Desktop Window Manager recurring fatal errors every 10-20 minutes (0x8898009b). I get the crashes in the display windows manager two or three times an hour while running Windows 6 64 guest. Only happens when CPU's are set greater than 1. (Runs like a dog if only set to 1). This is quite annoying. I have reported this multiple times. Most recently in case [Parallels #1322515]. Have been told that this is being sent to development and that they would close the ticket. Not a fan of that process as ten the issues tends to fall off of the radar and will never be addressed. I am running triple 24" cinema display monitors in full screen mode with single ATI/AMD video card. Was able to produce requested dump. The windows log captures the DWM crash at each occurrence. The screens flicker, go blank for a minute as the DWM restarts and then all is OK until the next occurrence. Very annoying if you are in the middle of something and then have to wait for the screens to restart. I saw another thread with a similar problem a few product releases ago that was supposedly corrected yet I am getting this with current release? Will post any progress and results in case other are experiencing anything similar. Anyone else see anything similar or have any fixes for this?
Zendog, your issue is indeed being handled by development team, it is not off the radar. The cause of dwm failure you're seeing is not within graphics subsystem but is timer related. It may only happen on older CPUs lacking certain hardware feature. The only quick remedy is to set your VM to single core.
The reported issue is running on a Mac Pro with 2 x 2.8 Ghz Quad-Core Intel Xeon processors, what hardware feature are you referencing that is not supported in the Xeon architecture and is creating this issue? The single CPU remediation is not a good option as the VM guest runs like a dog on single CPU. Please keep me updates on the development progress towards a fix for this issue. Many thanks ...
Xeon is a marketing name for server processors since before 2000, newer chips are more capable. In this particular case the lack of invariant TSC is at the root of issue.
Thanks ... This information is useful in my better understanding. My configuration is a 2008 Mac Pro. So am I to understand that this issue exists with all 2008 and Older Mac Pro's or is there something additionally unique in my configuration that is triggering the DWM display crashes? Would have thought this issue to be more widespread and publicized if this is prevalent on all MAC PRO 2008 models. Thanks for continued digging in to this. I look forward to a fix when it can be crafted. Let me know if I can assist in any testing when the appropriate code fixes are available.
Try disabling time sync? If you haven't tried it, could you try disabling time sync with the host for the Windows guest? I've been having the same problem in another hypervisor running a Windows 8 guest - same error code, and it only happens with >1 VCPU. Time will tell if this works, but it may be the case that the host is bumping the guest time backward a couple ticks to maintain clock sync, and this causes the composition engine to bail. It's too early to tell if my problem is solved, but I ran across this thread while looking for solutions, and it'd be interesting if it's the same issue across multiple HVs. I don't use Parallels anymore, but I believe the setting is under: Configure > Features > Services > Synchronize time with the host computer Good luck.
Previous fix didn't work but I found one that did. While the fix I suggested worked well the first day I tried it out, it did not resolve the issue in the longer term. However, I got a suggestion from another forum that worked in my environment; don't know if it can be applied in Parallels. 1.) If there is a host setting to enable HPET (high precision event timer) support in the guest, enable HPET. 2.) Boot the guest, and issue the following command: bcdedit /set useplatformclock true 3.) Reboot the guest.
