[gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Walter Dnes]

  With the recent speed bump on my ADSL service from 5 megabps to 6
(don't laugh), I can now download 1080p Youtube Flash videos almost fast
enough to keep up.  E.g. a 20 or 25 second headstart will allow me to
play a 5 minute video before it has to buffer.  On some html5 videos
(Firefox with USE="webm"), The download is actually a touch faster than
the playback, and there's no buffering at all.

  Some of you may remember my struggles to get my 4-year-old Dell to
eventually display hockey games on NHL GameCenter even at the lowest
available speed using the onboard Intel GPU.  Well, I can play the HD
Youtube videos with the "small player" or "large player", but fullscreen
is hopeless.  The onboard GPU can't keep up.  So I'm looking at getting
a PCI video card.  Any relatively new PCI video card that is supported
by an open-source driver, including hardware acceleration?  Any
experiences, good/bad/so-so?

-- 
Walter Dnes <waltdnes@waltdnes.org>

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Dale]

Walter Dnes wrote:
>   With the recent speed bump on my ADSL service from 5 megabps to 6
> (don't laugh), 


I get under 1Mb/sec so I won't laugh.  I may cry tho.  :-(

Then again, this beats dial-up big time.  ;-)

Dale

:-)  :-)


-- 
I am only responsible for what I said ... Not for what you understood or
how you interpreted my words!

Miss the compile output?  Hint:
EMERGE_DEFAULT_OPTS="--quiet-build=n"

[gentoo-user] Re: PCI video cards, hardware accel, upported by open-source drivers?

[Nikos Chantziaras]

On 10/04/12 14:26, Walter Dnes wrote:
>    With the recent speed bump on my ADSL service from 5 megabps to 6
> (don't laugh), I can now download 1080p Youtube Flash videos almost fast
> enough to keep up.  E.g. a 20 or 25 second headstart will allow me to
> play a 5 minute video before it has to buffer.  On some html5 videos
> (Firefox with USE="webm"), The download is actually a touch faster than
> the playback, and there's no buffering at all.
>
>    Some of you may remember my struggles to get my 4-year-old Dell to
> eventually display hockey games on NHL GameCenter even at the lowest
> available speed using the onboard Intel GPU.  Well, I can play the HD
> Youtube videos with the "small player" or "large player", but fullscreen
> is hopeless.  The onboard GPU can't keep up.  So I'm looking at getting
> a PCI video card.  Any relatively new PCI video card that is supported
> by an open-source driver, including hardware acceleration?  Any
> experiences, good/bad/so-so?

This is a CPU problem, not GPU.  Try to install 
media-video/smplayer-0.8.0 (older versions don't support YouTube), and 
open the YouTube video link in it.  In the preferences ("performance" 
section) you can select the quality at which to open the videos.

Re: [gentoo-user] Re: PCI video cards, hardware accel, upported by open-source drivers?

[Michael Mol]

On Tue, Apr 10, 2012 at 10:08 AM, Nikos Chantziaras <realnc@gmail.com> wrote:
> On 10/04/12 14:26, Walter Dnes wrote:
>>
>>   With the recent speed bump on my ADSL service from 5 megabps to 6
>> (don't laugh), I can now download 1080p Youtube Flash videos almost fast
>> enough to keep up.  E.g. a 20 or 25 second headstart will allow me to
>> play a 5 minute video before it has to buffer.  On some html5 videos
>> (Firefox with USE="webm"), The download is actually a touch faster than
>> the playback, and there's no buffering at all.
>>
>>   Some of you may remember my struggles to get my 4-year-old Dell to
>> eventually display hockey games on NHL GameCenter even at the lowest
>> available speed using the onboard Intel GPU.  Well, I can play the HD
>> Youtube videos with the "small player" or "large player", but fullscreen
>> is hopeless.  The onboard GPU can't keep up.  So I'm looking at getting
>> a PCI video card.  Any relatively new PCI video card that is supported
>> by an open-source driver, including hardware acceleration?  Any
>> experiences, good/bad/so-so?
>
>
> This is a CPU problem, not GPU.  Try to install media-video/smplayer-0.8.0
> (older versions don't support YouTube), and open the YouTube video link in
> it.  In the preferences ("performance" section) you can select the quality
> at which to open the videos.

Yes and no. You can use GPU acceleration for video decoding. I'd
suggest the low-end nVidia GeForce cards, as any nVidia card from the
last couple years will do hardware h264 decode (which even Linux
versions of Flash will take advantage of, now), but I don't think the
novou drivers implement vdpau support yet.

My favorite is the nVidia GeForce 210; cheap and effective. I picked
up a couple of them retail for $50 USD. At the time, I think there
were PCI versions available, but I don't know if that's still true.

