VDPAU testing tool

[jyavenard]

Originally Posted by Quantumstate

Here is 100% of the HD content I watch:

Your point being ?

your card can decode all those content just fine... with ample margin too

[jyavenard]

Originally Posted by Quantumstate

I notice in my tests above that 185.18.10 decodes faster than 185.18.14, but it deinterlaces slower. I'd rather have faster decoding. Is powermizer fixed in .10?

Who cares when in both cases as your decoding is fast enough for any media you're likely to even encounter?

When are you ever going to find mpeg2 1920x1080 at a faster framerate than 60fps ?

[Quantumstate]

My point being, that I am trying to confirm my understanding with others, OK?

[Quantumstate]

Originally Posted by jyavenard

Who cares when in both cases as your decoding is fast enough for any media you're likely to even encounter?
When are you ever going to find mpeg2 1920x1080 at a faster framerate than 60fps ?

I'm not, but when both my cards do a max of 60 pic/s, isn't that pretty freakin' marginal?

Why wouldn't I want to back out to an older driver that's faster?

And furthermore, it is not clear at all whether this "1080" is 1080i or 1080p. Big difference.

[Erik_]

Here are the results for a Q9550 (@2.83Ghz) with a Quadro FX 3700M (1GB, 550Mhz/799Mhz). OS is Fedora 10 (x86_64).

Drivers 180.60

SURFACE GET BITS: 1008.18 M/s
SURFACE PUT BITS: 1088.05 M/s

MPEG DECODING (1920x1080): 78 pic/s
H264 DECODING (1920x1080): 45 pic/s
VC1 DECODING (1440x1080): 121 pic/s

MIXER WEAVE (1920x1080): 1998 frames/s
MIXER BOB (1920x1080): 3673 fields/s
MIXER TEMPORAL (1920x1080): 792 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 301 fields/s

Drivers 185.18.14

SURFACE GET BITS: 1005.87 M/s
SURFACE PUT BITS: 1082.69 M/s

MPEG DECODING (1920x1080): 78 pic/s
H264 DECODING (1920x1080): 45 pic/s
VC1 DECODING (1440x1080): 121 pic/s

MIXER WEAVE (1920x1080): 2003 frames/s
MIXER BOB (1920x1080): 3673 fields/s
MIXER TEMPORAL (1920x1080): 792 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 325 fields/s

[jyavenard]

Any FTA media is a maximum of 1920x1080 @ 60Hz interlaced, so 30 fps.

For any H264 content, it all depends on the complexity of the profile and same thing FTA is 30 fps max ; with a bluray being 24fps.
I don't know what h264 video he's using on his test, but it's very unlikely performance wouldn't vary based on the video being played.
Like a capture from a HD-PVR is extremely complex to decode and require a huge processor, yet VDPAU does it perfectly on even a 8400gs which is one of the slowest card available.

From what I've seen, decoding is never an issue in real life. All VDPAU cards seem to be able to handle any content. Deinterlacing is what make a big difference between cards.

[Quantumstate]

I know you're in France jyavenard, but it seems to me that US ATSC is 1080i 59.940frames/ps, or deinterlaced is 29.970 pictures/s.

So the question I am trying to ask is, which is relevant to the test results?

[jyavenard]

Originally Posted by Quantumstate

I know you're in France jyavenard, but it seems to me that US ATSC is 1080i 59.940frames/ps, or deinterlaced is 29.970 pictures/s.

So the question I am trying to ask non-hostile ppl is, which is relevant to the test results?

For a start I'm not in France, only happen to have a French first name..

US ATSC is 29.976 fps or 59.94Hz interlaced.

It's the framerate per second that is relevant... Because it's interlaced, a full frame is made of 2 interlaced ones.
60Hz interlaced = 30 fps.
50Hz interlaced = 25 fps
60Hz progressive = 60 fps, but this is never broadcast, the best that can be broadcast is 60Hz interlaced or 30 fps.

I'm not hostile, I'm answering your questions, but you stubbornly choose to ignore them continuing to ask even after you got it answered.
In my first answer to you, previous page I said the exact same thing.
Your card is fast enough for any media you will ever encounter...

[crisalide]

Originally Posted by jyavenard

I don't know what h264 video he's using on his test, but it's very unlikely performance wouldn't vary based on the video being played.

Of course, results would vary with other streams, depending on profile/level and bitrate. (For example, the 25 first frames from Grey.ts are decoded at 68pic/s)
So, i've tried to take samples from average real-life streams.

[jyavenard]

Originally Posted by crisalide

Of course, results would vary with other streams, depending on profile/level and bitrate. (For example, the 25 first frames from Grey.ts are decoded at 68pic/s)
So, i've tried to take samples from average real-life streams.

Actually, a test that would be very relevant: use a H264 file generated by a HD-PVR.
They are known to be very hard to decode and even a 3GHz C2D is known to struggle on some.

[crisalide]

Originally Posted by jyavenard

Actually, a test that would be very relevant: use a H264 file generated by a HD-PVR.
They are known to be very hard to decode and even a 3GHz C2D is known to struggle on some.

Could provide a sample?

[cthulhu]

Results for my laptop, Asus U6V w/ 9300M GS

SURFACE GET BITS: 845.825 M/s
SURFACE PUT BITS: 576.458 M/s

MPEG DECODING (1920x1080): 69 pic/s
H264 DECODING (1920x1080): 65 pic/s
VC1 DECODING (1440x1080): 84 pic/s

MIXER WEAVE (1920x1080): 246 frames/s
MIXER BOB (1920x1080): 396 fields/s
MIXER TEMPORAL (1920x1080): 74 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 22 fields/s