On Wed, Sep 05, 2018 at 08:38:53PM +0200, Paul de Weerd wrote: > Hi all, > > Last year in May, i wrote to bugs@ about vmm VMs having slow time[1]. > Mike Larkin replied that this was a known issue[2], so I mostly > ignored it, thinking things would improve over time. And indeed they > have: these days, the play VM I have on my home workstation runs time > at normal speed. > > So I set about building another VM on another machine. Here time > seems to run slower again, but I don't really understand why. Of > course, the hardware of the host machines differs; my workstation uses > an i7-4770, the server with the slow VM runs on a Xeon E31260L, full > dmesgs at the end of this mail. > > I noticed that the VM on the Xeon runs at 100% interrupt according to > top / systat, but then found out the VM on the i7 is (almost) the > same, this one runs at 80-90% interrupt. > > Then I noticed that the interrupt rate on the i7 guest is half that of > the Xeon guest: >
Also: -proxy- ~> uptime 2:20PM up 229 days, 15:39, 1 user, load averages: 0.00, 0.00, 0.00 that VM has been up for 229 days and the clock has drifted less than 3 seconds over that time. Use tsc. -ml