virtualbox.org

michaln wrote:

skoehler wrote:I powered the VM off and then fired it up again each time to measure the startup time. So yes, it must have been a different issue.

OK, thanks for the confirmation. Just for reference, how exactly do you define "startup time"?

Way ahead of you. My posting included a precise definition of startup time. Anyhow, here is the defintion:
I measured the time from clicking the Start button in the VirtualBox GUI instead I could see the Windows desktop of the user (auto-login is enabled in the guest, so XP boots right through to the desktop of the main user account).

skoehler wrote:Now I'm terribly curious: did XP run fast with 4 CPUs in Parallels?

Yep, it ran full-speed, although I do not remember if windows did indeed see multiple CPUs; I suspect it did, 80% certain.

skoehler wrote:What exactly from that thread did you try?

Tried the fix, as in the first post in the thread.

thanar wrote:

skoehler wrote:What exactly from that thread did you try?

Tried the fix, as in the first post in the thread.

That cannot be considered a fix, if you want multiple CPUs inside the guest. The ACPI-PC HAL (the guy didn't change the processor driver, he changed the HAL) does not support multiple CPUs. Neither does it support I/O-APIC. Hence, the guest will use only one CPU and no I/O-APIC. In result, Windows will run faster then with multiple CPUs or a HAL that uses the I/O-APIC (at least for I/O-heavy loads). But then you're basiscally simply running a single CPU guest.

skoehler wrote:But then you're basiscally simply running a single CPU guest.

Thank you for confirming my findings!!!

Fixing the extremely poor performance is still considered a fix for me; switching to single-CPU under VirtualBox then.

I checked again the XP VM in Parallels just now and it still uses 4 CPUs with the HAL driver that makes the same VM under VirtualBox sluggish.

Creating a new XP (no sp) VM with I/O APIC disabled and VT-x/AMD-V and Nested Paging enabled.

thanar wrote:Fixing the extremely poor performance is still considered a fix for me

Sure thing... unless you actually need to run a multi-threaded, multi-core aware, CPU hungry, number-crunching program. And it's not even close to be called a fix. Not even a workaround.

The new XP (no sp) VM set up with 1 CPU (with I/O APIC disabled and VT-x/AMD-V and Nested Paging enabled) and most other settings default is BLAZING FAST. XP uses ACPI-PC HAL. I need the VM to run 2D CAD applications, so I don't think I need extra CPUs. Will stick with this one VM for now, unless there's a new 4.3.x version out in the next couple of days that fixes this issue. Have to check the beta builds I guess.

Just had a look on VirtualBox's bug traker and I did not see a related bug, seems strange to me. Am I missing something?

For what it is worth, my experience is similar to others...VB version 4.2.18...fresh install of WinXP SP3 is horribly slow while installing updates with SMP enabled (2 CPUs), but when I drop back to 1 CPU the system flies