[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] VGA passthrough with Xen 4.3 and xl toolstack - performance degradation resolved?


  • To: Gordan Bobic <gordan@xxxxxxxxxx>, David Sutton <kantras@xxxxxxxxx>
  • From: "H. Sieger" <powerhouse.linux@xxxxxxxxx>
  • Date: Wed, 12 Mar 2014 13:12:03 -0700 (PDT)
  • Cc: "xen-users@xxxxxxxxxxxxx" <xen-users@xxxxxxxxxxxxx>
  • Delivery-date: Wed, 12 Mar 2014 20:13:17 +0000
  • Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:X-Rocket-MIMEInfo:X-Mailer:References:Message-ID:Date:From:Reply-To:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type; b=pMaYXMcpKQ6rYMyfKIhafEPSMxinkDi+R0Slz5jfPYIkDRoPLVe23alTtyraVynUQ93batLhui/kPmWBS6sEdYsU9sD04tGcikRZVKzqeLzJU+cMywaVolNgnZoIK149NlOPD8kRYvLjm7iLuOqBYlnuLbva2u3Kb8LVYDs8avE=;
  • List-id: Xen user discussion <xen-users.lists.xen.org>

Now you really made me curious. I have two ATI cards (a Sapphire Radeon 7770 and a Sapphire Radeon 6450) in two different PCs and could use them for a test in my rig, running it from a Xen installation on an external USB drive so I won't have to compromise my regular system. Just need to find the time.

Your experience is also in contradiction with the Xen wiki: http://wiki.xen.org/wiki/Xen_VGA_Passthrough_Tested_Adapters.

I have collected user reports of successful VGA passthrough - see here: http://www.overclock.net/t/1307834/xen-vga-passthrough-compatible-graphics-adapters. There aren't too many posts there, some do mention guest reboot issues, but it seems there are few. Each card that's been listed should have a link to a user report.


On Wednesday, March 12, 2014 8:24 PM, Gordan Bobic <gordan@xxxxxxxxxx> wrote:
On 03/12/2014 06:14 PM, David Sutton wrote:
> Gordon,
>
> On Wed, Mar 12, 2014 at 12:01 PM, Gordan Bobic <gordan@xxxxxxxxxx
> <mailto:gordan@xxxxxxxxxx>> wrote:
>
>    On 03/12/2014 03:42 PM, H. Sieger wrote:
>
>        Thanks for the reply! Good to know the old ATI patch still works
>        with
>        Xen 4.4.
>
>
>    If you are talking about the patch I think you are talking about,
>    IIRC that was a patch for primary passthrough. This side-steps some
>    of the problems of lack of reset by POST-ing the VBIOS in domU, so
>    VBIOS gets the card into a state where the driver knows what to do
>    with it.
>
> The patch I was referring to is the same one that you are talking about;
> basically passing through the card as a primary display, with access to
> the VBIOS so the card can go through POST and be in a better internal
> state for when the driver starts to use it.

Thanks for clarifying. I was also referring to a much more recent patch
I saw floating around that adds bus reset as a means of resetting a PCI
device when FLR or power management based resetting isn't available.

The only ATI card I ever saw survive a reboot was a HD7450. It didn't
survive the reboot every time, but any number greater than 0 is a huge
improvement. It still wasn't reliable enough for "production" use, and I
needed something way more performant than a 7450 so I didn't stick with
it for more than preliminary testing, but I always assumed that the
reason it did better than higher end cards was because it was powered
only from the slot, no auxiliary power input - which meant that
resetting it via the power management trick (put it in powered off state
and then bring it back online). It's just a guess, though.


Gordan


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.