Hey, I'm seeing beween 12-22% CPU usage with CentOS 4.4 (all updates). I've not done any kernel or system parameter tweaking. Just applied updates via yum.

Host: MBP, 2 cores, 2GB RAM
Guest: CentOS 4.4, 1 virtual CPU.

I assume you've looked at the daemons that are enabled, e.g. with Applications > System Settings > Service Settings > Services ("Service Configuratin")

Yes, 120% (and anything over a few percent for that matter) is abnormal.


Usually we flip the DEBUG switch when we go from the "beta" denomination to the "release candidate" denomination. I'm not allowed to give you explicit dates.

Even after restart, both cores are hitting 20-30% just with X running. It's no problem though, I was just so excited I told a couple clients I could demo some stuff, but maybe I'll just hold off a little bit and get the non DEBUG version before I go out on a limb there.


No worries, good to know that it is in the works. So, technically, if I knew which bit to flip, I could patch the binary and turn DEBUG off?

Same issue here. My MBP uses around 30% cpu when XP Home Edition idles. I configured fusion to use an 8 Gb partition, 2 cpu's and 256 Mb memory space. My machine's configuration:

MacBookPro, 15", 2 GHz Intel Core Duo, 1GB 667 MHz DDR2 SDRAM, OS X 10.4.8, VMware Fusion e.x.p (36932)

Thanks for the quick reply. Yes, I run VMware tools. I just checked, XP is ~97% idle. But wait... VMware now uses only about 10-13 % cpu when idle...? Some experimenting shows that my eToken USB stick (for VPN) seems to be the culprit. When I plug it in, vmware-vmx-debug goes up to ~30% and stays there (XP is still ~97% idle). When I unplug it, vmware drops to 10-13 % again. Is that expected behaviour? What cpu load would you expect in an idle state (with or without eToken)?

(BTW: The eToken is in fact the reason I'm considering VMware to run XP. My company only supports VPN for a Windows client).

Yes, it is normal for host CPU usage to be a little higher when USB devices are connected to the guest, even if the guest isn't actively using those devices.

I usually see a difference of under 5% between idle load with and without USB devices: so the more dramatic difference you're seeing is probably due to debug code in the beta build. This is also probably something we'll be able to optimize more before releasing the next beta.

Thanks!
--Micah