Hi, I'm looking for some feedback from others with a similar setup to myself, to judge whether I am getting the performance that I should be.
I am running a black Macbook (original, Core Duo) 2Ghz, 2Gb RAM
I have no other apps running in OSX
I am using Beta 4 of Fusion
My VM is a Win XP SP2 32bit
The VM is set to single processor.
The VM is set to use 1Gb of the available 2Gb.
The VM is using a SCSI disk (cloned from the original IDE disk)
With no apps (discounting a couple of small services) running in the VM, following bootup, Activity Monitor is OS X is showing around 25% CPU utilization.
This sounds high compared to what I have read elsewhere on the forum...
Thoughts?
Thanks in advance,
If you're creating a new VM based off an existing vmdk, Fusion doesn't check if that vmdk is bootable. You can get around the CD check by deselecting "Start virtual machine and install operating system now".
This does sound high (I'd consider ~5% normal for most idle guests), though you didn't say if you checked if the guest is truly idle.
Do you have any USB (or other) devices connected?
Hmm thats what Im reading elsewhere,
Yes the guest is truly idle (ive stopped all non windows essential services also), no apps are running, VM is completely booted already, all startup disk activity is finished.
puzzling....I might try a fresh install of XP in a new SCSI VM just to compare and contrast.
Thanks again etung.
Ok, not the best of news - a new VM created via easy install of XP sp2, configured in exactly the same way (memory, processor etc) gives around 6% utilization through Activity Monitor....
Any ideas VMware gurus?
Correct, that is the only difference, other than how the machines were created.
But.............. no non-windows software is running in either VM.
In both VM's the services running are identical.
Windows Task Manager should 0% to 2% CPU usage (system idle process takes 98%)
Hmm, where did the original VM come from - was it a Parallels VM? Parallels uses a non-ACPI HAL, while Fusion uses (by default) an ACPI HAL.
Could you take the .vmx file from both VMs and compare them to see what settings might be different?
The original was created either in Workstation 5.5.3 or in Fusion beta 1 - cant quite remember - is there a way to find out?
If you haven't upgraded the virtual hardware, WS5 uses version 4 while Fusion (and WS6) uses version 6.
Actually - I can recall - it was a WS created VM, as I remember running the 'Upgrade Virtual Machine' command.
Furthermore, the directory contents of the VM are not in a package, as they are in the new VM.
Do you think this difference matters?
I also notice that the original VM has many Virtual Disk files, most of which have not been used in months.
I did delete the old IDE disk from the setting window, so I would have expected these older disk files to have gone. are they still used somehow? Can I delete them?
I have attached a screenshot(s) of the directory
Message was edited by:
tlog
Furthermore, the directory contents of the VM are not
in a package, as they are in the new VM.
Do you think this difference matters?
I highly doubt it matters - the difference is merely the extension of the folder, and how OS X treats it.
I also notice that the original VM has many Virtual
Disk files, most of which have not been used in
months.
I did delete the old IDE disk from the setting
window, so I would have expected these older disk
files to have gone. are they still used somehow? Can
I delete them?
Deleting a disk from the settings window just removes the reference to the disk from the vmdk. It's still taking up space on your hard drive, and (assuming nothing else is referencing the virtual disk) should be safe to delete.
Thanks etung,
So Im still left wondering why this VM is consuming significantly more resource than the new VM....any other suggestions?
With B3 I did add the
aiomgr.buffered = "TRUE"
line to the Fusion config file, but I guess that would affect both VM's identically?
Either way, should that line still be in that file with B4?
With B3 I did add the
aiomgr.buffered = "TRUE"
line to the Fusion config file, but I guess that
would affect both VM's identically?
Either way, should that line still be in that file
with B4?
Well...Beta 4 uses that switch by default, so the two VMs are identical in that regard. However, some people have had better experience after they disabled buffered I/O.
Open your .vmx file and add the following line to the end:
hard-disk.useUnbuffered = "TRUE"
Restart your VM and see whether it makes a difference...
Ciao, Andreas
Open the vmx file with a text editor, it'll say in plain text virtualHW.version = "6" for instance.
thanks sparcdr,
i posted both VMX's above, and they are both virtual HW version 6
Cheers
Thanks Andreas. I'll give this a try as soon as I get a minute,
Ben
There are a lot of differences between the two VMX files that I posted above. Can anyone with more knowledge than I explain some of the differences, and whether they are likely to have any impact on performance?
Thanks