VDPAU testing tool

[Stephen Warren]

For the NTSC output that's reported as 50Hz, I imagine it's most likely because of the following FAQ:

Why is the refresh rate not reported correctly by utilities that use the XRandR X extension (e.g., the GNOME "Screen Resolution Preferences" panel, `xrandr -q`, etc)?


The XRandR X extension is not presently aware of multiple display devices on a single X screen; it only sees the MetaMode bounding box, which may contain one or more actual modes. This means that if multiple MetaModes have the same bounding box, XRandR will not be able to distinguish between them.

In order to support DynamicTwinView, the NVIDIA X driver must make each MetaMode appear to be unique to XRandR. Presently, the NVIDIA X driver accomplishes this by using the refresh rate as a unique identifier.

You can use `nvidia-settings -q RefreshRate` to query the actual refresh rate on each display device.

This behavior can be disabled by setting the X configuration option "DynamicTwinView" to FALSE.

From: http://us.download.nvidia.com/XFree8...hapter-07.html

[Enigma2175]

Originally Posted by Stephen Warren

For the NTSC output that's reported as 50Hz, I imagine it's most likely because of the following FAQ:

Why is the refresh rate not reported correctly by utilities that use the XRandR X extension (e.g., the GNOME "Screen Resolution Preferences" panel, `xrandr -q`, etc)?


The XRandR X extension is not presently aware of multiple display devices on a single X screen; it only sees the MetaMode bounding box, which may contain one or more actual modes. This means that if multiple MetaModes have the same bounding box, XRandR will not be able to distinguish between them.

In order to support DynamicTwinView, the NVIDIA X driver must make each MetaMode appear to be unique to XRandR. Presently, the NVIDIA X driver accomplishes this by using the refresh rate as a unique identifier.

You can use `nvidia-settings -q RefreshRate` to query the actual refresh rate on each display device.

This behavior can be disabled by setting the X configuration option "DynamicTwinView" to FALSE.

From: http://us.download.nvidia.com/XFree8...hapter-07.html

Thanks Stephen, I was indeed using xrandr to determine the refresh rate. Using nvidia-settings shows the refresh rate to be 59.94, which is what I would expect from NTSC.

[Zooloo]

Hello,

I would like to thank crisalide for this useful tool, and also contribute myself a bit in making it easily installable under Gentoo. Therefore I wrote an ebuild for it, as I do not install anything unmanaged on my Gentoo systems. So for those who want to try it in a local overlay, paste the code below as qvdpautest-0.5.ebuild under the category x11-misc, generate the manifest, keyword if necessary and give it a go:

# Copyright 1999-2010 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: $

EAPI=2

inherit qt4-r2

DESCRIPTION="VDPAU benchmarking tool"

HOMEPAGE="http://www.nvnews.net/vbulletin/showthread.php?t=133465"

SRC_URI="http://hftom.free.fr/${P}.tar.gz"

LICENSE="LGPL-2.1"

SLOT="0"

KEYWORDS="~x86 ~amd64"

IUSE=""

DEPEND="x11-libs/libvdpau
        x11-libs/qt-gui:4"

RDEPEND="${DEPEND}"

S="${WORKDIR}/${PN}"

src_prepare() {
        cd ${S}
        echo "#!/bin/bash" > ${PN}.sh
        echo '' >> ${PN}.sh
        echo 'CURRDIR="$(pwd)"' >> ${PN}.sh
        echo "cd /usr/share/${PN}" >> ${PN}.sh
        echo "./${PN}-exec \"\$@\"" >> ${PN}.sh
        echo 'cd "${CURRDIR}"' >> ${PN}.sh
}

src_install() {
        newbin ${PN}.sh ${PN}
        dodoc README
        exeinto /usr/share/${PN}
        newexe ${PN} ${PN}-exec
        insinto /usr/share/${PN}
        doins *.dat
}

@crisalide: I'd have a question, would you mind if I submit the ebuild to the official Gentoo bugtracker? For that you should also specify the license model under which you're publishing your code, for now I just assumed LGPL-2.1 as you can see in the ebuild, but you're the one to decide this.
As for installing in a distribution, I had to overcome the fact that your utility only loads the test data from the current directory by installing the executable and the data files into /usr/share/qvdpautest, renaming the executable into qvdpautest-exec and explicitly calling that renamed executable by the batch the Gentoo ebuild generates and actually installs it as /usr/bin/qvdpautest:

#!/bin/bash

CURRDIR="$(pwd)"
cd /usr/share/qvdpautest
./qvdpautest-exec "$@"
cd "${CURRDIR}"

So, the currently installed files under Gentoo, are the following:

/usr/bin/qvdpautest
/usr/share/doc/qvdpautest-0.5/README.bz2
/usr/share/qvdpautest/h264720p.dat
/usr/share/qvdpautest/h264hd.dat
/usr/share/qvdpautest/mpeg4hd.dat
/usr/share/qvdpautest/mpg720p.dat
/usr/share/qvdpautest/mpghd.dat
/usr/share/qvdpautest/mpgsd.dat
/usr/share/qvdpautest/qvdpautest-exec
/usr/share/qvdpautest/vc1hd.dat

You may want to address these things in a future version. Other than that, thanks again for this tool, it does a good job!

