VMware Communities: Message List - New Laptop = VMware slower?!

Re: New Laptop = VMware slower?!

lolz — Wed, 15 Feb 2006 09:00:56 GMT

Have you read the link sent by jrj? The solution it provides is working well on an Inspiron 6000:

#!/bin/bash
echo 1 > /sys/module/processor/parameters/max_cstate
vmware
echo 8 > /sys/module/processor/parameters/max_cstate

Re: New Laptop = VMware slower?!

Canada_ eh_ — Tue, 14 Feb 2006 23:23:54 GMT

Doesn't look like that...

This is a strange problem. One Virtual Machine I have solved this issue by itself. It just started working as it should for some reason. But, sad to say, after I restarted that VM everything was as before again...

skr,

What host OS are you using?

Re: New Laptop = VMware slower?!

skr — Fri, 27 Jan 2006 16:51:39 GMT

I'm having this same problem on a Dell Latitude D610. Has anyone found a solution yet?

Re: New Laptop = VMware slower?!

jrj — Thu, 05 Jan 2006 16:11:07 GMT

kl1278 has posted another solution (setting max_cstate) in this thread http://www.vmware.com/community/thread.jspa?messageID=312185&#312185 it requires root access and setting the scaling governor to performance but it works better for me than the busy loop (the fan isn't constantly on now).

Re: New Laptop = VMware slower?!

haerench — Thu, 05 Jan 2006 15:06:35 GMT

I'm using fedora core 4 as host OS with vmware workstation 5.5.1 on a dell D810 (2Ghz, 2GB mem, 80GB disk). Windows XP is extreme slow as guest OS. I also use solaris 10 as guest OS. The performance of Solaris 10 is normal.
My dell has a SATA hard disk and dvd+-rw. I tried everything from deleting usb, dvd, ... in the windows XP. Also disabled acpi, cpuspeed, .... ; but no luck. Performance for Win XP stays very slow.
Performance is good when I use the loop script in my host OS. Even with acpi, usb and dvd enabled performance is good (as long as I keep the script running). When I stop the script, the performance drops to very slow.

Re: New Laptop = VMware slower?!

jrj — Thu, 05 Jan 2006 08:52:32 GMT

I have this problem as well using suse linux 10 on a fujitsu laptop (so its not just dell machines). I have also found that the problem only occurs when I run vmware on a linux host (and not on a windows host).

I've found two ways to work around it.

1. set noacpi as a host kernel boot option
2. set the scaling govenor to performance and put a load on the host machine. I run a do nothing shell script at nice 19 to provide a load to the machine while running vmware, since turning off acpi breaks too many things.

nice 19 busyloop
cat bin/busyloop

while ( true ) ; do a=$a; done

As I understand the problem (and I could be wrong) there is a problem with the idle task using C3 sleep states, hence putting a load on the system keeps linux from scheduling the idle task and vmware runs much better (well at least for me).

There is a caveat about my machines acpi support, I have not been able to successfully recompile its dsdt, because it has several errors in it that I haven't been able to figure out how to fix yet.

Re: New Laptop = VMware slower?!

s.c.v — Thu, 05 Jan 2006 07:19:18 GMT

Dunno. The performance on my Dell M70 is pretty unusable now. Supposedly, my company has set up a Platinum support contract with VMware that should start tomorrow. I'll put in a request via that channel and post back to this thread if I learn anything.

Re: New Laptop = VMware slower?!

Canada_ eh_ — Tue, 03 Jan 2006 19:30:52 GMT

Loading the host OS didn't affect anything. (I used the search tool in Windows to load the OS.)

However, looking at the VM's log file I found the following messages:
============================================================
Jan 03 11:10:21: vmx| SCSI0:0: Command READ(10) took 1.061 seconds (ok)
Jan 03 11:10:24: vmx| SCSI0:0: Command READ(10) took 1.079 seconds (ok)
Jan 03 11:10:24: vmx| SCSI0:0: Command READ(10) took 1.078 seconds (ok)
Jan 03 11:10:24: vmx| SCSI0:0: Command READ(10) took 1.078 seconds (ok)
Jan 03 11:10:24: vmx| SCSI0:0: Command READ(10) took 1.059 seconds (ok)
Jan 03 11:10:24: vmx| SCSI0:0: Command READ(10) took 1.056 seconds (ok)
Jan 03 11:10:27: vmx| SCSI0:0: Command READ(10) took 1.126 seconds (ok)
Jan 03 11:10:27: vmx| SCSI0:0: Command READ(10) took 1.068 seconds (ok)
Jan 03 11:10:27: vmx| SCSI0:0: Command READ(10) took 1.059 seconds (ok)
Jan 03 11:10:27: vmx| SCSI0:0: Command READ(10) took 1.059 seconds (ok)
Jan 03 11:10:30: vmx| SCSI0:0: Command READ(10) took 1.125 seconds (ok)
Jan 03 11:10:30: vmx| SCSI0:0: Command READ(10) took 1.116 seconds (ok)
Jan 03 11:10:30: vmx| SCSI0:0: Command READ(10) took 1.116 seconds (ok)
Jan 03 11:10:30: vmx| SCSI0:0: Command READ(10) took 1.116 seconds (ok)
Jan 03 11:10:30: vmx| SCSI0:0: Command WRITE(10) took 1.106 seconds (ok)
============================================================