-- 
:wq

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Stroller]

On 10 April 2012, at 12:26, Walter Dnes wrote:
>  With the recent speed bump on my ADSL service from 5 megabps to 6
> (don't laugh), I can now download 1080p Youtube Flash videos almost fast
> enough to keep up.  E.g. a 20 or 25 second headstart will allow me to
> play a 5 minute video before it has to buffer.  On some html5 videos
> (Firefox with USE="webm"), The download is actually a touch faster than
> the playback, and there's no buffering at all.
> 
>  Some of you may remember my struggles to get my 4-year-old Dell to
> eventually display hockey games on NHL GameCenter even at the lowest
> available speed using the onboard Intel GPU.  Well, I can play the HD
> Youtube videos with the "small player" or "large player", but fullscreen
> is hopeless.

Have you tried net-misc/youtube-dl ?

This may not be a direct answer to your question, but youtube-dl allows you to select the video's encoding and quality (-f and -F options) and just grab the video as an MP4 or .flv file.

You may find the video plays more smoothly without the overhead (??) of being played in a browser (or by a browser plig-in). When you've downloaded with youtube-dl you can then just double-click on it and open in mplayer or vlc - at least you get the choice of video players that way, and you may find one of them smoother and less sputtery.

I'm sceptical over the benefits of upgrading a 4 year old PC (short of ripping most all the guts out and starting again). I know the industry has currently settled on PCIe, but haven't bus speeds increased in the last 4 years? Are all the latest cards compatible with your Dell? If not, then you'll probably end up buying an older model, and then that will be sub-optimal when you want to upgrade your motherboard in a year's time.

I'm sorry if this reply is unhelpful, but you give a lot of information, and perhaps that means you might be open to considering alternative solutions to the core problem.

Stroller.

[gentoo-user] Re: PCI video cards, hardware accel, upported by open-source drivers?

[Nikos Chantziaras]

On 10/04/12 17:19, Michael Mol wrote:
> On Tue, Apr 10, 2012 at 10:08 AM, Nikos Chantziaras<realnc@gmail.com>  wrote:
>> On 10/04/12 14:26, Walter Dnes wrote:
>>>
>>>    With the recent speed bump on my ADSL service from 5 megabps to 6
>>> (don't laugh), I can now download 1080p Youtube Flash videos almost fast
>>> enough to keep up.  E.g. a 20 or 25 second headstart will allow me to
>>> play a 5 minute video before it has to buffer.  On some html5 videos
>>> (Firefox with USE="webm"), The download is actually a touch faster than
>>> the playback, and there's no buffering at all.
>>>
>>>    Some of you may remember my struggles to get my 4-year-old Dell to
>>> eventually display hockey games on NHL GameCenter even at the lowest
>>> available speed using the onboard Intel GPU.  Well, I can play the HD
>>> Youtube videos with the "small player" or "large player", but fullscreen
>>> is hopeless.  The onboard GPU can't keep up.  So I'm looking at getting
>>> a PCI video card.  Any relatively new PCI video card that is supported
>>> by an open-source driver, including hardware acceleration?  Any
>>> experiences, good/bad/so-so?
>>
>>
>> This is a CPU problem, not GPU.  Try to install media-video/smplayer-0.8.0
>> (older versions don't support YouTube), and open the YouTube video link in
>> it.  In the preferences ("performance" section) you can select the quality
>> at which to open the videos.
>
> Yes and no. You can use GPU acceleration for video decoding.

Not with Adobe Flash on an Intel GPU.  His problem is that Flash uses 
way too much CPU, and mplayer (which SMPlayer is using) does not.  It's 
really a CPU problem.

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Walter Dnes]

On Tue, Apr 10, 2012 at 06:45:55PM +0100, Stroller wrote

> I'm sceptical over the benefits of upgrading a 4 year old PC (short
> of ripping most all the guts out and starting again). I know the
> industry has currently settled on PCIe, but haven't bus speeds
> increased in the last 4 years? Are all the latest cards compatible
> with your Dell? If not, then you'll probably end up buying an older
> model, and then that will be sub-optimal when you want to upgrade
> your motherboard in a year's time.

  I've posted a snapshot of the Dell's internals on my ISP's personal
webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
the long black slot PCIe?  What's the short black slot?

> I'm sorry if this reply is unhelpful, but you give a lot of information,
> and perhaps that means you might be open to considering alternative
> solutions to the core problem.

  If it's PCIe, so be it.  Actually, a post that prevents me wasting
money is helpful <G>.  Would PCIe be significantly better on the same
CPU+GPU, or is it hype?

-- 
Walter Dnes <waltdnes@waltdnes.org>

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Paul Hartman]

On Wed, Apr 11, 2012 at 1:11 AM, Walter Dnes <waltdnes@waltdnes.org> wrote:
>  I've posted a snapshot of the Dell's internals on my ISP's personal
> webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
> the long black slot PCIe?  What's the short black slot?

