View Full Version : New version of Parallels

10-12-2006, 10:19 AM

Tempted to try it.

Wonder how it compares to Rosetta when running Lightwave.


10-12-2006, 12:50 PM
Parallels doesn't give you solid OGL acceleration yet, so little difference other than PC plugins work when you're emulating W2K or XP. Render threading seems a little weird, so performance isn't as good as it could be, but i know the Parallels guy are taking advantage of the processor-based thread managers.

I've fired up Parallels from time-to-time to to run PC-only stuff and it has been great compared to VPC! (I have Parallels set up both on a MacBook Pro and MacPro 3Ghz Quad.)


11-08-2006, 09:40 PM
Have you run LW in Parallels on your MBP? I'm curious if there's a difference than just running LW9 under Rosetta or Parallels.

Tonight I just installed Parallels on my wife's Macbook (not the pro) and it's great. It's kind of freaky hearing the Windows start chimes as it boots up! And it's REALLY cool/weird switching to full screen mode. The screen rotates like a giant cube, like what OS 10.4.8 is going to do.

I've really been holding off installing LW on my MBP until the UB comes out. Arrghh. So close. Once it's released, Chilton deserves a VERY well deserved break for all the work he's been doing.


11-08-2006, 09:49 PM
I'm kinda curious about the performance of the new MacBooks, myself.

11-10-2006, 02:44 AM
Would be really intrested to know the stabilility and speed of LW on Parallels, and if it's a good alternative to Bootcamp?

When you are talking about OGL acceleration would this effect the overall speed or just certain aspects like previewing or rendering?

11-10-2006, 04:21 PM
I can't see how Parallels can compete with Bootcamp when running 3d apps or any software that requires full processing power. I might be wrong, but isn't Parallels similar to running rosetta, only worse, because you are emulating an operating system instead of just the application?

In my opinion, if you need Windows for applications that are not memory intensive, then go with Parallels because it allows for easy switching. But for anything else, go with Bootcamp to take advantage of the raw power instead of OS emulation. Once virtualization becomes available then the situation will be different.

11-10-2006, 08:49 PM
You are not really emulating an operating system. We are running Parallels on a Macbook (not pro) and it runs pretty darn fast. Not sure exactly how, but since it's an Intel chip, it gives Windows access to the chip itself, not done through emulation.

I remember some benchmarks showing Bootcamp was fastest, then Parallels (not far behind) then Rosetta.

Of couse, UB would give us the BEST performance! *GRIN*

11-11-2006, 12:53 AM
I did actually get LW9 up and running under parallels. I didn't do anything hardcore with it other than render a few scenes. It's definitely an advantage over running it under Rosetta, by a large margin, at least in terms of speed for rendering. Rosetta is only giving me about 40% the overall power of the system, but rendering under Parallels is closer to 90% or even higher.

Rosetta is emulating the entire PPC architecture which is obviously a lot of overhead and really going to slow things down a lot, whereas Parallels is really doing very little work and offloading almost the entire thing to the CPU. Recent AMD and Intel CPUs support what is known as "vt" mode, for "virtualization", which is basically CPU-level support for things like VMware and Parallels. VMware will work for vt-enabled CPUs or not, but Parallels has specifically been developed since the vt-enabled CPUs have been released and won't run on anything without. vt is a way for any application to tell the CPU that it wants to run a virtual machine and to please handle all the difficult tasks like saving off of registers and securing memory areas and stuff so that the app doesn't have to do a lot of what used to be really hard work. The actual processing is sent directly to the CPU except for a few opcodes that are trapped (by the CPU with vt, or by the app without) and either emulated or translated in some way to maintain the integrity of the CPU state.

What all that means is that running under Parallels isn't emulating anything much at all, it's just telling the CPU to pretend that this is "another" computer using the same CPU. So CPU-intensive apps should run at nearly native CPU speed.

And yes, what you don't get from Parallels is any kind of OpenGL acceleration. You still get OGL, or at least it appears to work, in purely software form. So don't think you're going to be modeling a lot of high-poly objects under Parallels.

I'm much too much of a hardware nerd I think...