This is interesting because it coincides with the "stalls" on my guest OS. Every 3-4 seconds it stalls for a second.

Does anyone know what this message means?

Re: New Laptop = VMware slower?!

s.c.v — Fri, 30 Dec 2005 02:52:21 GMT

Have any of you tried loading down the host OS slightly (e.g., "nice +10 ls -lR /")? I found this to help the guest performance for some reason on my Dell M50 (see yesterday's thread at http://www.vmware.com/community/thread.jspa?threadID=29012&tstart=40

I would be interested in hearing if you observe the same effect.

Is this strictly a problem with the Dell laptops?

Re: New Laptop = VMware slower?!

rtd.img — Thu, 29 Dec 2005 09:54:40 GMT

Canada, eh?,

I have exactly the same problem and - coincidence or not - with a Dell laptop too (D505, Centrino 1.7GHz, 512MB, XP+SP2). I installed an XP guest on an XP host and every 3..4 seconds the guest freezes. I have also tried: with ot without VMware tools, reinstalling the guest, reinstalling VMware, reinstalling the host, disabling speed stepping, fixed the timing, disabling DVD, USB, PCMCIA controller and the WiFi. No use at all.

And it is very annoying because:
1. I watched the performance of both machines (host and guest) at high speed. During the freeze time, the CPU used by the vmware-vmx process on the host drops to 0 (leading to almost 99% idle), while in the guest the task manager freezes too, but when it comes back it looks like nothing has happened;

2. Because of these "freeze points" the Windows input events are incorrectly processed: some keys input from the keyboard are lost, some keys input from the keyboard are duplicated, some mouse clicks - or, more specifically, mouse button ups or downs - are lost. This mouse behaviour results into a very, very ugly Windows behaviour: clicks are forgotten, clicks become drag-and-drop, drag-and-drops can't be completed, and many other.

Beside this... the guest works Ok and the host is not influenced of such problems.

Re: New Laptop = VMware slower?!

Canada_ eh_ — Fri, 09 Dec 2005 18:32:56 GMT

ScatterBrain,

I'm trying to figure out if we hve the same problem. Every two, three seconds or so, my WMware freezes for half a second. Not WMware itself, but the guest OS running in WMware.

I have a Dell Latitude D810 with XP pro. The virtual machines I'm running are W2000 installations. I have tryed everything - resinstalling WMware, disabling speed stepping, disabling DVD and USB in the virtual machine.

It would be interesting if we have the same problem, since you are running Linux and I'm running XP.

Re: New Laptop = VMware slower?!

tepsipakki — Wed, 07 Dec 2005 16:31:58 GMT

I don't have performance problems with Ubuntu, but the dmesg-messages were familiar and there is a bug opened:

https://bugzilla.ubuntu.com/show_bug.cgi?id=17569

to quickly fix the issue, run "sudo /etc/init.d/hotkey-setup stop"

Re: New Laptop = VMware slower?!

lolz — Sat, 26 Nov 2005 15:46:08 GMT

hi,

sorry, I had very little time to try to solve this issue, but here is what I did today:
- Adjusted Windows for "best performances" (probably the most important thing to do)

Here is my VMWare config file :
--------------------------------------------------
#!/usr/local/bin/vmware
config.version = "8"
virtualHW.version = "4"
memsize = "320"
displayName = "Windows XP Professional"
guestOS = "winxppro"
nvram = "Windows XP Professional.nvram"
#paevm=true

svga.maxWidth = "1920"
svga.maxHeight = "1200"
svga.vramSize = 9216000

ide0:0.present = "TRUE"
ide0:0.fileName = "Windows XP Professional.vmdk"
ide0:0.redo = ""

isolation.tools.hgfs.disable = "TRUE"
snapshot.disabled = "TRUE"

ide0:1.present = "TRUE"
ide0:1.fileName = "/dev/scd0"
ide0:1.deviceType = "atapi-cdrom"
floppy0.startConnected = "FALSE"

ide0:1.startConnected = "TRUE"
tools.syncTime = "FALSE"
--------------------------------------------------

With all this, I was able to obtain some pretty good performances. The trouble is that I need networking for my guest OS. As soon as I reinstalled the network adapter, the VMWare tools started to take 2 or 3 min again before becoming "enabled"... I disabled DHCP and it is now much better, but still not perfect.

