Decoding performance?

[zeflash]

Hi

 

I just bought this cool little app because I found all the OSD stuff so well done, and it's cool and all, except one thing: it's quite slow compared to other DVB proggies.

 

In DVBViewer, when looking a euronews for example, I'm getting like 75% CPU usage - compared to 20% for the same channel in progDVB.

 

Any reason for that? any way to have it working faster? I might try DVBViewer GE but if it doesn't have the OSD .... it's getting far less interesting.

[Guest Oliver]

Maybe you should use a video-decoder that supports the hardware-decoding features (IDCT) of your graphic card. Believe me: ProgDVB is definitely not faster.

[Guest Lars_MQ]

Any reason for that? any way to have it working faster? I might try DVBViewer GE but if it doesn't have the OSD .... it's getting far less interesting

The cpu usage is caused by the video(mpeg)decoder. So playing a little bit around with the settings in options -> directX will show a much better performance. Your milage may vary but I got (athlon XP 2000+) round about 10 -15 % cpu usage.

[zeflash]

Maybe you should use a video-decoder that supports the hardware-decoding features (IDCT) of your graphic card. Believe me: ProgDVB is definitely not faster.

<{POST_SNAPBACK}>

 

Humm... that's weird. I'm pretty sure I'm using the same decoders in both cases (the "old" free nvidia one).

Are the graph file used by DVBViewer more complex than the ones in progdvb? Given we can only choose the decoders, I assume things are done automatically in the background?

 

By the way, I'm currently using an "old" video card (gf4 ti4200) - maybe it doesn't handle the hardware decoding well. But still, it would also be slow in progdvb.

 

In any case, I'll keep trying.

Edited August 4, 2005 by zeflash

[zeflash]

I've figured out the problem: I was using the dvbserver.

I wanted to be able to remotely watch something over the network, so I set up the dvbserver as a multicast server, and set up the DVBViewer on the same computer to use the server instead of using the card directly.

 

Somehow, in this case it takes a LOT more of CPU than when using the SS2 card directly. I wonder why?