The latest update of Parallels Desktop 3.0 has improved the resources usage significantly, but it still is eating a lot more resources in the same conditions than its competitor, VMWare Fusion. This is particularly important for laptop users, as the less resources Parallels use, the longer the battery life is when using it.
I can't confirm what you're saying. There is no evidence that VMWare Fusion is using less resources on my MacBook Pro than Parallels Desktop 3 build 5160 is. In what circumstances do you see a difference of that order of magnitude ("a lot more resources in the same conditions"). What are those conditions?
I agree with Oliver. Your conclusion may or may not be true, but your logic is bordering stupidity, and any self-respecting person should not even consider your results. What you said is like saying: The latest Lamborghini Gallardo goes faster than the Diablo. But, the Ferrari goes faster. Well, maybe it is true, maybe it isn't.
Just look the CPU usage. Ok I exaggerated by saying "a lot" more, Parallels eats more resources, but not a lot more. Also at least one computer magazine agree with me, "Vous et votre Mac" (French magazine), as they state also in their Parallels VS Fusion review that Parallels use more CPU usage. For example, and just like them, I observe that when doing nothing on Windows, Parallels eat up to 7-10% of CPU (while processus in Windows only use 1-3% in total usage (according to CPU usage monitor of Windows), while Fusion only eat about 1-3%. In a nutshell Parallels is using about 7-8% more CPU than what the Windows apps total usage is (according to Windows CPU usage monitor), while Fusion use almost the same (1-2% more maximum) as what Windows apps total usage is.
On my computer, I have two different virtual machines. They have the same disk size, the same windows XP installed, the same set of software, the same XP settings. One is a Parallels VM and the other one a VMWare Fusion. Both show similar host cpu usage when idle (from 7 to 11 % for parallels, from 7 to 11% for Fusion). Depending on weird factors (speed of wind and moon phase), I sometimes get Parallels around 8 and Fusion around 10 or the reverse, so according to my tests under my circumstances, they are totally on par regarding host cpu useage. Please check the publication date of that copy and the date at which they tested, so what build they tested. Both products evolve and paper magazine have long delays between test, press and availability in the book stores. Unless proved different, I am convinced they are right: versions of Parallels before 3.0 build 5160 were using more host cpu (10% more easily) than Fusion when a typical XP VM was mainly "idle". This is not true today, at least for what I can see on my machine with my VMs. You should remove Windows cpu useage as reported inside Windows from your equations. It will drive you nowhere. What Windows measures is unmaterial but virtualized. It's mostly correct if you consider that a percent in Windows is a percent of the 100 percent it can get from the host. But those 100% are not 100% of the host cpu. I'm not sure I make myself clear, but I think you get the picture. The only thing significant is that you say your idle Parallels VM uses somewhere 7 to 10%, while your Fusion VM uses 1 to 3%, as reported by Mac OS X activity monitor. Oh, then, yes, it looks like on your configuration Fusion uses some percent less host cpu when the VM is mainly idle. I'm so happy for you. Yet it doesn't generalize. Slight differences in the settings of both your VMs could account for this. But then what does it change in the end of the day? I never could get Fusion either configured for 2 cpus or for a single one to beat my Parallels VM when doing my typical C++ build (best I could get out of Fusion was 9 minutes and 15 seconds - with a single cpu - with two it was more than 11 minutes, go figure). I get that same build done in 6 min 45 to 7 min 30 routinely everyday. My conclusion on my machine for what I am doing: Parallels is the performance winner for me (even if it is not a terrible difference around 2 min). It might not compare to other's results, but these are mine and so are the only results which count to my work. If I could tweak Fusion to use only 1 to 3% host cpu with an idle VM, it would change nothing for me, as I'm doing nothing with an idle VM anyway. ;-) Maybe the only conclusion is that both competitors are somewhat stable enough for real work to be done, are close to each other regarding performance and host resource useage. Maybe one or the other might be slightly better depending on the kind of tasks it is used for. What remains then is a matter of personal liking regarding the features and 'gadgets' sets. Why do I have both systems configured? Because at some point in time I took the opportunity to buy a Fusion license when it was offered for a time-limited heavily reduced price. It was tempting to get both systems and have time to decide which way to go. Having nothing real to gain from switching, I kept my main VMs on Parallels, but continues to keep a close eye on Fusion as I'm a very old-time VMw customer on other systems than Apple computers.
Your exaggeration is not pertinent to our criticism. So with my analogy, Car and Driver says the Ferrari is faster, but does not reference if it is a top speed, speed on windy road, or speed on track, or average speed on a Le Mans. I ask how? Is it really? Did they test it on a straight-away? Windy road? Windows apps total usage? What apps, I thought you said you were running nothing? Furthermore, hopefully you aren't idling your VM the whole time... Hopefully you actually use it. It would be more useful if you actually told us usage numbers while doing set tasks in programs, etc. Take a page out of Oliver's book when you are evaluating things. He did it the right way, he timed what he actually did usefully with the VM. For the record, my idle is just a couple percent, as it seems that is what you care about...
It's impossible to have nothing running, as the OS itself at least is running and start already a bunch of processes.
It's the latest, October 2007, and it's Parallels 3.0 build 5160 vs Fusion 1.0. Note that if you would think they are biased, think again, because they conclude that Parallels is preferred for them as it has better integration features. So they prefer Parallels, but the cons they give are: Resources usage, No 64 bits support, No multicore support.
Well, technically, the very act of interfacing with or turning the hardware on is doing something... And as you point out impossible. Figurative and generally understood use of the word nothing. i.e. just the OS.
It would change a lot to me. I have Parallels running a lot in the background, and I do feel that it slows down my computer considerably. This is probably mainly because of the huge amount of RAM required by virtual machines in general, but I'm sure that processor usage also adds to the slowdown. Also, as iFrodo already pointed out, processor usage is especially important to laptop users: High resource usage means shorter battery life. Another problem I have with parallels is that it loads all it's drivers right after logging in. I don't think that's a good idea. The drivers should be loaded only when Parallels starts, not when I start the computer. The time right after logging in is the time I definitely want to use: I don't want to look at my desktop for 10 seconds and wait for my computer to load all the parallels crap -- I want to start using it right away!
Only happens when using USB for me. Which sucks, cause I use a USB Mic (better gain 7 sound quality for my use).
What build are you running? Do you have the USB set to auto-connect, there were some leaks with the auto-connect, I wouldn't know now as I never set-up auto-connect anymore, I always set it to prompt the user.