Anyway, I now re-enabled acpi support on my host OS and powernowd, and I really don't see any difference. Even the fglrx drivers are probably not useful to speed up VMWare (I will verify this latter because I need to restart X).

To summarise, the best things to do to get decent performances without loosing too much time are probably:
- and maybe also to remove the usb adapter

Re: New Laptop = VMware slower?!

dfranz — Tue, 22 Nov 2005 00:03:23 GMT

I have been working with an Inspiron 6000 running Libranet Gnu/Linux(Debian 3.1 with some admin
tools and some other software added) and with kernel 2.6.14-2. I have made sure that I am
in performance mode and have used gkrellm (compiled from latest source) and the gkfreq plugin
to monitor the frequency. It is at 1600 mHz continuously and I have a 1.6 GHz Pentium M CPU.
Also the real-time clock is rtc not genrtc. Currently I'm not concerned with poor time keeping in
the W2K guest-later on that one.

I am using Workstation 5.0 with a W2K guest at SP2. Boot and shutdown times are
very slow. Also opening any application takes many seconds. I have my W2K desktop set to
open apps with a single click. When I single click on an icon, nothing happens for many seconds
but finally the selected application will open. Also the cursor is often missing in text dialogs,
such as the login dialog, as well as sometimes in a command-prompt window. In the later the
cursor comes and goes based on what has been done in the window.

Run-time performance of my applications, at least those that do not write much to the screen,
appears to be unaffected. However, I can type at the command line faster than the system
can keep up. Even during login, typing the password has the asterisks one or two characters
behind the key being struck. I have never seen this behavior before. Its like the refresh operations
on the screen are orders of magnitude too slow. But compute intensive and even disk I/O
intensive operations complete as I expect!?

The really strange result is that I can speed the boot process to nearly what I have had on
my previous laptop, an Inspiron 8100, by resizing the Vmware window continuously during the
boot process. The same works for shutdown. Also when W2K is finally up, double clicking on an
icon will bring the application up immediately. In fact the second click can be any place on the
desktop(outside of another icon or command-prompt window). The other characteristic that
is strange is that the task manager shows 100% CPU use on W2K even with the most minor
operation. At the same time gkrellm shows essentially zero CPU on Linux. This occurs outside
of CPU intensive applications. Examples, are opening and closing Explorer, loading Internet
Explorer, etc. The mouse pointer shows the hourglass for a long time after the app is open.
Sometimes just clicking on the desktop clears the hourglass!

It seems to me that something is amiss with interupt handling or with the W2K graphics being
starved for CPU cycles. I was using APCI and not APM. However, I have just recompiled
the kernel using only APM and Vmware is back! The bootup and shutdown are like they should
be, a single click on an icon brings up the app rapidly, etc.

The CPU appears to be running at full speed always with no way of reducing it. The centrino
cpu frequency scaling system appears to depend on APCI being present.

Thus ScatterBrain and other posters with similar problems might want to try recompiling the
kernel using APM and not APCI. Perhaps there is a kernel boot parameter that will fix the
interupt handling with graphics that seems to be at the root of the problem. My BIOS is a bit
out of date, at A07 and with A09 available. The A08 BIOS update mentioned some changes
to the video BIOS, but I'm not sure that is related to the current problem. My video card
is an ATI M22 (Radeon Mobility M300) according to lspci.

This fix comes at the cost of some other functionality. Hopefully there is a fix via a boot
parameter for the problem with APCI on this latptop.

Delbert Franz

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 11 Nov 2005 13:47:34 GMT

OK, I've made yet another discovery...

Last night when you asked me to run the VM without the USB attached, I basically removed the checkbox in the USB Adapter that said (to the effect) automatically attach USB devices when the VM has the focus.

This morning, I actually removed the USB adapter completely from the VM setup. Things are *MUCH* better. I'm not willing to say that things are cured, but they are better. I still have to place the CPU in the "performance" goverenor, but I can handle that.

I'd like to have lolz do the same thing and see if he gets improved performance. Just to make sure that I'm not wishing it to happen.

