Parallels Hypervisor vs Apple's Hypervisor?

What are the pros/cons? Any performance benefits in either? If you use "Apple" for one VM and try to launch another with "Parallels" hypervisor, is that allowed?

The help dialog in v12 does not talk about the different Hypervisors at all..

 

I would also like to know this information.

 

Could you please provide some information on why a user would want to use the Apple hypervisor?

There must be some advantages to using it if Parallels took the time to add it as an option.

 

Apple has a hypervisor service that developers can call. It's running in user space, and doesn't require developers to write kernel extensions to use.

Personally, I've had erratic stability and performance issues that I can roughly tie to Parallels. I'm going to use it in Apple Hypervisor for a while and see how that goes.

It's quite possible that Parallels can create a Mac App Store version of Parallels, and keep not all but most of it's advantages.

 

i have ran my virtual machines with both enabled and cannot seem to tell a difference performance wise. I know that people having issues with there Vm's not running or booting into a black screen after the parallels desktop 12 update, changing the hypervisor to apple will usually solve this right away and everything goes back to working as it should. It seems to me that the apple hypervisor is at least decently stable. I also did some benchmarking with each hypervisor enabled and didnt notice a difference in the test result. Nothing to speak of anyway.

 

Just my $0.02 worth... I recently tried the Apple Hypervisor and have not noticed much difference in MacOS Sierra and Parallels Pro Version 12.0.2 (41353). Perhaps they have made improvements? I am still testing on day 2 but I run Windows 10 minimally in the background. Mainly for web testing and such. I am also monitoring my console log. So far all my errors are related to ... iCloud sigh. (I swear it seems Cloudkit spams my log more and more with every update and upgrade... I digress)

It definitely started fast. I am using an SSD with native drive access. (not quite bootcamp. I cloned a Para-VM to the drive), 6GB of RAM allocated to VM, 768 GpuRAM, and the usually sandboxing.

 

I just did a quick test today on my RMBP early '13 with i7 CPU and 8GiB RAM. I was using Ubuntu inside Parallels running some memory-demanding applications. From my observations, Parallel Hypervisor seems to be very extravagant in claiming memories, causing a large chunk of data being put into host's cache files. And it constantly drive my memory pressure into yellow zone. Then I tried the same set of programs with Apple Hypervisor. Surprisingly, the performance is quite satisfying, and memory pressure is green all the time.

Therefore, give Apple hypervisor a try.

 

Just a brief status update:

I stopped monitoring the hypervisor framework for a few months. Last week I decided to start doing more testing in other OS's.

I did more tests but am Apple all the way. Hands down it seems more efficient. I can run headless Ubuntu and it shows only 520Mb's in Activity Monitor. Hardly any CPU utilization. Though I gave the VM 8 gb of RAM, it appears to only use that which it needs. Many thumbs up.

I am using Apple Hypervisor all the way.

 

I was having major issues with the latest parallels 13 on High Sierra using Windows in my VM. I develop code in Windows Visual Studio and when running my code in the debugger the CPU would peg (400%) on the Mac side while the windows side would only be at 40%. Fans would turn on, battery life sucked etc. I turned coherence mode off, and the problems seem to have gotten better (but not great and I love coherence mode). Then while browsing the settings I found the hypervisor setting. Using the Apple hypervisor solved all my issues. I can run in coherence mode with the Apple hypervisor and my Mac side CPU usage is less than half, fans don't come on etc. I am also Apple hypervisor all the way.

 

Paul, most feedback in this thread points to people preferring/seeing advantages to the Apple hypervisor over the Parallels hypervisor. It's hard to argue with real world, final user feedback. With that in mind, does Parallels thoroughly test its hypervisor, comparing it to Apple's? Is there any documentation, test benchmarks, white-papers that you would be willing/allowed to share? As a paying user of Parallels Desktop Pro 13, I believe we are all entitled to more detailed technical information, please? Thank you in advance for considering this request.

