On Sun, 2009-03-08 at 20:37 +0100, Frank Mehnert wrote: > Brian, Hi Frank,
> Please could you be a bit more specific: Most definitely. > * I assume that you are running Windows XP / 32, is that right? Indeed. > * Do you have VT-x / AMD-V enabled for that VM or not? Yessir. > * Which virtual hard disk controller are you using, IDE or SATA? IDE. > * Check the disk settings in the guest: Does the guest drive the > hard disk in DMA mode or PIO mode? Hrm. How do I determine that from Windows XP? > Check the session information > dialog for this as well -- PIO and DMA accesses are shown there. The Session Information shows: DMA Transfers 492,884 PIO Transfers 59,770 Data Read 5,062,400,000 B Data Written 2,773,438,000B > * Which benchmark did you use for measuring the throughput? Something called DiskTT.exe. It doesn't look terribly "scientific" but I ran it a number of times to ensure consistency at least and it was consistent enough on both VBox and native hardware to at least some somewhat trustworthy. If we were niggling about 10-20% difference between the two I would be more concerned but we are talking about a factor of 4 or more. So what about host CPU? On a dual core system, is VBox able to take advantage of both cores if necessary? While top shows the VM pegged at 100% all of the time, sar shows that that most of the time a single core is not pegged but usually a combination of load on each, such as (there is much disk access during this as I rsync from C: to an F: which is a VDI: 03:51:23 PM CPU %user %nice %system %iowait %steal %idle 03:51:28 PM all 2.20 0.00 50.55 0.50 0.00 46.75 03:51:28 PM 0 3.79 0.00 1.80 1.20 0.00 93.21 03:51:28 PM 1 0.40 0.00 99.60 0.00 0.00 0.00 03:51:28 PM CPU %user %nice %system %iowait %steal %idle 03:51:33 PM all 3.09 0.00 51.10 0.40 0.00 45.41 03:51:33 PM 0 1.40 0.00 71.86 0.00 0.00 26.75 03:51:33 PM 1 4.81 0.00 30.26 0.60 0.00 64.33 03:51:33 PM CPU %user %nice %system %iowait %steal %idle 03:51:38 PM all 3.00 0.00 51.05 0.00 0.00 45.95 03:51:38 PM 0 0.60 0.00 99.40 0.00 0.00 0.00 03:51:38 PM 1 5.60 0.00 2.40 0.20 0.00 91.80 03:51:38 PM CPU %user %nice %system %iowait %steal %idle 03:51:43 PM all 2.40 0.00 50.75 0.40 0.00 46.45 03:51:43 PM 0 2.20 0.00 74.55 0.00 0.00 23.25 03:51:43 PM 1 2.59 0.00 27.49 0.80 0.00 69.12 03:51:43 PM CPU %user %nice %system %iowait %steal %idle 03:51:48 PM all 5.29 0.00 51.15 0.30 0.00 43.26 03:51:48 PM 0 1.60 0.00 84.23 0.00 0.00 14.17 03:51:48 PM 1 9.02 0.00 17.84 0.60 0.00 72.55 03:51:48 PM CPU %user %nice %system %iowait %steal %idle 03:51:53 PM all 5.30 0.00 51.10 0.80 0.00 42.80 03:51:53 PM 0 2.00 0.00 80.36 0.60 0.00 17.03 03:51:53 PM 1 8.40 0.00 22.20 0.80 0.00 68.60 03:51:53 PM CPU %user %nice %system %iowait %steal %idle 03:51:59 PM all 2.79 0.00 50.70 0.20 0.00 46.31 03:51:59 PM 0 1.00 0.00 99.00 0.00 0.00 0.00 03:51:59 PM 1 4.58 0.00 2.39 0.40 0.00 92.63 03:51:59 PM CPU %user %nice %system %iowait %steal %idle 03:52:04 PM all 4.70 0.00 51.15 1.60 0.00 42.54 03:52:04 PM 0 2.00 0.00 91.02 0.00 0.00 6.99 03:52:04 PM 1 7.41 0.00 11.22 3.21 0.00 78.16 03:52:04 PM CPU %user %nice %system %iowait %steal %idle 03:52:09 PM all 8.19 0.00 51.65 0.60 0.00 39.56 03:52:09 PM 0 5.41 0.00 40.68 0.40 0.00 53.51 03:52:09 PM 1 11.00 0.00 62.20 0.80 0.00 26.00 03:52:09 PM CPU %user %nice %system %iowait %steal %idle 03:52:14 PM all 8.50 0.00 51.20 0.20 0.00 40.10 03:52:14 PM 0 7.58 0.00 42.12 0.20 0.00 50.10 03:52:14 PM 1 9.58 0.00 60.08 0.40 0.00 29.94 03:52:14 PM CPU %user %nice %system %iowait %steal %idle 03:52:19 PM all 1.40 0.00 50.55 0.00 0.00 48.05 03:52:19 PM 0 2.40 0.00 1.60 0.00 0.00 96.00 03:52:19 PM 1 0.20 0.00 99.80 0.00 0.00 0.00 Could this starve the VM from disk accessing cycles or is the 100% CPU usage of the VM a red herring in this case? b.
signature.asc
Description: This is a digitally signed message part
_______________________________________________ vbox-users mailing list [email protected] http://vbox.innotek.de/mailman/listinfo/vbox-users