I still have to wait several minutes for XP to shut down, but I've seen that happen when XP was the only OS on the system at all, so I'm so much concerned about that - although now it more like 2 minutes instead of 10.

Re: New Laptop = VMware slower?!

petr — Fri, 11 Nov 2005 06:00:01 GMT

In that case I have no idea. 83Hz is quite low value and you should not have any problems. Maybe that BIOS in your laptop always slows down processor, regardless of cpufreq setting. I cannot think of any other reason. Can you try running 'x86info --mhz', whether it will report 2GHz or no?

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 11 Nov 2005 02:47:18 GMT

The Host Clock does change, I see it going from 0 to 19, and then 19 to 83 and staying there until I close vmware down. Then it goes from 83 to 19 and then from 19 to 0.

Here's a 'dmesg' snippet:
=============================================================
[4297105.157000] usb 3-2: USB disconnect, address 2
[4297139.958000] /dev/vmnet: open called by PID 10136 (vmware-vmx)
[4297139.958000] /dev/vmnet: port on hub 8 successfully opened
[4297140.062000] /dev/vmmon[10142]: host clock rate change request 0 -> 19
[4297145.565000] /dev/vmmon[10142]: host clock rate change request 19 -> 83
[4297178.819000] /dev/vmnet: open called by PID 10142 (vmware-vmx)
[4297178.819000] /dev/vmnet: port on hub 8 successfully opened
[4297406.155000] /dev/vmnet: open called by PID 10142 (vmware-vmx)
[4297406.155000] /dev/vmnet: port on hub 8 successfully opened
[4297406.255000] /dev/vmmon[10142]: host clock rate change request 83 -> 19
[4297406.361000] /dev/vmmon[10136]: host clock rate change request 19 -> 0
[4297406.401000] vmmon: Had to deallocate locked 183279 pages from vm driver c21b4000
[4297406.406000] vmmon: Had to deallocate AWE 2213 pages from vm driver c21b4000
=============================================================

Yes, I'm running VMware 5.0, build 13124 that I downloaded from the VMware website around the first of this month.

I have tried the VM with and without the CDROM and Floppy, but I have _not_ tried it without USB until just now - it did not help.

Re: New Laptop = VMware slower?!

petr — Thu, 10 Nov 2005 20:57:46 GMT

Hm. To reiterate, you still use WS5.0, and when you are using your VM, last line about host clock frequency says that it was switched from ??? (probably 18) to 83 Hz, yes? And you have disconnected CDROM/floppy/USB from guest?

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 10 Nov 2005 19:38:23 GMT

petr,

I assure you, when it's in "performance" mode, the machine is most surely running at 2GHz.

root@enterprise:~# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 13
model name : Intel(R) Pentium(R) M processor 2.00GHz
stepping : 8
cpu MHz : 1995.624
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe nx est tm2
bogomips : 3940.35

root@enterprise:~# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
performance

I switch to 'performance' mode just before I fire up VMware - every time since the reload of Ubuntu. The very first VMware session was OK, all other since crappy.

Message was edited by:
ScatterBrain

Re: New Laptop = VMware slower?!

petr — Thu, 10 Nov 2005 19:14:58 GMT

In that case you are again not running at 2GHz with 'performance' governor, but at variable (~1GHz) frequency again...

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 10 Nov 2005 19:06:18 GMT

And just to confuse everybody a little more...I've taken a "VM Movie" of the system showing what I'm experiencing. THe movie is located here:

http://www.klcollins.org/files/vmware-perf

I looked at the movie and what bothers me is that it looks almost like the movie is playing a twice the normal speed of what's actually happening. So just imagine everything in that movie taking _at_least_ twice as long.

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 10 Nov 2005 18:29:50 GMT

Yet another bit of information...

I had to re-install Ubuntu yesterday (long story) and I reinstalled VMware as well. After the re-install, I actually had VMware running very well for one session. Here's what I did differently:

1). I used the latest ATI fglrx drivers - 8.18.8. I had been using 8.18.6 up until this point.
2). Switch the CPU Freqency Governor from "userspace" to "performance". This set the CPU at 2GHz.
3). I used EXT3 for my file systems instead of XFS or Reiser.
4). I pre-allocated the VM hard drive space. Before I had it configured to grow as needed.

For the first session, immediately after installing the VMware tools in XP Pro, everything was fine. I thought I was on a different machine.

That's the good news.

The bad news is that I left the machine running Windows Updates and walked downtown to get lunch. When I can back the machine was locked. I couldn't use the keyboard, the mouse moved, but the system would not respond to "clicks". So I hit the power button, killing Ubuntu.