Kind regards,
Lucian

[Gillian00]

Genuine Intel(R) CPU @ 2.40GHz
NVIDIA GPU GeForce 9500M GS (G84) at PCI:1:0:0 (GPU-0)

VDPAU API version : 1
VDPAU implementation : NVIDIA VDPAU Driver Shared Library 195.36.03 Mon Feb 1 19:22:18 PST 2010

SURFACE GET BITS: 850.622 M/s
SURFACE PUT BITS: 703.213 M/s

MPEG DECODING (1920x1080): 72 frames/s
MPEG DECODING (1280x720): 146 frames/s
H264 DECODING (1920x1080): 42 frames/s
H264 DECODING (1280x720): 91 frames/s
VC1 DECODING (1440x1080): 109 frames/s

MIXER WEAVE (1920x1080): 612 frames/s
MIXER BOB (1920x1080): 999 fields/s
MIXER TEMPORAL (1920x1080): 230 fields/s
MIXER TEMPORAL + IVTC (1920x1080): 143 fields/s
MIXER TEMPORAL + SKIP_CHROMA (1920x1080): 310 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 75 fields/s
MIXER TEMPORAL_SPATIAL + IVTC (1920x1080): 61 fields/s
MIXER TEMPORAL_SPATIAL + SKIP_CHROMA (1920x1080): 83 fields/s
MIXER TEMPORAL_SPATIAL (720x576 video to 1920x1080 display): 283 fields/s

MULTITHREADED MPEG DECODING (1920x1080): 72 frames/s
MULTITHREADED MIXER TEMPORAL (1920x1080): 201 fields/s

9500m gs at 500/1000/475

quite low isn't it? Is it normal?

[primerib]

Originally Posted by Stephen Warren

For the NTSC output that's reported as 50Hz, I imagine it's most likely because of the following FAQ:

Why is the refresh rate not reported correctly by utilities that use the XRandR X extension (e.g., the GNOME "Screen Resolution Preferences" panel, `xrandr -q`, etc)?


The XRandR X extension is not presently aware of multiple display devices on a single X screen; it only sees the MetaMode bounding box, which may contain one or more actual modes. This means that if multiple MetaModes have the same bounding box, XRandR will not be able to distinguish between them.

In order to support DynamicTwinView, the NVIDIA X driver must make each MetaMode appear to be unique to XRandR. Presently, the NVIDIA X driver accomplishes this by using the refresh rate as a unique identifier.

You can use `nvidia-settings -q RefreshRate` to query the actual refresh rate on each display device.

This behavior can be disabled by setting the X configuration option "DynamicTwinView" to FALSE.

From: http://us.download.nvidia.com/XFree8...hapter-07.html

I have that problem as well since I use xrandr to check resolution & refresh. What "section" of xorg.conf should DynamicTwinView be set in? (You may want to add that information to your faq btw.)

[crisalide]

Originally Posted by Gillian00

9500m gs at 500/1000/475

quite low isn't it? Is it normal?

Yes, it is, for this GPU.

[crisalide]

Originally Posted by Zooloo

@crisalide: I'd have a question, would you mind if I submit the ebuild to the official Gentoo bugtracker? For that you should also specify the license model under which you're publishing your code, for now I just assumed LGPL-2.1 as you can see in the ebuild, but you're the one to decide this.
As for installing in a distribution, I had to overcome the fact that your utility only loads the test data from the current directory by installing the executable and the data files into /usr/share/qvdpautest, renaming the executable into qvdpautest-exec and explicitly calling that renamed executable by the batch the Gentoo ebuild generates and actually installs it as /usr/bin/qvdpautest

This tool isn't meant to be installed system wide nor to be packaged. Some samples may be copyrighted (even if i doubt anyone would complain). Target users are expected to have the skills to compile it. But feel free to do what you want with it.

[cehoyos]

Originally Posted by Gillian00

9500m gs at 500/1000/475

quite low isn't it? Is it normal?

Low?
You can decode H264 1080p30 material with your hardware, and you can use temporal de-interlacing for every content. What else do you expect (or need)?

Carl Eugen

[Gillian00]

I don't know, it's just by looking at results of other people, the number of frames and other stuff is low.

will it be ok to run a 1080p movie with a hd screen with this config ?? no lag ??

[cehoyos]

Originally Posted by Gillian00

will it be ok to run a 1080p movie with a hd screen with this config ?? no lag ??

That depends:
I am sure you can play H.264 1080p30 (This is not used in Europe, but I saw some uploaded samples from television).
Playback of H.264 1080p50 (produced by some xacti camera) is not possible (in real-time) with any current Nvidia hardware (but I could not find a GT 240 benchmark, so who knows).

Carl Eugen

[crisalide]

Originally Posted by cehoyos

Playback of H.264 1080p50 (produced by some xacti camera) is not possible (in real-time) with any current Nvidia hardware (but I could not find a GT 240 benchmark, so who knows).
Carl Eugen

Never tried such 1080p50 sample, but at least a GT220 is able to play two 1080i50 (@ 10650kb/s) at the same time.
Is anyone able to provide such a 1080p50 sample for benchmark?

[cehoyos]

Originally Posted by crisalide

Never tried such 1080p50 sample, but at least a GT220 is able to play two 1080i50 (@ 10650kb/s) at the same time.

You mean 1080i25?
25 interlaced frames per second with a horizontal resolution of 1080 lines.

Is anyone able to provide such a 1080p50 sample for benchmark?

The dog sample on samples.mplayerhq is the one mentioned in the long MPlayer VDPAU thread.
Is there nobody with a GT 240 who could run the testing tool?

Carl Eugen