Discuss VirtualBox 3.0 Release

[petrossa]

version 3.02. On host Vista SP2 x64 guest windows are correctly sized, Windows 7 7600 x64, same hardware quest windows are fixed size and scrollbars are visible.

[bitti]

Upgraded to 3.0.2

Install of windows xp in a VM with 2 CPUs took 5x longer than installing in a VM with 1 CPU.

With 2 CPUs enabled, it dogged my machine so bad that seamless windows took several seconds to render when dragged around.

I have a similar experience when I upgraded from 2.x.x to 3.0.2. My host is Opensuse 11.1 64-bit, and I'm running Windows XP 32-bit guest on a Intel Core 2 processor.

Running my old virtual machine (from 2.x.x) works as before, but utilises naturally only 1 CPU. When I create a new virtual machine with 2 CPUs, it is unbearably slow. Installing Windows XP and its updates took hours, and running the virtual machine after it is equally slow. Later editing virtual machine settings to allow only 1 CPU changes nothing in this respect, the virtual machine is still slow.

Host CPU utilisation drops when the guest is idle, but when the guest is actually doing something, both CPU cores show almost 100 % most of the time.

[carfield]

Version 3.0.2 work good for me, all problem gone, thx

[mykes]

Once windows (during install) sniffs the 2 CPUs, it installs its SMP kernel and you're SOL with respect to the poor performance of vbox.

It did boot up the install CD reasonably fast, but once it got into the install, it dogged.

Upgraded to 3.0.2

Install of windows xp in a VM with 2 CPUs took 5x longer than installing in a VM with 1 CPU.

With 2 CPUs enabled, it dogged my machine so bad that seamless windows took several seconds to render when dragged around.

I have a similar experience when I upgraded from 2.x.x to 3.0.2. My host is Opensuse 11.1 64-bit, and I'm running Windows XP 32-bit guest on a Intel Core 2 processor.

Running my old virtual machine (from 2.x.x) works as before, but utilises naturally only 1 CPU. When I create a new virtual machine with 2 CPUs, it is unbearably slow. Installing Windows XP and its updates took hours, and running the virtual machine after it is equally slow. Later editing virtual machine settings to allow only 1 CPU changes nothing in this respect, the virtual machine is still slow.

Host CPU utilisation drops when the guest is idle, but when the guest is actually doing something, both CPU cores show almost 100 % most of the time.

[bobdevis]

I had slowness and lockups with 2 cores with my Linux guests, also in 3.0.2.
It seems it's better to assume the SMP support in VBox is experimental/broken at this point.

[SSCBrian]

I had slowness and lockups with 2 cores with my Linux guests, also in 3.0.2.
It seems it's better to assume the SMP support in VBox is experimental/broken at this point.

Which guests?

So far, I've had very good SMP results with the latest Fedora and SuSE.

[bobdevis]

Which guests?
So far, I've had very good SMP results with the latest Fedora and SuSE.

Ubuntu 8.10 mostly. It runs fine and takes advantage of the extra core but tends to lock up and takes the whole VBox UI with it so I have to kill the guest with the killall command from the console (this is on a Ubuntu 9.04 host). I don't wanna get in to it too much because I haven't tested it good enough to make any useful bug hunting comments.
I have been more into testing the Direct3D in this thread: http://forums.virtualbox.org/viewtopic.php?f=2&t=19789

[Two]

Installed 3.0.2 and created a new WinXP32-box from scratch (WinXp32 host with Intel DualCore, default settings with SATA controller and SB16) was running fine in the beginning. After a while of testing I noticed the following issues:

- If running with 2 cores caused the Visual Studio 2008 C++ compiler to crash frequently at random places/files. Rerunning it caused it to either complete the code or crash somewhere else.
- After a while of working with the Box (1 CPU, programming and compiling), my host system reports "Out of resources", the VBox reports a "Delayed write" issue and about one or two minutes later the system bluescreens with "A kernel thread was terminated while waiting for a mutex". Letting VBox idle for 3 days did not cause any issues.

Anyone have any ideas on what might be wrong here?

[teddy]

Please understand that when using VBox on a dual core you should not enable SMP want more processing power. To use SMP for the reason of having multiple cores for sharing the load you
should follow an easy rule. actual cores = virtual cores + 1, turn this around and you get virtual cores = actual cores - 1. You should have at least one free core to run your underlying OS and
other stuff outside the VM. The Vbox team should probably have a warning or something, but on the other hand the multiple virtual core feature could be used for development of SMP enabled
applications and in the development phase speed isn't always necessary.

[BubbatheComputerGuy]

Should I be able to standby/hibernate my host computer while running a virtual machine?

Would it be normal for the screen to go black, but the power light stay on, and hard drive light stay active? Then when I attempt to turn the host back on, it comes to "Windows XP Preparing to Standby..." and is frozen. The only thing apparent to do then is hold the power button and cut off the host.

Thank you.

[BubbatheComputerGuy]

If I just downloaded the 3.0 binary, which option should I pick as my version on my profile here? Thank you.

[vbox4me2]

Should I be able to standby/hibernate my host computer while running a virtual machine?

This is not recommended, savestate of shutdown Guest when pausing the Host.

If you have any specific questions open a new topic, this one is for specific 3.0 issues such as comparing with previous versions.

[tlu]

To use SMP for the reason of having multiple cores for sharing the load you
should follow an easy rule. actual cores = virtual cores + 1, turn this around and you get virtual cores = actual cores - 1. You should have at least one free core to run your underlying OS and other stuff outside the VM.

Aha - if your logic is correct this would mean that you couldn't use a VM without a multi-core CPU at all

[SSCBrian]

Should I be able to standby/hibernate my host computer while running a virtual machine?

Yes, that functionality was added/fixed back in a previous version (2.x series I believe?). I've not had any problems with either standby or hibernate since (Windows hosts). What host OS are you using?

[SSCBrian]

To use SMP for the reason of having multiple cores for sharing the load you
should follow an easy rule. actual cores = virtual cores + 1, turn this around and you get virtual cores = actual cores - 1. You should have at least one free core to run your underlying OS and other stuff outside the VM.

Aha - if your logic is correct this would mean that you couldn't use a VM without a multi-core CPU at all

Yes, there seems to be a misunderstanding that cores are "allocated" to certain things. They are, but only very briefly during processing. There's no problem having a dual-core machine, with 2 VMs open, each of which is configured as dual processor. If there's work to be done, they can use 100% of both. If there's not work to be done, then they're not being used (with rare exceptions such as specific OS that use idle loops, rather than actually halting).