Avivo vs Purevideo article

[gEd]

Do you guys know if AVIVO is available just as a software CODEC like PureVideo? What I've seen about it on ATI's site is that it's a hardware platform, and I have an aging Radeon 9200 card. I've been using PureVideo and have not been too satisifed, so I'd love a chance to try an ATI product instead.

I bought a 9200 for pvr duties but got rid of it immediately as the svideo out picture was very soft and fuzzy.

I am using Purevideo with my ATI 9600 card and I am generally very pleased with the results.

The standard AVIVO encoder/decoder only works with the current X1xxx series of cards. It includes an encoder app and a mux filter but I haven't tried to get the decoder working with gbpvr (assuming it exists).

A "modded" version of AVIVO that works with any card can still be downloaded from the link refered to in my original thread on AVIVO.

http://forums.nextpvr.com/showthread.php...ight=avivo

[Torque]

I had been using the Intervideo codec with my Geforce 6600 and was happy with the results. Recently switched to the Purevideo codec and although the picture looks better, it uses about 10% more CPU with HW accelleration enabled (and 20% more without) compared to Intervideo. That seems to be just enough to make the video playback from my aging system stutter a bit for 5-10 seconds after I start to watch a recording or live TV.

[wtg]

I installed an AGP 6200 about a year ago and was generally disappointed with the well documented bordering garbage that appeared on screen using the card with VMR9. I kept hoping a driver update would fix it, but it never came.

I'm driving a HD TV at 1920x1080 and perhaps I'm just expecting more from the cards than I should, but the CPU usage has been frustratingly high. Next, I purchased the PureVideo decoder after all the rave reviews on how much better the image quality was and NVidia claim of it's great use of the special PureVideo hardware, but have been very disappointed. The image quality is good but the CPU usage was worse unless I used it in Overlay mode. It would only provide Mode A directx hardware acceleration - the lowest type of DxHA I understand. Nero decoders w/o hardware acceleration in VMR9 used less CPU than NVidia decoders with hardware acceleration. I documented my results in this thread and thanks to some good advice, I did find the Overlay performance could be considerably improved by downgrading to the 84.21 drivers. VMR still was lousy and since the screen corruption continued to occur, it didn't matter anyway.

HD playback was spotty at best. Impossible to get smooth playback in any VMR modes and even Overlay was touchy.

So recently I purchased a ATI X1600 Pro. I had hoped the performance would be better using the ATI drivers because I had such good performance with a wee 9200 in the past using the special ATI-optimized Cyberlink drivers. Nope... no such luck. The card performs ok but it's supplied ATI MPG Decoder doesn't offer much for performance either. I mean, it's ok but I have the ATI 9200 with ATI Cyberlink drivers in a P4 1.8 GHz PC and get 16-20% CPU utilization in Overlay mode (running 1280x1024 though). I get higher utilization with my AMD 2500 and the X1600 and it's decoder - 40-50% or so if memory serves. I even installed the Cyberlink drivers from my 9200 and tried using them with the card but no improvement.

Finally, I tried the PureVideo drivers with my ATI card... bingo. At least as far as Overlay goes this is the best performance so far, <10% utilization playing SD content, and very nice stable playback of HD 1080i content at 30% CPU. I get mode C DirectX hardware acceleration. VMR9 mode CPU usage is still way too high in my opinion, at around 55-60% for SD content but the Overlay numbers are so low I'm willing to live with it. It's strange to think that I get the best performance out of a ATI card using the NVidia decoder, and makes me think that neither ATI or NVidia have really tried very hard to deliver a good video decoder card. I think the lion's share of their efforts are spent on gaming and video is an afterthought.

My next step is to upgrade my sound card and see if that offers any improvement. The PC is using integrated sound and I wonder if there's something going on there in DirectX that causes the dramatic slowdown in VMR modes. Maybe not. The hardest part of all of this is determining what the real bottlenecks are, and all the potential variables. And unfortunately there are few reviews or articles that really dive into what it takes to measure performance and quality on HTPCs. The article that started this thread was one of the better ones that I've read, and it only scratched the surface.