qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: geoff@hostfission.com
To: qemu-devel@nongnu.org
Subject: [Qemu-devel] PCI Passthrough + AMD + NPT
Date: Sun, 22 Oct 2017 21:47:57 +1100	[thread overview]
Message-ID: <088c92d5c3be0715c46f20e8b0ba0a0f@hostfission.com> (raw)

Hi All,

I have started to dig into why ntp seems to slow down graphics 
performance on
AMD systems using PCI passthrough and figured I would report what I have 
so far
discovered. I have noted the primary point of failure seems to be 
specifically
with PhysX. This is why people only see a slow down in certain games, 
not
everything uses PhysX.

Using FluidMark[1] the problem is immediately obvious, showing extremely 
low
FPS on light/medium workloads with ntp enabled, and extreme fluididy and 
high
FPS with ntp disabled.

Switching nVidia to use CPU makes no difference to the performance when 
ntp
is enabled, which seems to indicate that PhysX is falling back to CPU 
due to
a failure of some kind to initialize.

With ntp turned off, and nVidia set to use the CPU for PhysX I see an
identical performance drop off in FluidMark as I see when ntp is 
enabled, this
would seem to confirm this suspicion.

Since other features such as APIC is only available if ntp is enabled, 
it
could be something down stream of ntp that is getting disabled as a
consequence of turning off ntp. It might be interesting to see if we can 
get
some diagnostics information out of PhysX to see what if any error or 
debugging
information it might provide when it falls back to CPU.

1: 
http://www.geeks3d.com/20130308/fluidmark-1-5-1-physx-benchmark-fluid-sph-simulation-opengl-download/

Kind Regards,
Geoffrey McRae

                 reply	other threads:[~2017-10-22 14:23 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=088c92d5c3be0715c46f20e8b0ba0a0f@hostfission.com \
    --to=geoff@hostfission.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).