When I restarted, the performance was in the gutter again. Nothing I do now is any better than before the re-install.

I've also noticed that the VMWare tools icon in the XP systray takes a *very* long time to go from "not ready" to "ready".

Don't know if it helps or just muddies the water, but there it is.

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 10 Nov 2005 17:01:39 GMT

If anybody find a way to compile a kernel allowing
good performances with VMWare, I am really
interested. Even if it means getting poor
performances with the DVD player, I really don't
care...

I agree totoally. I'm looking every day for hacks to make this work.

It's definatley the Inspiron Hardware. I've tried Ubuntu Warty and even Gentoo. Hoary was better, Gentoo a little more so than Hoary, but nothing like running VMWare on my desktop or previous laptop.

As of right now, I've reduced my use of VMware to simply checking my corporate e-mail. Even that is *painfully* slow. Someone please help.

Kevin

Message was edited by:
ScatterBrain

Re: New Laptop = VMware slower?!

lolz — Wed, 09 Nov 2005 17:19:25 GMT

I just tried again with a newly compiled kernel (2.6.12, the one coming with Ubuntu, without applying any patches) and with VMWare 5.5 RC2, and I still have the same problem (The host OS (windows XP) takes more than 5 min to shutdown for example).

If anybody find a way to compile a kernel allowing good performances with VMWare, I am really interested. Even if it means getting poor performances with the DVD player, I really don't care...

Message was edited by:
lolz

Re: New Laptop = VMware slower?!

lolz — Wed, 09 Nov 2005 17:16:40 GMT

Ubuntu as well.

With exactly the same system, VMWare is running fine on my desktop computer (AMD XP 2600+, Ubuntu breezy, kernel 2.6.12), so obviously the problem comes from the Inspiron hardware.

I tried to compile a 2.6.14 kernel and made sure that in /usr/src/linux/include/linux/libata.h:
#undef ATA_ENABLE_ATAPI
#undef ATA_ENABLE_PATA
where not enable, but it didn't change anything...

Re: New Laptop = VMware slower?!

ScatterBrain — Wed, 09 Nov 2005 03:19:05 GMT

lolz: Just out of curiosity, what host OS are you running?

Re: New Laptop = VMware slower?!

ScatterBrain — Wed, 09 Nov 2005 03:13:31 GMT

As with lolz, it make no difference whether the DVD drive is "attached" the VM or not, performance still stinks.

Re: New Laptop = VMware slower?!

lolz — Fri, 04 Nov 2005 20:29:22 GMT

hi,

I have exactly the same problem (same computer), both with VMware Workstation v5 and v5.5 beta.
Unfortunately, disconnecting the DVD doesn't make any difference.

Re: New Laptop = VMware slower?!

petr — Wed, 02 Nov 2005 22:00:57 GMT

I do have the DVD connected to the virutal machine.

In that case I would disconnect it unless you need it. Or try loading some CD disc to the drive... it could be that IDE driver uses some other call than SCSI driver to find whether device is ready, and on IDE this call is fast, while in the SCSI/SATA driver it is slow... It was seen in the past on Windows hosts as guest hiccuping every two seconds for second or so... Especially if guest are Windows, with insert media notification polling drive every two seconds...

Re: New Laptop = VMware slower?!

ScatterBrain — Wed, 02 Nov 2005 21:04:28 GMT

I do have the DVD connected to the virutal machine.

The Hard Drive has always shown up as "/dev/sda". From what I can tell, the patch only enables normal IDE access (PATA) for the second controller channel - the channel that the DVD is connected to.

Re: New Laptop = VMware slower?!

petr — Wed, 02 Nov 2005 20:27:00 GMT

And do not you have CD/DVD connected to the VM? I cannot imagine why 'hdc=noprobe' would make difference otherwise...

BTW, do you see your harddrive as hda or sda ? Do not you see it as /dev/hda when 'ATA_ENABLE_PATA' is not set, while as /dev/sda when 'ATA_ENABLE_PATA' is set? Though hdc should not make any changes in this...

Re: New Laptop = VMware slower?!

ScatterBrain — Wed, 02 Nov 2005 14:35:28 GMT

OK, I've made an intresting discovery.

I installed Gentoo on the laptop over the weekend. On Monday morning, I was able to run VMware (using the same Windows XP VM that I created in Ubuntu) and everything was *very* useable. I took this to mean that something in the Gentoo kernel was the cure. But at the time, I couldn't investigate as I was backup up from not having a useable machine for nearly a week. So off to work I went.