If you look closely, they are labeled on the motherboard itself. :)

The white slots are 5-volt 32-bit conventional PCI slots.

The long black slot is a PCI Express x16, while the short black slot
is a PCI Express x1.

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Pandu Poluan]

[-- Attachment #1: Type: text/plain, Size: 1574 bytes --]

On Apr 11, 2012 1:15 PM, "Walter Dnes" <waltdnes@waltdnes.org> wrote:
>
> On Tue, Apr 10, 2012 at 06:45:55PM +0100, Stroller wrote
>
> > I'm sceptical over the benefits of upgrading a 4 year old PC (short
> > of ripping most all the guts out and starting again). I know the
> > industry has currently settled on PCIe, but haven't bus speeds
> > increased in the last 4 years? Are all the latest cards compatible
> > with your Dell? If not, then you'll probably end up buying an older
> > model, and then that will be sub-optimal when you want to upgrade
> > your motherboard in a year's time.
>
>  I've posted a snapshot of the Dell's internals on my ISP's personal
> webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
> the long black slot PCIe?  What's the short black slot?
>

The long black slot looks like PCIe. To be precise, PCIe x16. The short
black slot is PCIe x1, (originally) meant for low-bandwidth devices like a
fax modem.

> > I'm sorry if this reply is unhelpful, but you give a lot of information,
> > and perhaps that means you might be open to considering alternative
> > solutions to the core problem.
>
>  If it's PCIe, so be it.  Actually, a post that prevents me wasting
> money is helpful <G>.  Would PCIe be significantly better on the same
> CPU+GPU, or is it hype?
>

For games with huge 3D textures, absolutely. For video playback, not so
much.

But the main point would be that the newest graphics cards are all released
in PCIe version only, and future mobos will all support PCIe, so it's a
future-safe investment.

Rgds,

[-- Attachment #2: Type: text/html, Size: 2026 bytes --]

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Volker Armin Hemmann]

Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
> On Tue, Apr 10, 2012 at 06:45:55PM +0100, Stroller wrote
> 
> > I'm sceptical over the benefits of upgrading a 4 year old PC (short
> > of ripping most all the guts out and starting again). I know the
> > industry has currently settled on PCIe, but haven't bus speeds
> > increased in the last 4 years? Are all the latest cards compatible
> > with your Dell? If not, then you'll probably end up buying an older
> > model, and then that will be sub-optimal when you want to upgrade
> > your motherboard in a year's time.
> 
>   I've posted a snapshot of the Dell's internals on my ISP's personal
> webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
> the long black slot PCIe?  What's the short black slot?
> 
> > I'm sorry if this reply is unhelpful, but you give a lot of information,
> > and perhaps that means you might be open to considering alternative
> > solutions to the core problem.
> 
>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
> money is helpful <G>.  Would PCIe be significantly better on the same
> CPU+GPU, or is it hype?

a lot, lot lot lot better. No hype.
-- 


#163933

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Volker Armin Hemmann]

Am Mittwoch, 11. April 2012, 13:51:28 schrieb Pandu Poluan:
> On Apr 11, 2012 1:15 PM, "Walter Dnes" <waltdnes@waltdnes.org> wrote:
> > On Tue, Apr 10, 2012 at 06:45:55PM +0100, Stroller wrote
> > 
> > > I'm sceptical over the benefits of upgrading a 4 year old PC (short
> > > of ripping most all the guts out and starting again). I know the
> > > industry has currently settled on PCIe, but haven't bus speeds
> > > increased in the last 4 years? Are all the latest cards compatible
> > > with your Dell? If not, then you'll probably end up buying an older
> > > model, and then that will be sub-optimal when you want to upgrade
> > > your motherboard in a year's time.
> >  
> >  I've posted a snapshot of the Dell's internals on my ISP's personal
> > 
> > webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
> > the long black slot PCIe?  What's the short black slot?
> 
> The long black slot looks like PCIe. To be precise, PCIe x16. The short
> black slot is PCIe x1, (originally) meant for low-bandwidth devices like a
> fax modem.

oh so wrong. Even a single PCIe lane is faster than an entire PCI bus.

More like SATA controllers, usb-3.0 controllers, high end sound cards.

For slow crap you have usb.

> 
> For games with huge 3D textures, absolutely. For video playback, not so
> much.
> 

em, just compare a 1080p with amd+working va-api backend in vlc and without. 
Huge difference. 

