Announcement: VirtualBox 4.3 Beta 1 released

[Frank Mehnert]

Please do NOT use this VirtualBox Beta release on production machines!
A VirtualBox Beta release should be considered a bleeding-edge release
meant for early evaluation and testing purposes.

You can download the binaries here:

http://download.virtualbox.org/virtualbox/4.3.0_BETA1

Please do NOT open bug reports at our public bugtracker but use our
VirtualBox Beta Feedback forum at

http://forums.virtualbox.org/viewforum.php?f=15

to report any problems with the Beta. Please concentrate on reporting
regressions since VirtualBox 4.2.16!

Version 4.3 will be a new major release. Please see the forum at

http://forums.virtualbox.org/viewtopic.php?f=15&t=56983

for the extended announcement and for a preliminary list of all changes
since VirtualBox 4.2.16.

Thanks for your help!

Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany

Hauptverwaltung: Riesstr. 25, D-80992 München
Registergericht: Amtsgericht München, HRA 95603
Geschäftsführer: Jürgen Kunz

Komplementärin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
Geschäftsführer: Alexander van der Ven, Astrid Kepper, Val Maher

[Stephan von Krawczynski]

On Wed, 21 Aug 2013 10:44:44 +0200

Post by Frank Mehnert
[...]
Version 4.3 will be a new major release. Please see the forum at
http://forums.virtualbox.org/viewtopic.php?f=15&t=56983
for the extended announcement and for a preliminary list of all changes
since VirtualBox 4.2.16.

Hm, read the changelist, but networking performance or latency aren't even
mentioned.
That probably means it makes no sense to await 4.3 because it will not improve
that issue.
In the meantime we tested kvm/qemu with virt-manager and have to say that it
contains all the things we suggested for vbox networking and works just
brilliant...
It looks to me that linux host/guest support isn't really a target any more
for the vbox crew.

[Frank Mehnert]

Post by Stephan von Krawczynski
Hm, read the changelist, but networking performance or latency aren't even
mentioned.
That probably means it makes no sense to await 4.3 because it will not
improve that issue.
In the meantime we tested kvm/qemu with virt-manager and have to say that
it contains all the things we suggested for vbox networking and works just
brilliant...
It looks to me that linux host/guest support isn't really a target any more
for the vbox crew.

I already mentioned that the changelog isn't complete. Apart from that
we expect some performance improvements from the VT-x rewrite and other
internal VMM changes. It makes definitely sense to check this and the
next betas.

And it is _wrong_ that Linux host/guest support isn't a primary target
for the VBox development team.

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany

Hauptverwaltung: Riesstr. 25, D-80992 München
Registergericht: Amtsgericht München, HRA 95603
Geschäftsführer: Jürgen Kunz

Komplementärin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
Geschäftsführer: Alexander van der Ven, Astrid Kepper, Val Maher

[Stephan von Krawczynski]

On Wed, 21 Aug 2013 11:47:45 +0200

Post by Frank Mehnert

Post by Stephan von Krawczynski
Hm, read the changelist, but networking performance or latency aren't even
mentioned.
That probably means it makes no sense to await 4.3 because it will not
improve that issue.
In the meantime we tested kvm/qemu with virt-manager and have to say that
it contains all the things we suggested for vbox networking and works just
brilliant...
It looks to me that linux host/guest support isn't really a target any more
for the vbox crew.

I already mentioned that the changelog isn't complete. Apart from that
we expect some performance improvements from the VT-x rewrite and other
internal VMM changes. It makes definitely sense to check this and the
next betas.
And it is _wrong_ that Linux host/guest support isn't a primary target
for the VBox development team.
Kind regards,
Frank
From my current experience with former vbox versions and several qemu versions

I can tell you that you won't improve vbox network performance if you don't
implement multiqueuing. _Exactly that_ makes the difference in latency and
throughput performance.
Even VMWare understood that and implemented it as big new feature in their
vmxnet3 drivers.
Don't get me wrong, some of the vbox network stuff is pretty nice (like the
802.1q vlan implementation in bridging (which explodes only in 3.10.7 kernels
on guest when using around >100 vlans)), but in the end it does not save you
because the latency is simply not good enough for todays' networking
experience with LAN latencies of less than 0,1 ms.

