I will soon be using Rhino 4 in PD. I currently have a Windows XP guest with 128MB 3D hardware acceleration on a MBP 2.2GHz 4GB 10.6.2 setup and am considering upgrading from Parallels Desktop 4 to 5. The upgrade pages claims improved OpenGL support, but I do not see any improvements. OpenGL Extensions Viewer http://www.realtech-vr.com/glview/ returns these results: Mac Native Fullscreen: 1.1-2.0: ~1300fps, 2.1: 120fps Parallels 4 Fullscreen: 1.1-2.1: 30fps Parallels 5 Fullscreen: 1.1-2.1: 30fps Am I reading the v5 notes wrong? Are the improvements limited to Windows Vista or 7? Anyone else use Rhino in any Parallels configuration? How is the performance?
No response from Parallels, so I decided to do more testing on my own. The above results were with Windows XP. Here are the same tests with Windows 7: Parallels 4 Fullscreen: 1.1-2.1: 24-27fps Parallels 5 Fullscreen: 1.1-2.0: 35-37fps, 2.1: 21 These are with a GeForce 8600 GT at 1386x878 (1440x900 less menu & left dock) in coherence view with tools installed. So, it appears there is a minor improvement using Win7 from 4 to 5, but insignificant compared to the native capability, and a decrease in OpenGL 2.1 performance. Looks like I will be sticking with Parallels 4, unless anyone else can provide more information, or settings that can improve the OpenGL performance.
EEFranklin, unfortunately Coherence mode is not the fastest case for 3D apps yet. I'd suggest you do your measurements in Full Screen given that you're going to use Rhino which takes full control of mouse pointer in some operations. But you'll likely see no difference between 4 and 5 in full screen because the frame rate will be limited by 60Hz refresh (which is forced for GL). It's possible to turn it off, but doing so will bring no real benefit except larger numbers in GLView. On the other hand it's always advisable to check performance of real world applications (Rhino in your case) than extrapolating from a benchmark that measures something different. Rhino draws a lot of geometry while GLView drops a single cube, thus those crazy numbers in OS X. I don't have much to say about numbers you've got except that you have to be very carefull when measuring because secondary factors (like background process activity and overall system state) affect results significantly. Multiple measures must be made and basic statistical analysis applied to validate the results. Needless to say that averaging results over different test scenarios must be avoided as non-scientific. We'll double check GLView results in Coherence internally to make sure there's no degradation.
Thanks for the tip about Full Screen view. With Parallels 4 and WinXP in Full Screen, the results were about 60fps, which is plenty adequate. Rhino is responsive enough in the simple tasks I have tried so far. I realize my GLView results were basic and unscientific, but is freely available and produces an easy number to compare. My main purpose was to see how much improvement there was in PD 5 to determine if I should upgrade. Though I should be fine just using Full Screen view in PD 4. Thanks.
Hi EEFranklin, Just as a thought, there's a native OSX version of Rhino that's under development. I don't know when it's due for release (they haven't published any dates yet), but there is a free beta that you can sign up for, I haven't tried it myself but according to their website it should be pretty usable at this stage. If you google on Rhino OSX you should be able to find it. BTW - I don't work for whoever produces Rhino, just thought it could be an option for you.
Thanks for the tip about Rhino OSX. I will try out their beta, though I still need plugins that are currently only available for the Windows version of Rhino.
