VLC Media Player To Receive VDPAU Support

[deanjo]

Which cards support VDPAU? (G7x, G8x, or?)

http://www.nvnews.net/vbulletin/showthread.php?t=123091

Cards that support VC-1 list is incomplete as well. G98 cards (newer 8400GS and 9400 cards) should be able to do it and 8200/8300 igps are proven to work as well.

[Kano]

I would say G9x, complete (incl. VC1) I guess only G98 right now.

[myxal]

Yet another player supporting VDPAU? Pfft.
At this rate, the real question is: When will VDPAU support come to xorg-video-ati or xorg-video-radeonhd? (I'm not counting on Xorg-video-intel)

Seriously, XvBA is long-time no-see, zero player supporting, likely-no-better than DxVA bullet-point on AMD's roadmap, XvMC is insufficient and VA-API is vaporware.

I'm really close to being in the market for a new graphics/mobo. If it weren't for nvidia's horrible track record with chipsets and recent problems with GPUs dying, I'd go for GeForce in a heartbeat. Since that just isn't the case, the current "purchase candidate" is a AMD 780G-based board, using 'that other OS' for now and hope for usable, supported solution to come along.

[deanjo]

I would say G9x, complete (incl. VC1) I guess only G98 right now.

Well you can't say G9x as the G92 does not support VC-1 decode.

[Kano]

I did not say that, i only mentioned G98 for VC-1. Anything wrong with it?

[bugmenot]

Now they only need to start supporting VA API, too.

[deanjo]

I did not say that, i only mentioned G98 for VC-1. Anything wrong with it?

Nope sorry, just read your reply the wrong way.

[makomk]

Isn't it be possible to dump the symbols and deduce the header from the mangled names? I think Michael said that Amd's API is a simple extension to the existing video decode acceleration (sorry, I can't remember yet another acronym ) - I'm not saying it's worth the effort, but it could make for a fun afternoon.

Nope. Firstly, the API is C, so no mangled symbols. Most of the functions used to be wrappers of some sort around C++ code, but it looks like this is no longer true. Secondly, this just gives you the type names of the arguments - there's still no way of figuring out the layout of structs passed to the functions (e.g. XVBA_BUFFER_DESCRIPTOR), the argument names, or the return types. Any magic constants are right out.

Also, with 8.12 the API seems to have changed to become more freestanding and less of an extension to XvMC. Plus, at the same time, the C functions stopped being simple wrappers around C++ function calls, so no guessing the API that way.