[Frank Mehnert]

Post by Frank Mehnert
From my current experience with former vbox versions and several qemu
versions I can tell you that you won't improve vbox network performance if
you don't implement multiqueuing. _Exactly that_ makes the difference in
latency and throughput performance.

This is a bold message. Implementing multiqueuing will perhaps help
but so far we know a couple of places which need optimizing. 4.3 will
address some places, later versions will address more. At the moment
we do much too many context changes for high-traffic operations like
networking so this is the first place we started optimizing. The final
results will show how successul these optimizations were.

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany

Hauptverwaltung: Riesstr. 25, D-80992 München
Registergericht: Amtsgericht München, HRA 95603
Geschäftsführer: Jürgen Kunz

Komplementärin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
Geschäftsführer: Alexander van der Ven, Astrid Kepper, Val Maher

[Alexey Eromenko]

Well, the guest-side things are already optimized with VirtIO network
(same as KVM), so only the host-side things should be improved.

Post by Frank Mehnert

Post by Frank Mehnert
From my current experience with former vbox versions and several qemu
versions I can tell you that you won't improve vbox network performance if
you don't implement multiqueuing. _Exactly that_ makes the difference in
latency and throughput performance.

This is a bold message. Implementing multiqueuing will perhaps help
but so far we know a couple of places which need optimizing. 4.3 will
address some places, later versions will address more. At the moment
we do much too many context changes for high-traffic operations like
networking so this is the first place we started optimizing. The final
results will show how successul these optimizations were.
Kind regards,
Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany
Hauptverwaltung: Riesstr. 25, D-80992 München
Registergericht: Amtsgericht München, HRA 95603
Geschäftsführer: Jürgen Kunz
Komplementärin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
Geschäftsführer: Alexander van der Ven, Astrid Kepper, Val Maher
------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and
AppDynamics. Performance Central is your source for news, insights,
analysis and resources for efficient Application Performance Management.
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
VBox-users-community mailing list
https://lists.sourceforge.net/lists/listinfo/vbox-users-community
_______________________________________________

--
-Alexey Eromenko "Technologov"

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and
AppDynamics. Performance Central is your source for news, insights,
analysis and resources for efficient Application Performance Management.
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
VBox-users-community mailing list
VBox-users-***@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/vbox-users-community
_______________________________________________
Unsubscribe: mailto:vbox-users-com

[Stephan von Krawczynski]

On Wed, 21 Aug 2013 12:26:51 +0200

Post by Frank Mehnert

Post by Frank Mehnert
From my current experience with former vbox versions and several qemu
versions I can tell you that you won't improve vbox network performance if
you don't implement multiqueuing. _Exactly that_ makes the difference in
latency and throughput performance.

This is a bold message. Implementing multiqueuing will perhaps help
but so far we know a couple of places which need optimizing. 4.3 will
address some places, later versions will address more. At the moment
we do much too many context changes for high-traffic operations like
networking so this is the first place we started optimizing. The final
results will show how successul these optimizations were.
Kind regards,
Frank

Dear Frank,

I know it sound a bit strange, but I really try to help you focus on what is
worth working on.
I can tell you for sure that you cannot beat kvm/qemu in terms of networking
currently because vhost_net does zero-copy networking and multiqueues. And I
saw the difference in production. It is huge. Whatever you improve in other
fields it will not allow to come close to these to points. It simply makes a
difference to copy a GB datastream once instead of null times. And for the
scheduling/multitasking multiqueuing on vcpu basis is not beaten either.
If you see the figures you can't even tell that its a virtualised guest system.

VBox windows guest additions are very good, disk performance on linux guest is
very good, almost everything you can handle in userspace is quite ok (besides
the front-end btw as you cannot see any performance figures in the guest list
like in virt-manager, or cpu pinning). But not networking for servers.