Late last night, I wanted to watch a DVD, but the performance of the DVD drive sucked. Video was very choppy. So I did some googleing and found that the problem was the SATA DVD drive. You see, this laptop has a SATA controller that drives both the Hard Drive and DVD+RW. The reason that DVD playback was so horrible was that DMA on the drive was disabled - and couldn't be enabled. (That's what I get for buying a machine with bleeding edge hardware in it.)

But my googleing also pointed out a hack that would treat the DVD as a SCSI device and improve the perfomance. So off I went and implemented it. My DVD viewing was pleasurable the rest of the night.

Here's the hack:
in /usr/src/linux/include/linux/libata.h I changed this:

#unde ATA_ENABLE_ATAPI
#undef ATA_ENABLE_PATA

to this:

#define ATA_ENABLE_ATAPI 1
#define ATA_ENABLE_PATA 1

and then on my kernel line I added this switch:
hdc=noprobe

When I came into work this morning and discovered that VMware was as slow as a snail. I about fell on the floor. I couldn't believe that the DVD drive was the cuplrit all along.

I did a bit of checking and discovered that Ubuntu implements this kernel hack as part of ti's kernel patch set. So just to confirm that I had found the problem, I removed the hdc=noprobe statement from the kernel line and VMware performance was back to useable.

That kernel switch is trigger. With it in place VMware's speed sucks, but DVD performance is OK - the drive is recogonized as /dev/sr0. With it off, VMware runs great, but DVD playback stinks - the drive is recognized as /dev/hdc.

Somebody help me out here. Why does this make a difference?

Re: New Laptop = VMware slower?!

petr — Fri, 28 Oct 2005 18:45:55 GMT

Type of your harddisk should not matter.

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 28 Oct 2005 17:44:03 GMT

Just out of curiosity, could the fact that my hard drive is a SATA drive have anything to do with it?

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 28 Oct 2005 17:43:07 GMT

Yes I did reload vmmon (and vmnet just for good measure). Neither made a difference.

I'm contimplating reloading the entire laptop. I don't know if it will have any effect otr not, but at this point, I'm not sure what else to do.

Re: New Laptop = VMware slower?!

slylos — Fri, 28 Oct 2005 16:13:50 GMT

Not sure if this helps (I see you already discussed CPU frequency) but I just had a very similar issue - WinXP SP2 running like a pig on my laptop (hp pavilion). I'm running OpenSuSE 10.0, and power management set my CPU Frequency to 'dynamic' - I changed it to 'performance' and my winxp guest os improved dramatically in performance - I hardly even notice a difference with two vm's running at the same time now (win2k server + winxp pro sp2). I'm running Visual Studio.NET 2003 in my guest OS for Windows development, and most of the time performance is very tolerable (not as good as a native install, but definately tolerable nontheless). I'm not sure how that would be changed in Ubuntu, but that may be something you should really look into . . .

Re: New Laptop = VMware slower?!

petr — Fri, 28 Oct 2005 16:10:16 GMT

Did you reload vmmon after making change?

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 28 Oct 2005 02:21:04 GMT

I do use the vmware-any-any update. And you are correct, Ubuntu uses "powernowd" by default for CPU frequency governing. I disabled that, set the performance governer (which put the CPU at 2GHz) and gave it a shot. No change.

Yet another example of how slow this setup is. When Windows XP boots, it has a fade-in effect going from black to a colored Windows XP flag in the middle and a green (or blue) progress bar going across the bottom. That fade in is not smooth. I see it in about 15 different chops. The progress bar is just as choppy.

All of this was as sooth as glass on my old laptop - which was Pentium M 1.4GHz with 512MB of RAM. Now I'm on a Pentium M 2GHz with 2GB of ram and I want the old machine back. Grrrr.

Re: New Laptop = VMware slower?!

petr — Fri, 28 Oct 2005 01:33:41 GMT

You are not using ondemand governor. You are running some cpu frequency selecting program on your host - probably cpudyn or something like that. Do not do that. Configure your host to use only one frequency - in any case kill cpudyn, and then either select performance governor, or select userspace governor and select one of frequencies supported by your processor.

Then if you are using vmware-any-any-update then just poweroff and poweron VM. If you do not use vmware-any-any-update then you have to stop VM, unload vmmon, reload it back, and poweron VM. It now should be fast and time should go as it should. If you are using dynamic frequency on host then time is not linear in the guest.

Re: New Laptop = VMware slower?!