> But the main point would be that the newest graphics cards are all released
> in PCIe version only, and future mobos will all support PCIe, so it's a
> future-safe investment.

and all current. Really, can you even get agp based boards anymore? agp is 
dead. PCI is as good as dead...
,
-- 
#163933

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Michael Mol]

On Wed, Apr 11, 2012 at 12:30 PM, Volker Armin Hemmann
<volkerarmin@googlemail.com> wrote:
> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
>> On Tue, Apr 10, 2012 at 06:45:55PM +0100, Stroller wrote
>>
>> > I'm sceptical over the benefits of upgrading a 4 year old PC (short
>> > of ripping most all the guts out and starting again). I know the
>> > industry has currently settled on PCIe, but haven't bus speeds
>> > increased in the last 4 years? Are all the latest cards compatible
>> > with your Dell? If not, then you'll probably end up buying an older
>> > model, and then that will be sub-optimal when you want to upgrade
>> > your motherboard in a year's time.
>>
>>   I've posted a snapshot of the Dell's internals on my ISP's personal
>> webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
>> the long black slot PCIe?  What's the short black slot?
>>
>> > I'm sorry if this reply is unhelpful, but you give a lot of information,
>> > and perhaps that means you might be open to considering alternative
>> > solutions to the core problem.
>>
>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
>> money is helpful <G>.  Would PCIe be significantly better on the same
>> CPU+GPU, or is it hype?
>
> a lot, lot lot lot better. No hype.

One thing worth noting about the difference between PCI and PCIe:

PCI typically has all devices on the same bus, or on bridged buses.
Traffic from one device at a particular instant means another device
can't communicate until that first device is done. That means your PCI
video card competes with your PCI hard disk controller and your PCI
USB card for bandwidth. Two high-throughput devices like DMA-enabled
video cards and disk controllers will get in each others' way.

With PCIe, the data channels (called lanes) are electrically distinct;
your video card and RAID card both communicate directly with the PCIe
controller, and don't have to wait for a clear channel from each other
before they can talk.

*Logically*, PCIe looks like PCI when you run lspci or similar. At the
hardware and electrical levels, though, they're quite different.

-- 
:wq

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Florian Philipp]

[-- Attachment #1: Type: text/plain, Size: 985 bytes --]

Am 11.04.2012 18:33, schrieb Volker Armin Hemmann:
> Am Mittwoch, 11. April 2012, 13:51:28 schrieb Pandu Poluan:
>>
[...]
>>
>> But the main point would be that the newest graphics cards are all released
>> in PCIe version only, and future mobos will all support PCIe, so it's a
>> future-safe investment.
> 
> and all current. Really, can you even get agp based boards anymore? agp is 
> dead. PCI is as good as dead...
> ,

OT:

I still have use for PCI graphics: I needed to refit two PCs with dual
port PCIe ethernet adapters. These only come with 4x PCie and many
boards only have 1x PCIe slots (aside of their PEG port). So I used an
old PCI graphics card and put the network adapter in the PEG slot.

With one board (cheap MSI) it worked; the other one (expensive Intel)
didn't work because the chipset apparently detected the network adapter
as a graphics adapter even though PCI had higher priority in the BIOS
settings.

Regards,
Florian Philipp


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Stroller]

On 11 April 2012, at 17:33, Volker Armin Hemmann wrote:
>>> 
>>> I've posted a snapshot of the Dell's internals on my ISP's personal
>>> 
>>> webspace at http://clients.teksavvy.com/~walterdnes/misc/dell2.jpg  Is
>>> the long black slot PCIe?  What's the short black slot?
>> 
>> The long black slot looks like PCIe. To be precise, PCIe x16. The short
>> black slot is PCIe x1, (originally) meant for low-bandwidth devices like a
>> fax modem.
> 
> oh so wrong. Even a single PCIe lane is faster than an entire PCI bus.

I think the confusion here may lie in the fact that (some?) motherboard manufacturers used to make motherboards with an onboard modem; they had the socket on a daughterboard so that OEMs could choose whether or not to install it.

The ones I'm thinking of used used a little socket which looks exactly like the small PCIe connector - I think it shared space with a PCI socket (so the OEM could choose to advertise either the modem or X+1 PCI slots).

Stroller.

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Walter Dnes]

On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
> > 
> >   If it's PCIe, so be it.  Actually, a post that prevents me wasting
> > money is helpful <G>.  Would PCIe be significantly better on the same
> > CPU+GPU, or is it hype?
> 
> a lot, lot lot lot better. No hype.

  I've done some looking, and I'm back with more questions.  I've also
read the Nouveau-versus-NVIDIA thread.  Questions...

1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
performance, I just want full backwards compatability.  PCIe 1.0 cards
seem to be rare, and have to be ordered online, while I can pick up a
2.0 card locally at a store.

2) My main "torture test" will be HD fullscreen video.  Will there be
major improvement in that?  That's 2D.  Forget 3D.

3) The 2 lowest-priced Nvidia's at the local store are...

