Firstly, props to nVIDIA for planning to bundle H.264 decoding software that utilises it's hardware for no additional charge. Hopefully they will eventually extend this policy to the rest of their PureVideo suite as well, which after the 30-day free trial requires an additional
charge, i.e. either Bronze ($19.95), Gold ($29.95) or Platinum ($49.95) to unlock the varying functionality.
Traditionally, ATI have not charged for similar functionality, but with the X1K series have outsourced their H.264 decoder to 3rd party developers such as Cyberlink, who
charge $14.95 to download it. Again, one hopes that nVIDIA's free H.264 bundling will encourage ATI to do the same and in turn encourage nVIDIA to bundle the whole PureVideo suite free of charge.
D.P. said:
Hopefully that will also signify the release of the H.264 supporting drivers I have been talking about which will allow everyone with a 6200 and up to play full 1080P H.264 compressed movies.
I will be very impressed if a 6200 receives significant benefit (i.e. much less CPU usage when decoding with the same CPU) from the H.264 "PureVideo driver" and can decode a full, high bit-rate (>9mbps) 1080p H.264 stream without dropped frames. However, as oftens happens with similar proclaimations from both IHV's, there is usually a caveat or two and I would wait for independent reviews before accepting the marketing spiel.
It seems peculiar that a feature (hardware HD H.264 decoding) which has been promised for quite a few months now as a simple driver release, should require such an extended (software) development period when PureVideo (while
programmable - c.f. Question 2) is largely a dedicated logic block on the ASIC and the base required hardware (GeForce 6200) has been available on the market for over a year. This begs the question of why the long wait and the cynic would say it is simply a stall tactic because your main competitor has a lead to market on a feature which you can't implement yet.
I have heard mention of PureVideo2, although I can't recall the source. While it is likely an updated software implementation mostly for marketing purposes, it is possible that additional/improved logic is included in nVIDIA's new 90nm products, i.e. 7300GS/7600GT/7900GT/GTX/etc that possibly includes superior H.264 performance among other things, although this would obviously preclude the 6200 et al. More likely it is just an updated frontend that exposes some more functionality, which could explain for the delay. We'll know in a few more weeks.
D.P. said:
...ATI cards use the pixel shaders to do the decoding, this just isn't as fast as dedicated hardware like Nvidia cards have...
While most of the AVIVO documentation, reviews and subsequent discussion suggest that AVIVO (X1K range) relies on processed shader code to a significant extent in the decoding process, it does not rely on it exclusively and does have some
dedicated logic (c.f. Question 2).
As Dave Baumann mentions in his ATI AVIVO
Winter Update:
Beyond3D AVIVO Winter Update said:
...ATI's H.264 decoding solution relies on both custom logic as well as shader code being processed over the pixel shader pipelines so the performance can be slightly reliant on the graphics capabilities of the board in question - the fewer the pipelines the lower the decoding power, hence the lower the resolution, in theory. In practice it's not quite as clear cut at that as not only are things going to be dependant on the resolution, but also the bit rate at which the video is encoded - X1300, for instance, may be fine at decoding full HD resolution sources encoded at lower bit-rates. Also, given the performances here it's evident that the decoding isn't highly optimised yet and we suspect that there is more performance to come for the entire line...
In many instances his testing shows similar performance metrics across the X1K range in H.264 decoding and a minimal delta between the software (CPU-only) implementation which would suggest that further gains should materialise primarily from the programmable (shader) aspect of the AVIVO engine as more efficient code is developed and optimised. In support of this, Dave goes onto suggest that the higher end boards are currently only implementing the video decoding over a single quad, which could potentially open up even greater decode performance if the other quads are tapped.
Some further highlights on the areas covered by the custom logic
here:
ATI AVIVO Video & Display Engine - Technology Discussion said:
Reverse entropy means rebuilding the larger dataset created in the first encode stage outlined on the previous page [...] CABAC works by analysing frame data to decide on the best compression scheme. Per-frame, that's a nice added cost in decode to get it done faster than real-time so you don't drop frames. R5-series GPUs from ATI have dedicated silicon for that.
[...]
The biggest thing to take from the Decode stage is that Avivo-capable products have dedicated gates for video decode. Computationally expensive formats like H.264 get significant assist by the GPU according to the Avivo literature, both by fixed function hardware, programmable video-only silicon and the 3D shader hardware if needed.
Anyway, the next few weeks should be very interesting -- nVIDIA seems to have a busy Cebit planned! Here's hoping they deliver on the expectations.
Cheers,
BrynS