[Frank Mehnert]

Stephan,

Post by Stephan von Krawczynski
I know it sound a bit strange, but I really try to help you focus on what
is worth working on.
I can tell you for sure that you cannot beat kvm/qemu in terms of
networking currently because vhost_net does zero-copy networking and
multiqueues. And I saw the difference in production. It is huge. Whatever
you improve in other fields it will not allow to come close to these to
points. It simply makes a difference to copy a GB datastream once instead
of null times. And for the scheduling/multitasking multiqueuing on vcpu
basis is not beaten either. If you see the figures you can't even tell
that its a virtualised guest system.

I don't doubt that the difference is huge. What I say is that we start
optimizing the areas we are aware of. Optimizing the VMM will not only
help to improve networking performance but it will help in other places
as well. What you say is that only multiqueuing matters any other
optimization is useless and this is what I don't agree.

Btw, we are happy to revise and integrate patches.

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany

Hauptverwaltung: Riesstr. 25, D-80992 München
Registergericht: Amtsgericht München, HRA 95603
Geschäftsführer: Jürgen Kunz

Komplementärin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
Geschäftsführer: Alexander van der Ven, Astrid Kepper, Val Maher

[Pau Garcia i Quiles]

On Wed, Aug 21, 2013 at 1:02 PM, Stephan von Krawczynski
<skraw-DcQCyzbjH0jQT0dZR+***@public.gmane.org>wrote:

If you see the figures you can't even tell that its a virtualised guest

Post by Stephan von Krawczynski
system.

Are those numbers available somewhere?

[Peter Ondruška]

Dear Frank,

sorry if this should be posted on beta forum but this is not an error just
a question. Is it possible to run this beta and if something does not work
as supposed and I want to revert back to 4.2.16 without guest VM restore?

Post by Frank Mehnert
Please do NOT use this VirtualBox Beta release on production machines!
A VirtualBox Beta release should be considered a bleeding-edge release
meant for early evaluation and testing purposes.
http://download.virtualbox.org/virtualbox/4.3.0_BETA1
Please do NOT open bug reports at our public bugtracker but use our
VirtualBox Beta Feedback forum at
http://forums.virtualbox.org/viewforum.php?f=15
to report any problems with the Beta. Please concentrate on reporting
regressions since VirtualBox 4.2.16!
Version 4.3 will be a new major release. Please see the forum at
http://forums.virtualbox.org/viewtopic.php?f=15&t=56983
for the extended announcement and for a preliminary list of all changes
since VirtualBox 4.2.16.
Thanks for your help!
Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany
Hauptverwaltung: Riesstr. 25, D-80992 MÃŒnchen
Registergericht: Amtsgericht MÃŒnchen, HRA 95603
GeschÀftsfÌhrer: JÌrgen Kunz
KomplementÀrin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
GeschÀftsfÌhrer: Alexander van der Ven, Astrid Kepper, Val Maher
------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and
AppDynamics. Performance Central is your source for news, insights,
analysis and resources for efficient Application Performance Management.
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
VBox-users-community mailing list
https://lists.sourceforge.net/lists/listinfo/vbox-users-community
_______________________________________________
?subject=unsubscribe

[Frank Mehnert]

Peter,

Post by Peter Ondruška
sorry if this should be posted on beta forum but this is not an error just
a question. Is it possible to run this beta and if something does not work
as supposed and I want to revert back to 4.2.16 without guest VM restore?

you can install the 4.3 Beta 1 and run your VMs. It will be possible to
switch back to 4.2.16 if you don't have saved states for the VM
involved. If you create any saved state with 4.3 (either by terminating
a VM session by saving the state or by creating a snapshot while the VM
is not powered off) you will not be able to restore such a state with
4.2.16. The other way around works: Saved states are backward-compatible
but not forward-compatible.

4.3 will also introduce some new settings. 4.2.16 will not be able to
read them and will just ignore them. So this shouldn't be a problem.

