPTIONAL_ARGS "-driver opengl".  However, this appears to not use hardware rendering and the VGL_LOGO does not appear.

All the other applications (Mechanical, Meshing, Design Modeler) launch from workbench with the VGL_LOGO, and VGL_VERBOSE shows VirtualGL being called, so these seem to be working correctly. 

I am using TigerVNC + VirtualGL.

All the glxspheres64 and glxgears, etc. tests work fine and the VGL_LOGO appear in those and VGL_VERBOSE shows that VirtualGL is being called.

Has anyone had any luck with Fluent from Workbench since v16.2?

Thanks.

Josh 

  

[Joshua]

If the VGL_LOGO appears using TigerVNC, does that mean hardware acceleration is working, or would the VGL_LOGO appear even if it was running using the GLX extensions, unaccelerated?

[DRC]

If you see the VGL logo, that means that hardware acceleration is
working. The logo is drawn by VirtualGL after the OpenGL-rendered
pixels are read back from the GPU and before they are drawn into the X
window, so it is guaranteed that, if you see the logo, the underlying
pixels were rendered by the GPU.

This has been a long-standing problem, and I'm not aware that anyone has
discovered a workaround. I'll repeat what I said in the message on the
old SF Mailing list: if ANSYS is claiming that they "support" VirtualGL
as a platform, then it's their responsibility to fix this or suggest how
I can work around it in VGL. Given that other ANSYS applications work
properly, I strongly suspect that this isn't my problem. At this point,
VirtualGL is 14 years old, and I have bent over backwards to accommodate
enterprise ISVs and their wonky software, but it would be nice if they
made even one step in my direction for a change. It would be even nicer
if companies like ANSYS stepped forward to fund my labor in supporting
their applications, given that they're claiming that they "support"
VirtualGL. Otherwise, I'm flat broke from giving away too much of my
work to the open source community for free. Can't do it anymore.

DRC

> --
> You received this message because you are subscribed to the Google
> Groups "VirtualGL User Discussion/Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to virtualgl-use...@googlegroups.com
> <mailto:virtualgl-use...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/virtualgl-users/6b00c485-4386-41e8-b8f5-e0b68f336378%40googlegroups.com
> <https://groups.google.com/d/msgid/virtualgl-users/6b00c485-4386-41e8-b8f5-e0b68f336378%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Joshua Moore]

Thank you DRC. I forwarded this thread to ANSYS and the previous one from SF. They haven’t proven to me that Fluent works with VirtualGL the way they say it does, but maybe we will get somewhere. We really just need to know how Workbench calls Fluent

Thanks for answering my questions.

Sent from my iPhone

> To unsubscribe from this group and stop receiving emails from it, send an email to virtualgl-use...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/virtualgl-users/5abac8de-9cbe-b96c-f420-004e7326a901%40virtualgl.org.

[Joshua Moore]

Hello,

Apparently there is an environmental variable that runs before Fluent.  It's named CORTEX_PRE

If you set this as an environmental variable, this will force vglrun to redirect Fluent when called from Workbench.  Now the VGL_LOGO appears when opening Fluent from workbench, and it also shows as being executed on the GPU using "nvidia-smi".

setenv CORTEX_PRE "/bin/vglrun"   

You may want to put the full path to vglrun just in case.

I am using modules, so added it to my ansys/192 module file

#%Module1.0
proc ModulesHelp { } {
        global version prefix

        puts stderr "\tANSYS 19.2"
}

module-whatis   "ANSYS 19.2"

setenv FLUENT_WB_OPTIONAL_ARGS "-driver opengl"
setenv CORTEX_PRE "vglrun"

prepend-path PATH /ansys_inc/v192/bin
prepend-path PATH /ansys_inc/v192/CFX/bin
prepend-path PATH /ansys_inc/v192/fluent/bin
prepend-path PATH /ansys_inc/v192/icemcfd/linux64_amd/bin
prepend-path PATH /ansys_inc/v192/Framework/bin/Linux64
prepend-path PATH /ansys_inc/v192/TurboGrid/bin
prepend-path PATH /ansys_inc/shared_files/licensing/lic_admin

prepend-path AWP_ROOT192 /ansys_inc/v192
prepend-path ANSYS192_DIR /ansys_inc/v192/ansys
prepend-path ANSYSLIC_DIR /ansys_inc/shared_files/licensing

prepend-path LD_LIBRARY_PATH /usr/lib64/nvidia
prepend-path LD_LIBRARY_PATH /ansys_inc/v192/Framework/Linux64/bin

Thanks.

Josh

To unsubscribe from this group and stop receiving emails from it, send an email to virtualgl-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/virtualgl-users/5abac8de-9cbe-b96c-f420-004e7326a901%40virtualgl.org.