Hi, we develop software on Ubuntu, and of course, we use VMware for this, for all the usual great reasons. We make use of an API that supports Ubuntu 12.04, 14.04 and now 16.04. We decided to go for 16.04, as its new and shiny. But there appear to be serious issues with OpenGL in 16.04, which are crashing OpenGL based apps. Or, they are working, and producing terrible performance.
My setup is...
Mac Pro Late 2013
32GB Ram
AMD Fire Pro D700 / 6GB RAM
VMWare Fusion 8.1.1
And we are runing Ubuntu 16.04 in VM
Running our own app, we get a stack trace, with the final error being...
oblong::noodoo::WindowedLinuxVisiFeld::DisestablishRenderConditions(oblong::basement::Atmosphere*) () from /opt/oblong/g-speak3.24/lib/libNoodoo.so.2
The problem appearing to be with WindowedLinuxVisiFeld, which in our API is an OpenGL window that items are drawn to. This only happens as we load in Bitmap files to be drawn to OpenGL primatives. So my guess is, its bombing as it tries to make mipMaps for the BitMaps. Why? Well...
When I run glmark2 on the same machine I get, this
=======================================================
glmark2 2014.03+git20150611.fa71af2d
=======================================================
OpenGL Information
GL_VENDOR: VMware, Inc.
GL_RENDERER: Gallium 0.4 on SVGA3D; build: RELEASE; LLVM;
GL_VERSION: 3.0 Mesa 11.2.0
=======================================================
[build] use-vbo=false: FPS: 2063 FrameTime: 0.485 ms
[build] use-vbo=true: FPS: 2198 FrameTime: 0.455 ms
[texture] texture-filter=nearest: FPS: 2205 FrameTime: 0.454 ms
[texture] texture-filter=linear: FPS: 2207 FrameTime: 0.453 ms
[texture] texture-filter=mipmap:VMware: vmw_ioctl_command error Invalid argument.
Program received signal SIGABRT, Aborted.
0x00007ffff6779418 in __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:54
54 ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) bt
#0 0x00007ffff6779418 in __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:54
#1 0x00007ffff677b01a in __GI_abort () at abort.c:89
#2 0x00007ffff351a99f in ?? () from /usr/lib/x86_64-linux-gnu/dri/vmwgfx_dri.so
#3 0x00007ffff3518629 in ?? () from /usr/lib/x86_64-linux-gnu/dri/vmwgfx_dri.so
#4 0x00007ffff352003d in ?? () from /usr/lib/x86_64-linux-gnu/dri/vmwgfx_dri.so
#5 0x00007ffff312b7da in ?? () from /usr/lib/x86_64-linux-gnu/dri/vmwgfx_dri.so
#6 0x00007ffff320463a in ?? () from /usr/lib/x86_64-linux-gnu/dri/vmwgfx_dri.so
#7 0x00007ffff73f3b04 in ?? () from /usr/lib/x86_64-linux-gnu/mesa/libGL.so.1
#8 0x00007ffff73f3ee7 in ?? () from /usr/lib/x86_64-linux-gnu/mesa/libGL.so.1
#9 0x000000000050f3c2 in ?? ()
#10 0x000000000041036f in ?? ()
#11 0x00000000004121d8 in ?? ()
#12 0x0000000000406d6f in ?? ()
#13 0x00007ffff6764830 in __libc_start_main (main=0x406710, argc=1, argv=0x7fffffffdf48, init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>,
stack_end=0x7fffffffdf38) at ../csu/libc-start.c:291
A colleague has run glmark2 on a physical Ubuntu16.04 machine, and it did not crash. BUT, it did appear to have pretty awful performance, with a score of 200, compared to 2000+ on Ubuntu 14.04 VMWare. It also was throwing a LOT of errors.
Does it look like VMWare can do anything to remedy this situation?
