prl_vm_app process 100% but task manager nil cpu usage

Seems like this started a few days ago and now I can't reduce the CPU usage without pausing the VM.
Anyone else observing similar behaviour?

Happens regardless of adaptive hypervisor, memory, number of CPUs, or any other
performance option for the VM.

 

This is with Lion, PD 7.0.15104, Macbook Pro 13" 2011 8GB.

 

One interesting observation that may be relevant: I cannot seem to remove Mac applications from my start menu even with "Share Mac applications" disabled. From earlier reports, it looks like sharing on the Dock might have been a problem.

Problem exists only with bootcamp Win7 VM - others seem fine.

Clean reinstall of Parallels didn't help.

 

Problem report ID: 16458108

 

Non-zero boot camp perf numbers and host info:

My Boot Camp - 09:23:27 +10.06 sec

devices.ahci.0.irq_reqs 156
devices.ahci.0.process 942

devices.ahci.1.irq_reqs 20
devices.ahci.1.process 20

devices.sata-sf0.read_req 781
devices.sata-sf0.write_req 5

devices.sata0.read_hit 776
devices.sata0.read_requests 781
devices.sata0.read_seq_requests 1
devices.sata0.read_total 655872
devices.sata0.write_requests 5

devices.sata0.write_total 151552

devices.sound.playback.bufd_size 882

kernel.activity.vcpu0.tsc_guest 26576187576
kernel.activity.vcpu0.tsc_host 287528400
kernel.activity.vcpu0.tsc_host_hlt 0
kernel.activity.vcpu0.tsc_mon 147097836

kernel.time.elapsed 10004874

kernel.vcpu0.async_thread_activate 184

kernel.vcpu0.guest_switch_hvt 16770

kernel.vcpu0.instr_emul 4383

kernel.vcpu0.int.injected_sum 1614

kernel.vcpu0.int.ipi 781
kernel.vcpu0.int.irq_process 964
kernel.vcpu0.int.redirect 5835
kernel.vcpu0.io_req 2438

kernel.vcpu0.mon_ret_to_host 6980
kernel.vcpu0.npt.fault 4809
kernel.vcpu0.npt.map 426
kernel.vcpu0.pic.setintrequest 1796
kernel.vcpu0.sendapirequest 961
kernel.vcpu0.tools_api_request 790

kernel.ws.active 400309
kernel.ws.alloc_limit 874117

kernel.ws.guest_pagein 279

kernel.ws.inactive 1446
kernel.ws.reclaimable 401755
kernel.ws.unreclaimable 20462

net.nic0.activate 8

net.nic0.bytes_in 112
net.nic0.bytes_out 1092

net.nic0.pkts_in 2
net.nic0.pkts_out 7

video.wait_tsc 26947625688
video.work_tsc 2232243


host - 09:23:27 +10.06 sec
cpu.idle 2858
cpu.inter 0
cpu.nice 0
cpu.system 1065
cpu.user 81
cpu.wait 0
disk.read.bytes 1847296
disk.read.req 187
disk.write.bytes 3452928
disk.write.req 48
mem.active -9600
mem.dirty 0
mem.free 7724
mem.inactive 1720
mem.swap 0
mem.wired 1592

 

I was able to trim prl_vm_app CPU utilization by disabling virtual memory in the Windows VM. I went from a minimum 26% value down to about 8%.

 

Thanks Todd, but this isn't a tweaking that is needed . Parallels needs to fix this bug in the VMM that sounds like it has existed for some time for others but just showed up here recently (even though I've been using PD for about 5 years!)

 

Wow, this really made a difference. This is really a candidate for hard to identify clock cycle gobbling activities in the parallels host that will not show up in Task Manager.

 

Doesn't help here - still 100% when doing nothing (not even virtual memory handling).