Are there any benchmarks showing difference between hardware virtualisation enabled/disabled?

Question

I have a 13" sub-laptop/large-netbook, it has an AMD Athlon Neo X2 L335, and I chose this one because it supports hardware virtualisation.

In the end, I hardly do any virtualisation on it, however, when I do... it is fast.

To my shock, I went in to the BIOS and saw that virtualisation was disabled!

I turned this on and, I see no speed difference.... or at least none that I can tell.

I do not have time to do a full set of benchmarks - and I run quite a bit of software on the host, so it wouldn't be scientific. I have searched quite a few places and I just can not find any benchmarks showing the difference of virtualisation bit enabled/disabled on the same hardware. Does anyone have any benchmarks they have seen that they can share?

In addition, I know there was an uproar a while ago as Sony disable the hardware virtualisation on some models and only offer it in their higher models as a premium feature, however, apart from forcing an up-sell, are there any benefits to having it disabled e.g. battery/heat? I just can't find any information and can't work out why it would be disabled by default.

Edit---

To add, The only thing I can find is that without it, you can not perform x64 virtualisation as fast. This is the only down side I can find. However, if this is the only difference, then I am still interested in the second part of the question - why offer the option to disable it?

Answer 1

There's actually a pretty good answer to "Are there any benchmarks for virtual machines with and without VT-x? over on Server Fault. Even though the answer is about a year and half old it is still reasonable. The takeaway regarding performance is "it depends" in general AMD-V and Intel VTx do increase stability and ease development of virtual machines.

As for why the option to disable it. Some processors don't support hardware virtualization and some BIOSes when combined with those processors don't report that correctly. The ability to specifically turn off the hardware virtualization is a plus in those instances. There are probably also some concerns about stability. It is a newer technology, if a stability or performance bug crops up it's easier to tell people to disable virtualization in the BIOS then to tell everyone to update their BIOS to a version that allows them to turn it off. Finally, it's probably also there for testing. It's easier to turn things on and off in the BIOS when testing your motherboard for stability and performance with random different configurations than it is to rebuild the BIOS and load it on to test it with and without any set of features.

Answer 2

My experience (i don't remember the numbers exactly)

Windows XP (host) Windows XP (guest)

I did a windows benchmark in both (host and guest)

Windows XP (host) CPU 100% Graphics100% Math 100% Disk 100%

Windows XP (Guest VMWARE Virtualization ON) : CPU :80% Graphics 80% Math 80% Disk : 120%

Windows XP (Guest VirtualPC Virtualization ON) : CPU :70% Graphics 50% Math 75% Disk : 90%

And without virtualization the performance was about the same, but i was unable to run a 64bits guest without virtualization (including OSX). So, apparently it is about 64 bits. May be there are some difference when you are running several virtual machines at once.

My conclusion :virtualization does not change the performance, at least not for a home-usage but it is a must for 64bits, VirtualPC does not support correctly graphics acceleration and vmware speed up my disk (by fragmentation?)