How many GFLOPs are actually in use to decode hd level VC-1/mpeg-4?

Bohdy said:
I've always assumed that the PSP had dedicated AVC decoding logic to be able to decode UMD's.
Hmm, the PSP has all sorts of processing units, see here. I guess what you're referring to is the mysterious VME. Anyway, it's clearly not correct to say that the PSP decodes 480P H.264 on a paltry 222mhz R4000 ;)
 
deathkiller said:
Not in my PC, I hate that game trailers come in 720p60 wmv(VC-1), I don't have problems with 720p24-30 in any format.


Where can I find samples of 720p60 and 720p24-30? I want to test my pentium-m.
 
Mmmkay said:
Yes I'm quite sure of that. People have ripped the contents of UMD movies to find that the .mps files have a resolution of 720x480.
I didn't know that. Do you know if the content is interlaced since that makes a significant difference to the decode logic.

Also do we know if they use CABAC?

What I am getting at here is that there is no need for Sony to stick to the defined profiles of H.264 for this closed system so they can cherry pick features to support (such as B frames) whilst ignoring other features which are harder to implement (such as CABAC).
 
Compression can be an interesting benchmark, but it tends to degenerate to your ability to read and fill memory, although Mpeg2 has enough work to do to stop this being the only factor.

I've run tests on LZH type dictionary compressors on both next gen platforms and you'd be surprised how poorly they compare to current gen platforms. They are obviously faster, but not by a large margin. In real terms taking into account both tranfer rate and latency the memory on PS3/X360 is only about 2x the speed of Xbox , and simple decompression algorithms tend to bear this out.
 
Also cruise the Microsoft website for their WMV-HD collection.

If you thought you were getting by just fine with your older, sub-2 Ghz computer to get your basic tasks done, running these videos will humble you pretty quick. They aren't even the most intense examples of hd content, either. ;)
 
ERP said:
In real terms taking into account both tranfer rate and latency the memory on PS3/X360 is only about 2x the speed of Xbox , and simple decompression algorithms tend to bear this out.
It sounds kinda expected to me - but I gather you were working off a shared codebase.
Given the streaming nature of LZW type coders and their usual 'window' size, I imagine that something that did most work in local store (or locked cache) should be relatively straightforward to implement, and probably yield significant improvements on either polatform.
 
randycat99 said:
Also cruise the Microsoft website for their WMV-HD collection.

If you thought you were getting by just fine with your older, sub-2 Ghz computer to get your basic tasks done, running these videos will humble you pretty quick. They aren't even the most intense examples of hd content, either. ;)

I can play fine all hd (720p, my laptop screen is only 1280x800) videos with my proccessor at 1.3 ghz (pentium-m), including 720p quicktime h.264.
 
Fafalada said:
It sounds kinda expected to me - but I gather you were working off a shared codebase.
Given the streaming nature of LZW type coders and their usual 'window' size, I imagine that something that did most work in local store (or locked cache) should be relatively straightforward to implement, and probably yield significant improvements on either polatform.

It was pretty much what I expected, I just wanted to demonstrate another side to these new processors performance, we have a 30x+ improvement in floating point power, but only a 2x performance improvement in memory performance.
 
ERP said:
It was pretty much what I expected, I just wanted to demonstrate another side to these new processors performance, we have a 30x+ improvement in floating point power, but only a 2x performance improvement in memory performance.

Well memory latency has always been a problem with faster and faster processors anyway. If anything the faster execution exacerbates the problem (vector units magnify the problem even moreso with the amount of data they can swallow) as a complex problem on one processor becomes a simple blitter algorithm on a faster one...
 
Another thing I wanted to ask is more specific to the Toshiba hd-dvd player, itself. I can understand the P4 CPU to do the heavy lifting for VC-1 decoding. However, what is the 1 GB of RAM about? How does one function (playback) need all that memory? Is it there to cover the most insane interactive features? It just seems way out there to me. What kind of stuff is needing all this space?

Off the top of my head, I'd think maybe 128 MB could do the job, but maybe there is more going on here than is obvious? I never thought about how much RAM is used in a typical DVD player, but is that considerable, as well? Anybody know?
 
Well, that's very bizarre, indeed. What in the world justifies 2.5 Ghz P4 and 1 GB of RAM for a videoplayer then? Is it all there just to run iHD?
 
Probably. iHD is JavaScript + XML + CSS + other stuff. When's the last time a web browser running DHTML fit in a small amount of RAM and ran 60Hz interactivity via JavaScript? :)

Another dumb Microsoft design. If Disney had gone with Macromedia Flash, they'd have everything iHD could do, in a small, compact, superfast executable that has excellent development tools.
 
Flash makes sense. I've just got into Flash and the tools are pretty easy and you get fantastic results good and quick. It's so well established, why bother inventing a new standard? Redesigning the wheel over and over...
 
Advocating for the devil....
Shifty Geezer said:
Flash makes sense. I've just got into Flash and the tools are pretty easy and you get fantastic results good and quick. It's so well established, why bother inventing a new standard? Redesigning the wheel over and over...
...because it's a closed/proprietry?
 
Back
Top