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany

Hauptverwaltung: Riesstr. 25, D-80992 MÃŒnchen
Registergericht: Amtsgericht MÃŒnchen, HRA 95603
GeschÀftsfÌhrer: JÌrgen Kunz

KomplementÀrin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
GeschÀftsfÌhrer: Alexander van der Ven, Astrid Kepper, Val Maher

[Peter Ondruška]

Thanks...

Post by Frank Mehnert
Peter,

Post by Peter Ondruška
sorry if this should be posted on beta forum but this is not an error

just

Post by Peter Ondruška
a question. Is it possible to run this beta and if something does not

work

Post by Peter Ondruška
as supposed and I want to revert back to 4.2.16 without guest VM restore?

you can install the 4.3 Beta 1 and run your VMs. It will be possible to
switch back to 4.2.16 if you don't have saved states for the VM
involved. If you create any saved state with 4.3 (either by terminating
a VM session by saving the state or by creating a snapshot while the VM
is not powered off) you will not be able to restore such a state with
4.2.16. The other way around works: Saved states are backward-compatible
but not forward-compatible.
4.3 will also introduce some new settings. 4.2.16 will not be able to
read them and will just ignore them. So this shouldn't be a problem.
Kind regards,
Frank
--
Dr.-Ing. Frank Mehnert | Software Development Director, VirtualBox
ORACLE Deutschland B.V. & Co. KG | Werkstr. 24 | 71384 Weinstadt, Germany
Hauptverwaltung: Riesstr. 25, D-80992 MÃŒnchen
Registergericht: Amtsgericht MÃŒnchen, HRA 95603
GeschÀftsfÌhrer: JÌrgen Kunz
KomplementÀrin: ORACLE Deutschland Verwaltung B.V.
Hertogswetering 163/167, 3543 AS Utrecht, Niederlande
Handelsregister der Handelskammer Midden-Niederlande, Nr. 30143697
GeschÀftsfÌhrer: Alexander van der Ven, Astrid Kepper, Val Maher
------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and
AppDynamics. Performance Central is your source for news, insights,
analysis and resources for efficient Application Performance Management.
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
VBox-users-community mailing list
https://lists.sourceforge.net/lists/listinfo/vbox-users-community
_______________________________________________
?subject=unsubscribe

[Eric S. Johansson]

On Wed, 21 Aug 2013 04:44:44 -0400, Frank Mehnert

Post by Frank Mehnert
for the extended announcement and for a preliminary list of all changes
since VirtualBox 4.2.16.

Frank, thank you again for your work on virtual box. it's my favorite go
to virtual machine environment, I've asked about this before but I'm
hoping you or one of your team can spare some cycles to look at this.

My goal is to run NaturallySpeaking in a windows virtual machine. Lots of
reasons why.

The problem with doing this in the past is that the USB bridge from the
native host to the virtual machine has been unkind to real-time audio.
Both usb playback and microphone input have been very choppy. Obviously
choppy audio input will ruin speech recognition accuracy.

It is possible to fix this, great, if not I can live without it. I have
some workarounds I've been using such as not using speech recognition on
the guest but instead using it only on the host.

[Fernando Cassia]

Post by Eric S. Johansson
The problem with doing this in the past is that the USB bridge from the
native host to the virtual machine has been unkind to real-time audio.

Have you tried a cheap $15 usb audio interface (with mic in, audio
out) assigned exclusively to the guest?

FC
--
During times of Universal Deceit, telling the truth becomes a revolutionary act
Durante épocas de Engaño Universal, decir la verdad se convierte en un
Acto Revolucionario
- George Orwell

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and
AppDynamics. Performance Central is your source for news, insights,
analysis and resources for efficient Application Performance Management.
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
VBox-users-community mailing list
VBox-users-***@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/vbox-users-community
_______________________________________________
Unsubscribe: mailto:vbox-users-community-request

[Eric S. Johansson]

Post by Fernando Cassia

Post by Eric S. Johansson
The problem with doing this in the past is that the USB bridge from the
native host to the virtual machine has been unkind to real-time audio.

Have you tried a cheap $15 usb audio interface (with mic in, audio
out) assigned exclusively to the guest?

not lately but I will give it a shot rsn (linux laptop HD just failed.
ugh)