Asus EN210 SILENT/DI/512MD3/V2(LP) NVIDIA GeForce 210 Chipset (589Mhz)
512MB (1333Mhz) GDDR3 DVI/VGA/HDMI PCI-Express 2.0 Graphics Card


and EVGA GeForce 8400GS 1GB (01G-P3-1302-LR) nVidia GeForce 8400GS
Chipset (520Mhz) 1GB (520Mhz) DDR3 Dual Display DVI/HDMI/VGA PCI Express
2.0 Graphics Card

  Any preferences?

-- 
Walter Dnes <waltdnes@waltdnes.org>

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Drew]

> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
> performance, I just want full backwards compatability.  PCIe 1.0 cards
> seem to be rare, and have to be ordered online, while I can pick up a
> 2.0 card locally at a store.

They should work just fine. PCIe gen1/2/3 slots will work just fine
with gen1/2/3 cards, the speed will just be limited to the slower of
the two parts. Kinda like plugging an USB2 device into an older 1.1
port. It'll work but will be slower.

From personal experience working with a mix of Gen1 & Gen2 devices &
slots in high end workstations/servers, you won't really notice the
performance hit of running a gen2 Video Card in a gen1 slot unless you
play the latest 3D games or run some really high end 3D apps. Only
place I really noticed performance differences was in running higher
end RAID cards & NICs. Some can max a gen2 x8 slot and running on a
gen1 you can see the performance hit.

But I digress. ;-)


-- 
Drew

"This started out as a hobby and spun horribly out of control."
-Unknown

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Florian Philipp]

[-- Attachment #1: Type: text/plain, Size: 1673 bytes --]

Am 15.04.2012 15:18, schrieb Walter Dnes:
> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
>>>
>>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
>>> money is helpful <G>.  Would PCIe be significantly better on the same
>>> CPU+GPU, or is it hype?
>>
>> a lot, lot lot lot better. No hype.
> 
>   I've done some looking, and I'm back with more questions.  I've also
> read the Nouveau-versus-NVIDIA thread.  Questions...
> 
> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
> performance, I just want full backwards compatability.  PCIe 1.0 cards
> seem to be rare, and have to be ordered online, while I can pick up a
> 2.0 card locally at a store.
> 

PCIe-2.0 is fully downward compatible to 1.1 and 1.0.

> 2) My main "torture test" will be HD fullscreen video.  Will there be
> major improvement in that?  That's 2D.  Forget 3D.
> 

2D video is still rendered using OpenGL if your video player supports it.

> 3) The 2 lowest-priced Nvidia's at the local store are...
> 
> Asus EN210 SILENT/DI/512MD3/V2(LP) NVIDIA GeForce 210 Chipset (589Mhz)
> 512MB (1333Mhz) GDDR3 DVI/VGA/HDMI PCI-Express 2.0 Graphics Card
> 
> 
> and EVGA GeForce 8400GS 1GB (01G-P3-1302-LR) nVidia GeForce 8400GS
> Chipset (520Mhz) 1GB (520Mhz) DDR3 Dual Display DVI/HDMI/VGA PCI Express
> 2.0 Graphics Card
> 
>   Any preferences?
> 

1GB of GPU RAM looks like overkill if you just want to watch videos. I
suggest selecting the model which is cheaper, has more connectors or is
more silent.

Regards,
Florian Philipp


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Volker Armin Hemmann]

Am Sonntag, 15. April 2012, 09:18:15 schrieb Walter Dnes:
> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
> 
> > Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
> > >   If it's PCIe, so be it.  Actually, a post that prevents me wasting
> > > 
> > > money is helpful <G>.  Would PCIe be significantly better on the same
> > > CPU+GPU, or is it hype?
> > 
> > a lot, lot lot lot better. No hype.
> 
>   I've done some looking, and I'm back with more questions.  I've also
> read the Nouveau-versus-NVIDIA thread.  Questions...
> 
> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot? 

yes

> 
> 2) My main "torture test" will be HD fullscreen video.  Will there be
> major improvement in that?  That's 2D.  Forget 3D.
> 

if you get va-api working. Is it mpeg4?

> 3) The 2 lowest-priced Nvidia's at the local store are...
> 
> Asus EN210 SILENT/DI/512MD3/V2(LP) NVIDIA GeForce 210 Chipset (589Mhz)
> 512MB (1333Mhz) GDDR3 DVI/VGA/HDMI PCI-Express 2.0 Graphics Card
> 

you could also stay with onboard graphics.

> 
> and EVGA GeForce 8400GS 1GB (01G-P3-1302-LR) nVidia GeForce 8400GS
> Chipset (520Mhz) 1GB (520Mhz) DDR3 Dual Display DVI/HDMI/VGA PCI Express
> 2.0 Graphics Card

you should stay with onboard graphics.

