Can Parallels use more RAM than the Maximum?

Is it possible to install say 4 Gigs of RAM in a MacBook and use the 2 extra Gigs for Parallels only?

The reason I ask is currently the MacBook can only use up to a maximum of 2GB, however there are paired memory modules from 3rd party vendors that one could install which contain 2GB on each memory card.

Hence my curiosity about the possibility of using one 2GB memory card for the Mac side and the other 2GB that Mac can't see for the windows side?

Is there any logic in this approach, has it been tried?


If possible, this could make for some VERY fast VM's.

 

If 2 gig of RAM is unavailable to OS X, and Parallels is an OS X application, how do you think it is going to be able to access that hidden RAM?

 

Parallels isn't really an OS X application in the traditional sense as I understand it. Parallels is a hypervisor with a nice GUI and setup features which allows Windows (or another OS) to run concurrently with OS X on the same hardware.

The answer to your question is that I don't know. What I do know is there are a lot of really smart people out there who like to think about this kind of stuff and try to find solutions to problems like this. It would be interesting and usefull if something like this could be done.

 

The 2GB limit on the MacBook is a chipset limitation, like the 3GB limitation on MacBook Pros (now 4GB). You can use 2 - 2GB memory modules, but because of the way the chipset works on that particular Mac, you can only use 2GB. I believe it would even show 4GB installed, but only 2GB available.

 

A little OT but VMWare has 'memory overcommit' which means you can assign more mem to the vm than exists on your system. It's swapped out a page file. (Don't know if it exists on Fusion, but it's on workstation.)

The net gain is that you can assign 2 gigs to your VM when you only have 2 gigs on your system. The host can lock the mem that it needs and the guest looks like the mem is committed to another process. Close an app on the host and you have more free ram on the guest. It allows you to really share the RAM nicely between the two environments.

 

I see real issues with that. If I have 2GB in my system and assign it all to the vm, and the rest is paging to disk, what happens when everything is running and windows is trying to access it's own page file? Seems like this could bog down things to an unacceptable level.

 

There could be issues and VMWare recommends turning the feature off for better performance, but it could be desirable.

I haven't worked much with the feature but this is how I think it works:

Assign 2 GB of RAM to the guest. When VMWare tools loads, it locks the amount of memory that the host is using, so even though the guest is assigned 2 gb the guest only sees say 1 gb as free. As you launch applications in the host, it locks the same amount of memory in the guest. As you launch applications in the guest, VMWare locks the same amount of memory in the host.

So now you don't need to 'decide' how much RAM to give your guest. It uses what it needs, the host uses what it needs and VMWare notifies the other as to how much is available total.

I've suggested this as a great feature to the parallels folks before, but nothing happened.

 

Yeah, but I wonder if the hyperthread could be tweaked to see the extra ram independent of the Mac OS?

I know this is probably not possible, but stranger things have been done.