Opened 15 years ago
Closed 14 years ago
#5294 closed defect (wontfix)
VirtualBox performance is much worse when Intel VT-x is enabled
Reported by: | Michał Gołębiowski | Owned by: | |
---|---|---|---|
Component: | other | Version: | VirtualBox 3.0.8 |
Keywords: | intel, vt, vt-x, virtualization, 64, 32 | Cc: | |
Guest type: | Windows | Host type: | Linux |
Description (last modified by )
I have Dell Latitude E6500 with P8600 CPU. When I run my Windows XP guest on VirtualBox OSE without virtualization VT-x option inactive it runs much faster than in the case I turn it on. It is really visible - with VT-x on I even see how screen flickers during the boot process.
I did Peacekeeper tests in Firefox and I got these results:
- VT-x off, 3D support off - 1596
- VT-x off, 3D support on - 1679
- VT-x on, 3D support off - 900
- VT-x on, 3D support on - 1314
Google Chrome behaves even more strange - with VT-x off it runs faster than Firefox, while with VT-x on it runs *VERY* slow but looses no frames in graphic tests and it finally gets about 8800 points.
I checked both on OSE version and a closed one. My OS is 64-bit Ubuntu Karmic. For more information, see here: https://bugs.launchpad.net/bugs/455112
Attachments (6)
Change History (22)
comment:1 by , 15 years ago
Description: | modified (diff) |
---|
comment:2 by , 15 years ago
comment:4 by , 15 years ago
A person called Fale here: http://ubuntuforums.org/showpost.php?p=8167746&postcount=8 claims the problem didn't exist in Ubuntu Jaunty (9.04).
comment:5 by , 15 years ago
@sandervl73 Is that wht You asked for or should I provide additional information? If so, what kind?
comment:9 by , 15 years ago
by the why, if you wanna see how slow do VT-x on old CPU,
compile GCC.
I am a damn Gentoo user, I do know how SLOW my system is.
enable VT-x on T7x00 series Intel CPU result about 5 times slower GCC compiliation, I can duplicate this result on any host/guest os combination with same hardware.
VT-x just slow, and really slow on old hardware.
by , 15 years ago
Attachment: | VBox.2.log added |
---|
enable VT-x, host Vista x64 SP2, guest WinXP SP2, IE 8 Peacemaker 277
comment:10 by , 15 years ago
@AndCycle I guess AMD offers better virtualization capabilities (and Nested Paging in ALL modern CPUs - Intel still releases processors with and without even VT-x and Nested Paging is available only in Core i7). But I thought that it at least should behave better than software solutions? Strange.
BTW, there are some people saying that VT-x efficiency in VBox run on Ubuntu Jaunty (9.04) was rather OK, so maybe it's an Ubuntu issue? Hard to say, I hope logs will say sth.
by , 15 years ago
Attachment: | VBox.3.log added |
---|
disable VT-x, host Vista x64 SP2, guest WinXP SP2, IE 8 Peacekeepr 355d
comment:11 by , 15 years ago
VT-x is slow on old Pentium 4 machines (unusable imho). On a Core 2 Duo it's fine. On Nehalem machines is extremely fast. AMD-V has been fast from day one. All mostly related to world-switch overhead.
If you're going to compile, then use SATA drives. IDE is virtualization unfriendly, which is more visible with VT-x and AMD-V; especially on older machines.
comment:12 by , 15 years ago
@m_gol
you can take a look at this
AnandTech: Hardware Virtualization: the Nuts and Bolts
comment:13 by , 15 years ago
@sandervl73 I use Core2 Duo (P8600, specifically) on a new laptop (it has a month or so), so I guess it should work fine - and it doesn't.
comment:14 by , 15 years ago
I get quite different results. Windows XP guest, FireFox 3.5.4, Peacekeeper:
- VT-x on, 3d off - 2832 (nested paging off) - VT-x on, 3d off - 2877 (nested paging on) - VT-x off, 3d off - 2363
Core i7 3.2 Ghz; Windows 7 x64.
comment:15 by , 15 years ago
@sandervl73 Core i7 is a new CPU, the difference between virtualization on and off is certainly bigger than in older CPUs, especially in mobile ones (like my P8600).
comment:16 by , 15 years ago
Hardware ======== Motherboard: GA-P55-UD3P CPU: I5 750 Memory: 4GB Software ======= Host: Ubuntu 9.10 64bit Guest: Windows XP SP3
Anything but disabling VT-x causes terrible performance of Windows XP guest
Unlike the I7 family I5 does not have certain virtualization features like VT-d.
Is it possible that the missing features are relied upon by VirtualBox?
comment:17 by , 14 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
Just did this test with a T9550 (so comparable to your P8600) with VBox 4.0.4. With VT-x the test is a bit faster than without (WinXP guest, IE8, Debian/Sid host). Please reopen if still relevant for you with VBox 4.0.4.
by , 14 years ago
Attachment: | VBox.4.log added |
---|
Ubuntu 10.10 x64, VirtualBox 4.0.4: VT-x off, 3D acc. on
by , 14 years ago
Attachment: | VBox.log.2.1 added |
---|
Ubuntu 10.10 x64, VirtualBox 4.0.4: VT-x on, 3D acc. on
comment:18 by , 14 years ago
Resolution: | fixed |
---|---|
Status: | closed → reopened |
I'm sorry but I have to reopen. I've just did some tests on the newest Ubuntu 10.10 (Maverick) x64 with VirtualBox 4.0.4. I was unable to run a Peacekeeper test as it failed on both Firefox 4 and Opera 11.10, so I run SunSpider on Opera 11.10 and Kraken on Firefox 4. In both cases browsers were slightly faster when VT-x was off. Even if you consider it within error margin (it was up to 1.1 times faster) there was certainly no gain.
Also, slow screen refreshing during Windows loading (when text is displayed if you boot it with /sos option added to boot.ini) when VT-x is on still exists.
Maybe it's just VT-x without nested paging on P8600 is so poorly designed that VirtualBox with its software implementation does better, I don't know. But then mark it as invalid or wontfix. If you think it's not the case, I'll be happy to assist you further (I'm sorry but I didn't have time back in time and doing these tests properly (like running it just after a system boot with no other applications run etc.) requires some effort).
comment:19 by , 14 years ago
Resolution: | → wontfix |
---|---|
Status: | reopened → closed |
There will be improvements in future releases but on some hardware there might be indeed a difference. And 'slightly' faster is nothing which justifies such an open ticket.
I'm sorry, I forgot to mention that I changed my VBox version to non-OSE one (from official repository for Ubuntu Karmic) and the problem still exists.