> 
>   Any preferences?

yes, AMD, they support open source.

also, va-api does work.

-- 
#163933

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Michael Mol]

On Sun, Apr 15, 2012 at 10:09 AM, Florian Philipp <lists@binarywings.net> wrote:
> Am 15.04.2012 15:18, schrieb Walter Dnes:
>> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
>>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
>>>>
>>>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
>>>> money is helpful <G>.  Would PCIe be significantly better on the same
>>>> CPU+GPU, or is it hype?
>>>
>>> a lot, lot lot lot better. No hype.
>>
>>   I've done some looking, and I'm back with more questions.  I've also
>> read the Nouveau-versus-NVIDIA thread.  Questions...
>>
>> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
>> performance, I just want full backwards compatability.  PCIe 1.0 cards
>> seem to be rare, and have to be ordered online, while I can pick up a
>> 2.0 card locally at a store.
>>
>
> PCIe-2.0 is fully downward compatible to 1.1 and 1.0.
>
>> 2) My main "torture test" will be HD fullscreen video.  Will there be
>> major improvement in that?  That's 2D.  Forget 3D.
>>
>
> 2D video is still rendered using OpenGL if your video player supports it.

I'm not aware of any video decoders using CUDA, OpenCL, or pixel
shaders for video decoding; AFAIK, unless you're using VDPAU you're
still using the CPU to render the video to a frame buffer. The most a
video player is going to use OpenGL for is stretching that frame
buffer to fit a window or screen, and possibly as a compositor to
place overlays like subtitles or playback control elements..

>
>> 3) The 2 lowest-priced Nvidia's at the local store are...
>>
>> Asus EN210 SILENT/DI/512MD3/V2(LP) NVIDIA GeForce 210 Chipset (589Mhz)
>> 512MB (1333Mhz) GDDR3 DVI/VGA/HDMI PCI-Express 2.0 Graphics Card
>>
>>
>> and EVGA GeForce 8400GS 1GB (01G-P3-1302-LR) nVidia GeForce 8400GS
>> Chipset (520Mhz) 1GB (520Mhz) DDR3 Dual Display DVI/HDMI/VGA PCI Express
>> 2.0 Graphics Card
>>
>>   Any preferences?
>>
>
> 1GB of GPU RAM looks like overkill if you just want to watch videos. I
> suggest selecting the model which is cheaper, has more connectors or is
> more silent.

Yeah, RAM isn't going to be the issue here.

Honestly, this is the card I'd recommend if you're not doing any heavy gaming.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125402

It's cheap, dead quiet and has full hardware decode of h.264.

-- 
:wq

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Florian Philipp]

[-- Attachment #1: Type: text/plain, Size: 2006 bytes --]

Am 15.04.2012 16:22, schrieb Michael Mol:
> On Sun, Apr 15, 2012 at 10:09 AM, Florian Philipp <lists@binarywings.net> wrote:
>> Am 15.04.2012 15:18, schrieb Walter Dnes:
>>> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
>>>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
>>>>>
>>>>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
>>>>> money is helpful <G>.  Would PCIe be significantly better on the same
>>>>> CPU+GPU, or is it hype?
>>>>
>>>> a lot, lot lot lot better. No hype.
>>>
>>>   I've done some looking, and I'm back with more questions.  I've also
>>> read the Nouveau-versus-NVIDIA thread.  Questions...
>>>
>>> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
>>> performance, I just want full backwards compatability.  PCIe 1.0 cards
>>> seem to be rare, and have to be ordered online, while I can pick up a
>>> 2.0 card locally at a store.
>>>
>>
>> PCIe-2.0 is fully downward compatible to 1.1 and 1.0.
>>
>>> 2) My main "torture test" will be HD fullscreen video.  Will there be
>>> major improvement in that?  That's 2D.  Forget 3D.
>>>
>>
>> 2D video is still rendered using OpenGL if your video player supports it.
> 
> I'm not aware of any video decoders using CUDA, OpenCL, or pixel
> shaders for video decoding; AFAIK, unless you're using VDPAU you're
> still using the CPU to render the video to a frame buffer. The most a
> video player is going to use OpenGL for is stretching that frame
> buffer to fit a window or screen, and possibly as a compositor to
> place overlays like subtitles or playback control elements..
> 

Agreed. Decoding is still usually done in software but offloading
scaling and YUV to RGB conversion helps none the less. Mplayer, for
example, allows a lot of customization depending on the amount of
texture units. With high resolution displays and slow CPUs, this can
have surprisingly large effects.