ScatterBrain — Fri, 28 Oct 2005 00:48:50 GMT

Output of 'cpufreq-info' on host at idle:

enterprise ~ # cpufreq-info
cpufrequtils 0.3: cpufreq-info (C) Dominik Brodowski 2004
Report errors and bugs to linux@brodo.de, please.
analyzing CPU 0:
driver: centrino
CPUs which need to switch frequency at the same time: 0
hardware limits: 800 MHz - 2.00 GHz
available frequency steps: 2.00 GHz, 1.60 GHz, 1.33 GHz, 1.07 GHz, 800 MHz
available cpufreq governors: userspace, powersave, ondemand, conservative, performance
current policy: frequency should be within 800 MHz and 2.00 GHz.
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 800 MHz (asserted by call to hardware).

I had not tried the cpufreq-set -g performance, but I have used the CPUFreq Monitor applet in Gnome to set the CPU to 2GHz while I was running the VM. I did try the command tonight and neither of the situations helped. I can't confirm this, but I think Ubuntu uses the "ondemand" governer by default. And I have seen the CPU bounce from 800MHz to 2GHZ and back (in various increments) on occasion - but mostly when I was running Linux Apps and not VMware.

Just as an asside: I have noticed that while I'm running th VM, that the clock on XP doesn't keep sync with the Linux clock. For example, that afternoon I ran the XP VM for more than 15 minutes (just let it idle) and the clock changed 1 time (ie it changed from 3:00 to 3:01) and then never moved again. Don't know if that helps or muddies the water, but it's what I'm seeing.

Re: New Laptop = VMware slower?!

petr — Thu, 27 Oct 2005 22:58:52 GMT

Thanks for dmesg. Apparently there are no problems with /dev/rtc.

Can you run 'cpufreq-info' on your host? And did you tried 'cpufreq-set -g performance' ?

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 27 Oct 2005 19:46:23 GMT

I had not looked at that article, but I followed the link and did as it directed. Unfortunetly, it didn't help. Sorry.

Kevin.

Re: New Laptop = VMware slower?!

Raymond — Thu, 27 Oct 2005 19:24:48 GMT

Hi there,

Since this is a laptop did you set the clock speed for Workstation.

Check out this link

http://www.vmware.com/support/kb/enduser/std_adp.php?p_sid=psxiG5Th&p_lva=&p_faqid=1227

Re: New Laptop = VMware slower?!

ScatterBrain — Thu, 27 Oct 2005 19:21:41 GMT

While reading some other posts on this forum, I saw a request from some to post the last 30 lines of dmesg, so I looked at the output of dmesg after I ran the virtual machine and here's what I recieved:

[4314954.311000] /dev/vmnet: port on hub 0 successfully opened
[4314954.869000] Device sr0 not ready.
[4314954.873000] Device sr0 not ready.
[4314954.880000] Device sr0 not ready.
[4314954.884000] Device sr0 not ready.
[4315012.707000] Device sr0 not ready.
[4315012.716000] Device sr0 not ready.
[4315226.273000] Device sr0 not ready.
[4315323.306000] device eth1 left promiscuous mode
[4315323.306000] bridge-eth1: disabled promiscuous mode
[4315323.306000] /dev/vmnet: open called by PID 23003 (vmware-vmx)
[4315323.306000] device eth1 entered promiscuous mode
[4315323.306000] bridge-eth1: enabled promiscuous mode
[4315323.306000] /dev/vmnet: port on hub 0 successfully opened
[4315323.483000] /dev/vmmon[23003]: host clock rate change request 83 -> 19
[4315323.498000] device eth1 left promiscuous mode
[4315323.498000] bridge-eth1: disabled promiscuous mode
[4315323.649000] /dev/vmmon[22997]: host clock rate change request 19 -> 0
[4315323.693000] vmmon: Had to deallocate locked 58184 pages from vm driver c4c1e000
[4315323.708000] vmmon: Had to deallocate AWE 4009 pages from vm driver c4c1e000[4318079.703000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4318079.703000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4318079.767000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4318079.767000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4318652.423000] ipw2200: Firmware error detected. Restarting.
[4319369.714000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319369.714000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319369.809000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319369.809000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319369.910000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319369.910000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319369.985000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319369.985000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.301000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319370.301000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.365000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319370.365000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.497000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319370.497000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.551000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319370.551000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.662000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319370.662000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319370.737000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319370.737000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319371.013000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319371.013000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319371.077000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319371.077000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319371.188000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319371.188000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319371.995000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319371.995000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319372.199000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319372.199000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319372.273000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319372.273000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319372.375000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319372.375000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.129000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319373.129000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.302000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319373.302000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.376000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319373.376000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.498000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319373.498000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.552000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319373.552000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.653000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319373.653000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.717000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319373.717000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.818000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319373.818000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319373.902000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319373.902000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319374.024000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319374.024000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319374.088000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319374.088000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319404.827000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319404.827000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319404.871000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319404.871000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319405.474000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319405.474000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319405.580000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319405.580000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319406.347000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319406.347000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319406.442000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319406.442000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319406.811000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319406.811000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319407.534000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319407.534000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319407.821000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4319407.821000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4319407.926000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4319407.926000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4320775.236000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4320775.236000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4320775.382000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4320775.382000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4320777.383000] atkbd.c: Unknown key released (translated set 2, code 0xaa on isa0060/serio0).
[4320777.383000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4320777.457000] atkbd.c: Unknown key pressed (translated set 2, code 0xaa on isa0060/serio0).
[4320777.457000] atkbd.c: Use 'setkeycodes e02a <keycode>' to make it known.
[4321751.442000] /dev/vmnet: open called by PID 27170 (vmware-vmx)
[4321751.442000] device eth1 entered promiscuous mode
[4321751.442000] bridge-eth1: enabled promiscuous mode
[4321751.442000] /dev/vmnet: port on hub 0 successfully opened
[4321751.696000] /dev/vmmon[27176]: host clock rate change request 0 -> 19
[4321756.178000] /dev/vmmon[27176]: host clock rate change request 19 -> 83
[4321763.870000] Device sr0 not ready.
[4321763.883000] Device sr0 not ready.
[4321764.378000] device eth1 left promiscuous mode
[4321764.378000] bridge-eth1: disabled promiscuous mode
[4321764.378000] /dev/vmnet: open called by PID 27176 (vmware-vmx)
[4321764.378000] device eth1 entered promiscuous mode
[4321764.378000] bridge-eth1: enabled promiscuous mode
[4321764.378000] /dev/vmnet: port on hub 0 successfully opened
[4321765.007000] Device sr0 not ready.
[4321765.011000] Device sr0 not ready.
[4321765.018000] Device sr0 not ready.
[4321765.022000] Device sr0 not ready.
[4321794.349000] ipw2200: Firmware error detected. Restarting.
[4321813.313000] Device sr0 not ready.
[4321813.321000] Device sr0 not ready.
[4322122.572000] Device sr0 not ready.

I hope that helps someone.

New Laptop = VMware slower?!

ScatterBrain — Thu, 27 Oct 2005 18:57:00 GMT

On Monday of this week, I received my new laptop - a Dell Inspiron 6000d. This machine is far and away the most powerful laptop I have ever owned. I'm running Ubuntu 5.10 on it at the moment.

I installed VMware Workstation v5 build 13124 on it on Tuesday and everything *seemed* to work just fine. I never had a chance to install any guest OSes until Wednesday. On Wednesday, I loaded WinXP Pro SP2 on it and started off by getting XP updated. I noticed small things then. When a menu in XP would fade in or out, it wouldn't be smooth or quick, instead it would be choppy and slow. But I figured I was doing so much with the machine at the time that it really wasn't a big deal. Once I get the XP guest updated, I settled in to do some work. The the crap hit the fan. Nothing was peppy, everything was slow. I thought the guest OS was hosed, I re-created it. No help there.

A good example of the problem would be when I start Internet Explorer - it takes nearly two minutes for it to fire up. Or, when I tried to browse with "My Network Places" and it was nearly seven minutes before I could find any other hosts on the LAN.

(All of these performance problems were experienced _after_ I had the vmware tools installed.)

Something isn't right. I'm wondering if it may be the machine, if it's Ubuntu 5.10, or if it's just a bogus VMware build by me.

Because I hoping it's the latter, I'd like to start with a bogus build idea. This is what I had to do to get VMware on this machine (taken from my wiki):

-- I had to install the following packages: gcc-3.4 g++-3.4 linux-headers-2.6.12-9-386 build-essential

-- Next I downloaded the VMware application, untarr'ed and ran the vmware-install.pl script but did NOT run the configuration option toward the end of the installer.

-- Then I downloaded this utility: http://ftp.cvut.cz/vmware/vmware-any-any-update94.tar.gz.

-- Untarr'ed the vmware-any-any file and then ran:
CC=gcc-3.4 ./runme.pl

I don't know if that helps at all, but that's how I got VMware installed.

After the installation, I ediitted the .vmx file for the WinXP guest and added this directive:

paevm="true"

Finally, a bit of specs about the machine:
Pentium-M 2 Ghz
2GB RAM
80GB SATA hard drive

The guest OS is sitting on an XFS filesystem

Any help at getting VMware more responsive would be helpful.

Thanks,
Kevin