Best,

Raphael Ferreira
Senior DBA
Governors State University

 

For months I would get random reboots (kernel panics) on my MBP '17 using Parallels 13. Very infuriating and lost a lot of work over this.
Considered getting a refund for purchasing Parallels, but thought I'd see if I could fix the problem first.
I switched to Apple hypervisor, and my problem disappeared, no more random reboots!
It's been stable and memory utilisation is improved which is a bonus!
Message to Parallels: You are developing software in the virtualisation space... surely your hypervisor should be an improvement over Apples?!

 

My performance problems happened right after I updated my MacBook Pro with the latest OS version, I do believe it's 10.14.4. It's like the processes inside the VM is using more resources than what is available. Processes that have never caused issues are now flagged as sucking up 80-400% of the CPU (yes, 400%, it depends on if I view the CPU usage from within or outside of the VM) and freezing the entire VM. They run rampant and the end result is that the entire Windows UI freezes.

From what I've read above, changing the hypervisor seems to mitigate the problem. After reading what a hypervisor does it adds fuel to my own hypothesis regarding why the processes are acting so strange in the VM.

Something changed with 10.14.4 which Parallels is not compatible with, it's not a configuration problem, it's a software engineering problem. Looking forward to and hoping for a Parallels patch that makes Parallels compatible with 10.14.4 and later.

Update:

• I've confirmed that I'm running 10.14.4 and it was that update that caused the incompatibility with Parallels.
• I changed the hypervisor from Parallels to Apple, started multiple Visual Studio instances and did a build. The result was that the VM/Windows UI did not freeze.

Parallels & Parallels Hypervisor version 14.1.2 (45479) is not compatible with Mojave 10.14.4.

 

I have just been going round in circles with brand new iMac Pro and Parallels - whereby a build in Visual Studio is 100 times faster on my MacBook Pro from 2013 than it is on my new iMac Pro - switch Hypervisor to Apple and it's far better - I am running Mojave 10.14.5 - when I'm building in Visual Studio 2017 or 2019 it just hangs on projects and then suddenly has a burst then hangs.

 

Please let us know whether you have any anti-virus program installed on your Windows.

 

No I don't have any antivirus installed - plus switching from "Parallels" to "Apple" in the hypervisor settings makes windows run as expected so I don't see why you would think that antivirus software would be the issue.

 

Running Mojave 10.14.5 (but same experience with 10.14.4) and Parallels 14.1.3 - Windows 10 VM (Build 18362) was painfully slow starting programs, especially Firefox. Today, I switched to using the Apple Hypervisor instead of the Parallels Hypervisor. Wow - everything in the Windows 10 VM is now much snappier.

And, no, my machine is not:

• Underpowered hardware - No: 2018 Mac mini, 3.2GHz-4.6GHz Core i7 (6 cores + hyperthreading), 16GB, SSD
  
• Using an outdated Parallels Desktop version or build. - No: current version (just reinstalled)
  
• Too many resources assigned to the virtual machine - No: 2 CPUs, 6GB
  
• Multiple anti-virus programs installed - No: only MS Defender in Win10 VM
  
• Too many virtual machines are running - No: Only the Win10 VM.

 

I have used the parallels hypervisor for quite some time. the other day i decided to try Apple's. As far as performance of the VM I didn't notice much difference. Applications started in pretty much the same amount of time, nothing threw any errors, etc.

However, Parallels is much better as far as resource management goes.

When first booted, my Windows 10 VM (allocated 4GB ram) took a little over 2 minutes to boot and start Visual Studio. The same VM switched to Parallels hypervisor booted nearly 30 seconds faster. In addtion, the Apple hypervisor seems to let the VM grab as much memory as it needs, whereas Parallels kept it pretty low. ( 5.5gb for Apple's HV, ~700mb for Parallel's HV).

Using Apple's Hypervisor

Using Parallel's Hypervisor:

 

Curious. In my case (Windows 10 1903) and the RAD Studio IDE for Delphi, startup times are (averaged) 30s for Apple and 40s for Parallels and memory usage is an identical 1.6GB out of 6GB. Hardware: 2018 Mac mini, six-core i7 3.2GHz, 16GB, SSD.

The 10s difference is down to how long it takes the IDE to start and load my project. Actual VM boot times are identical. VM shutdown is marginally faster by 7s with Apple.

 

The two images are the Apple, the thumbnailed image is Parallels. (No post edit facility it seems)