Regards,
Florian Philipp


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Volker Armin Hemmann]

Am Sonntag, 15. April 2012, 16:44:43 schrieb Florian Philipp:
> Am 15.04.2012 16:22, schrieb Michael Mol:
> > On Sun, Apr 15, 2012 at 10:09 AM, Florian Philipp <lists@binarywings.net> 
wrote:
> >> Am 15.04.2012 15:18, schrieb Walter Dnes:
> >>> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
> >>> 
> >>>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
> >>>>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
> >>>>> 
> >>>>> money is helpful <G>.  Would PCIe be significantly better on the same
> >>>>> CPU+GPU, or is it hype?
> >>>> 
> >>>> a lot, lot lot lot better. No hype.
> >>>> 
> >>>   I've done some looking, and I'm back with more questions.  I've also
> >>> 
> >>> read the Nouveau-versus-NVIDIA thread.  Questions...
> >>> 
> >>> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
> >>> performance, I just want full backwards compatability.  PCIe 1.0 cards
> >>> seem to be rare, and have to be ordered online, while I can pick up a
> >>> 2.0 card locally at a store.
> >> 
> >> PCIe-2.0 is fully downward compatible to 1.1 and 1.0.
> >> 
> >>> 2) My main "torture test" will be HD fullscreen video.  Will there be
> >>> major improvement in that?  That's 2D.  Forget 3D.
> >> 
> >> 2D video is still rendered using OpenGL if your video player supports it.
> > 
> > I'm not aware of any video decoders using CUDA, OpenCL, or pixel
> > shaders for video decoding; AFAIK, unless you're using VDPAU you're
> > still using the CPU to render the video to a frame buffer. The most a
> > video player is going to use OpenGL for is stretching that frame
> > buffer to fit a window or screen, and possibly as a compositor to
> > place overlays like subtitles or playback control elements..
> 
> Agreed. Decoding is still usually done in software but offloading
> scaling and YUV to RGB conversion helps none the less. Mplayer, for
> example, allows a lot of customization depending on the amount of
> texture units. With high resolution displays and slow CPUs, this can
> have surprisingly large effects.
> 

and with vlc you can use vaapi which can make use of the video decoding engine 
of the graphic chip.

If the movie is using the right codec, of course.
-- 
#163933

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Michael Mol]

On Sun, Apr 15, 2012 at 4:45 PM, Volker Armin Hemmann
<volkerarmin@googlemail.com> wrote:
> Am Sonntag, 15. April 2012, 16:44:43 schrieb Florian Philipp:
>> Am 15.04.2012 16:22, schrieb Michael Mol:
>> > On Sun, Apr 15, 2012 at 10:09 AM, Florian Philipp <lists@binarywings.net>
> wrote:
>> >> Am 15.04.2012 15:18, schrieb Walter Dnes:
>> >>> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
>> >>>
>> >>>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
>> >>>>>   If it's PCIe, so be it.  Actually, a post that prevents me wasting
>> >>>>>
>> >>>>> money is helpful <G>.  Would PCIe be significantly better on the same
>> >>>>> CPU+GPU, or is it hype?
>> >>>>
>> >>>> a lot, lot lot lot better. No hype.
>> >>>>
>> >>>   I've done some looking, and I'm back with more questions.  I've also
>> >>>
>> >>> read the Nouveau-versus-NVIDIA thread.  Questions...
>> >>>
>> >>> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting 2.0
>> >>> performance, I just want full backwards compatability.  PCIe 1.0 cards
>> >>> seem to be rare, and have to be ordered online, while I can pick up a
>> >>> 2.0 card locally at a store.
>> >>
>> >> PCIe-2.0 is fully downward compatible to 1.1 and 1.0.
>> >>
>> >>> 2) My main "torture test" will be HD fullscreen video.  Will there be
>> >>> major improvement in that?  That's 2D.  Forget 3D.
>> >>
>> >> 2D video is still rendered using OpenGL if your video player supports it.
>> >
>> > I'm not aware of any video decoders using CUDA, OpenCL, or pixel
>> > shaders for video decoding; AFAIK, unless you're using VDPAU you're
>> > still using the CPU to render the video to a frame buffer. The most a
>> > video player is going to use OpenGL for is stretching that frame
>> > buffer to fit a window or screen, and possibly as a compositor to
>> > place overlays like subtitles or playback control elements..
>>
>> Agreed. Decoding is still usually done in software but offloading
>> scaling and YUV to RGB conversion helps none the less. Mplayer, for
>> example, allows a lot of customization depending on the amount of
>> texture units. With high resolution displays and slow CPUs, this can
>> have surprisingly large effects.
>>
>
> and with vlc you can use vaapi which can make use of the video decoding engine
> of the graphic chip.
>
> If the movie is using the right codec, of course.

Also depends on whether or not the graphics driver and vaapi like each
other. I'm not aware of NVidia cards supporting VA API yet.

