Allocating More Than 4GB

H

horrido

Bit poster
I have a Mid-2011 Mac mini with 8GB of physical memory. It runs OS X "Lion".

I recently purchased and installed Parallels Desktop 8 for Mac.

I created two VMs, one for 32-bit Windows 7 Pro and the other for 64-bit Linux Mint 14. In both cases, I cannot allocate more than 4GB (despite the warning to avoid more than 4GB). If I try to allocate more than 4GB (eg, 6GB), the OSes only recognize 4GB upon start-up.

I understand the 4GB limit recommendation, but since Parallels permits you to specify more than 4GB, I would expect the VM to honour that. Why doesn't it?

Is there something I must do in the Mac mini's bios to permit more than 4GB allocation? Is there a necessary setting in Parallels to permit this?

I know this isn't an issue with 32 bits vs 64 bits, since Linux Mint is 64-bits. This is clearly an issue between Parallels and my Mac mini (a Google search suggests that this isn't an issue with, say, the MacBook Pro).

I'd like an answer, not because I care to allocate more than 4GB, but because I'm curious. Is this an unknown bug?
 
It IS a 32 bit vs 64 bit issue, but in a different way.

* Windows 32 bit can't use more than 4 GB RAM, but that doesn't explain Linux 64 bit.

* If the OS X kernel is running in 32 bit, it can't allocate more than 4 GB to a single process (Parallels VM).

OS X defaults to 32 bit kernel on most Macs in Snow Leopard and Lion, in Mountain Lion there's only the 64 bit kernel (AFAIK).
32 bit kernel can run 64 bit apps, but can't allocate more than 4 GB to a single process.

You can check the process 'kernel_task' in Activity monitor to see what kind it is (32 or 64 bit).

There's no BIOS on Macs, there's PRAM, you may google on how to force OS X to start with the 64 bit kernel via PRAM command.

You didn't expect the answer to actually make sense, didn't you? ;)
 
Last edited by a moderator:
Wow, thanks! That's a very good explanation.

I think I'll stick with the 32-bit kernel. I only have 8GB of physical memory and I don't want honking big VMs anyway.
 
You must log in or register to reply here.
Back
Top