VA API is pretty new; it'll be interesting to see where it goes, and I
hope it takes hold. Right now, the most tested and working solutions,
AFAIK, are nVidia cards and VDPAU. At least, that combination has been
working well for me for 3-4 years.

-- 
:wq

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Volker Armin Hemmann]

Am Sonntag, 15. April 2012, 16:54:58 schrieb Michael Mol:
> On Sun, Apr 15, 2012 at 4:45 PM, Volker Armin Hemmann
> 
> <volkerarmin@googlemail.com> wrote:
> > Am Sonntag, 15. April 2012, 16:44:43 schrieb Florian Philipp:
> >> Am 15.04.2012 16:22, schrieb Michael Mol:
> >> > On Sun, Apr 15, 2012 at 10:09 AM, Florian Philipp
> >> > <lists@binarywings.net>
> > 
> > wrote:
> >> >> Am 15.04.2012 15:18, schrieb Walter Dnes:
> >> >>> On Wed, Apr 11, 2012 at 06:30:02PM +0200, Volker Armin Hemmann wrote
> >> >>> 
> >> >>>> Am Mittwoch, 11. April 2012, 02:11:35 schrieb Walter Dnes:
> >> >>>>>   If it's PCIe, so be it.  Actually, a post that prevents me
> >> >>>>> wasting
> >> >>>>> 
> >> >>>>> money is helpful <G>.  Would PCIe be significantly better on the
> >> >>>>> same
> >> >>>>> CPU+GPU, or is it hype?
> >> >>>> 
> >> >>>> a lot, lot lot lot better. No hype.
> >> >>> 
> >> >>>   I've done some looking, and I'm back with more questions.  I've
> >> >>> also
> >> >>> 
> >> >>> read the Nouveau-versus-NVIDIA thread.  Questions...
> >> >>> 
> >> >>> 1) Will PCIe 2.0 cards work in a PCIe 1.0 slot?  I'm not expecting
> >> >>> 2.0
> >> >>> performance, I just want full backwards compatability.  PCIe 1.0
> >> >>> cards
> >> >>> seem to be rare, and have to be ordered online, while I can pick up a
> >> >>> 2.0 card locally at a store.
> >> >> 
> >> >> PCIe-2.0 is fully downward compatible to 1.1 and 1.0.
> >> >> 
> >> >>> 2) My main "torture test" will be HD fullscreen video.  Will there be
> >> >>> major improvement in that?  That's 2D.  Forget 3D.
> >> >> 
> >> >> 2D video is still rendered using OpenGL if your video player supports
> >> >> it.
> >> > 
> >> > I'm not aware of any video decoders using CUDA, OpenCL, or pixel
> >> > shaders for video decoding; AFAIK, unless you're using VDPAU you're
> >> > still using the CPU to render the video to a frame buffer. The most a
> >> > video player is going to use OpenGL for is stretching that frame
> >> > buffer to fit a window or screen, and possibly as a compositor to
> >> > place overlays like subtitles or playback control elements..
> >> 
> >> Agreed. Decoding is still usually done in software but offloading
> >> scaling and YUV to RGB conversion helps none the less. Mplayer, for
> >> example, allows a lot of customization depending on the amount of
> >> texture units. With high resolution displays and slow CPUs, this can
> >> have surprisingly large effects.
> > 
> > and with vlc you can use vaapi which can make use of the video decoding
> > engine of the graphic chip.
> > 
> > If the movie is using the right codec, of course.
> 
> Also depends on whether or not the graphics driver and vaapi like each
> other. I'm not aware of NVidia cards supporting VA API yet.
> 
> VA API is pretty new; it'll be interesting to see where it goes, and I
> hope it takes hold. Right now, the most tested and working solutions,
> AFAIK, are nVidia cards and VDPAU. At least, that combination has been
> working well for me for 3-4 years.

va-api can use vdpau as backend ;)

I am not using nvidia anymore and some time ago vaapi-xvba started to work for 
mp4.

-- 
#163933

Re: [gentoo-user] PCI video cards, hardware accel, upported by open-source drivers?

[Claudio Roberto França Pereira]

Since we're talking about video playback with open source drivers, any
luck to intel onboard adapter users? I'm struggling with an HD3000
core i3 video adapter, can't playback 1080p without tearing. I was
impressed with this same adapter running a racing game on a colleagues
macbook air, smooth and aliased.

[gentoo-user] Re: PCI video cards, hardware accel, upported by open-source drivers?

[Nikos Chantziaras]

On 17/04/12 17:11, Claudio Roberto França Pereira wrote:
> Since we're talking about video playback with open source drivers, any
> luck to intel onboard adapter users? I'm struggling with an HD3000
> core i3 video adapter, can't playback 1080p without tearing.

Tearing shouldn't occur when you use Xv for playback.  Are you using GL 
instead?  That might explain it.