MPEG-4 Hardware Decoder For $99
secondsun writes: "Tom's Hardware has the story. Apparently sigma designs has made a PCI card that decodes DiVX movies in reltime with little processor overhead." Under a hundred bucks, too.
← Back to Stories (view on slashdot.org)
DivX codec changes so frequently, what are you gona do, flash your card every month?
But how long until there are Linux drivers?
LedgerSMB: Open source Accounting/ERP
Coming from sigma designs, you can expect them to get around to releasing the drivers for it in 2008.
Free software implementations of the MPEG standard (2,4) legally cannot be done because the MPEG standard is full of patents, usually requiring payment of licensing fees.. If hardware vendors implement MPEG on hardware, and open the specifications for it's hardware, it is possible to have 100% legal playback of these media on 100% free software systems.
MPEG-4 for older PCs
. To ensure that just about any Divx film can be played back without image dropouts, you should use at least an AMD Athlon with 800 MHz plus or an Intel Pentium III/733.
um right. i have a p2/500 and i can run all divx moviews flawlesly, so the "divx for old pcs" is kind of moot, isnt it?
It looks like a cool gadget none the less, but personally im more concerned by the direction the Divx project is going than what cool stuff i can do with it.
This sig was cut off by the sla
Not trolling. Just pointing out that not all that glitters is worth $99.
I mean sure I guess now you could maybe decode divx on a 486 or something, but otherwise I can't see what the point is. It's not like with graphics cards where the cpu cycles saved are desperately needed elsewhere within the same process (the game). What, if anything, does anyone think this card's selling points would be? Maybe if someone integrated this technology into pre-existing graphics or dvd-decoder cards. . .
This card has been out for a while. Nothing new... Only thing that is new is that the DivX codec keep changing every 3 months...
I wouldn't mind being flashed every month.
Actually, I wouldn't even mind being flashed every day.
I preordered one of these.
I use my software tv out for playing divx still. The drivers were shitty. (win2k fresh install) The only thing that they would have given me is the ability to fast forward and rewind with the remote control on my celeron 900.
It comes with it's own horribly skinned app, which crashes consistently on my computer.
It's still good for watching dvds. You could , in theory, throw this card into a machine that has no sound or tv out, and watch movies on your tv with it, since it provides high quality outs for both.
Note that the tv out only works for watching movies-- it's not an addon tv out card that'll let you use emulators and the like on your tv. For one of those, either get a scanline converter or tv out on your video card. I use a tnt2.
what the hell are you gonna use the cpu cycles for?
CPUs are getting faster and faster, and since processor intensive tasks are getting exported to cards what the hell do you need your cpu for.
If you are using that card that means that you are watching a movie so you are probably not doing anything else processor intensive.
I guess there are some porfessionals that need to do processor intensive tasks in the background but thats not true for most people.
I guess microsoft needs to work extra hard to ensure newer versions of windows soke up even moreprocessor power.
This product is (right now) of little use, as linux support is not out there yet... But I am a little amused by the "need a 733+" idea of DiVX decoding! Xine (http://xine.sourceforge.net/ ) lets me decode fullscreen divx on a measley PII266, and with the addition of a Creative DXR3 (20-30 on Ebay), you can display anything that xine will decode onto a TV (s-video) through the card. New codec, no problem. The card simply outputs the xine-decoded information. Beautifully, I might add. And the card includes an S/PDIF coax output that functions nicely as /dev/dsp!
Just a thought, as this is a linux-friendly solution, and is completely codec independent.
(caveat --> Using the DXR3 actually re-encodes the video stream into MPEG which the card can decode in hardware. Doing so is fast (using FAME), but takes a little more oompf that a pII266. 350-400 MHZ is more then enough, however, to decode DiVX, reencode at 100% MPEG, and still act as a fileserver : )
Check out my post here, it might give you an idea or two why this hardware is interesting. :)
"Derp de derp."
The em847x is a very nice chip and cries out for being used in STBs. It is the successor of the em8400 which was used in the netstream PCI cards and in some STBs and which provides a great MPG1/2 decoder with a very high quality TV picture. The great thing is that some of the em847x chips are pin compatible with the em8400, so that manufactures don't have to change their layout.
The only shortcomings are that it only provides overlay for the display on your PC, i.e. no DMA into the graphics memory like most TV capture cards. That's of course because of the paranoia of the DVD consortium.
There will probably also be Linux drivers, in the same fashion as for the em8400 (closed user space and with a pass through kernel module) which is unfortunate and ill designed. That means no video4linux or Linux DVB API support (although you can probably get the latter also closed source).
***Quis custodiet ipsos custodes***
Reasons why it might be useful to have a hardware MPEG-4 decoder:
1) Latency, latency, latency... You wouldn't want to miss 1 second of Baywatch just because you are compiling now, would you?
2) Embeddable solution... Look ma, no X! Just slap one of these puppies in and you can run your fav OS with high-quality TV out... assuming someone doesn't try to prevent drivers from being written for it...
3) Encoding possibility... Heck, if hardware decoders exist, hardware encoders can be built too! I just hope they wouldn't be too expensive.
4) Hiiiiiigh load... I can just see some bragging in the future: "Hey, I can play a DVD and write a CD and rip a CD and record a TV program all at once!"
5) The future... People, realise this, in a couple of years your PC architecture is going to be a CPU that delegates tasks to the dedicated sub-CPUs. Look at the 3D card industry if you want an example.
Things that might not be cool:
1) I don't need one! Nobody's going to buy this one because they can already play the stuff!.
2) Too expensive! $99 is a sizable chunk of salary where I come from. (Don't ask). I'd rather save that to invest in my next PC.
3) Not enough features! Bundled with VGA output for dual-screen, this would have been very, very useful. Bundled with TV capture, this would have been a shoe-in. Bundled with an encoder it would have been... glorious!
I'm not going to buy one. If they bring out an encoder, I will buy one!
I would rather have a DIVX hardware ENcoder. Something that allows you to rip^H^H^H make safety copies of you DVD collection in less time.
IANAL, but imagine a beowulf cluster of in Soviet Russia all your belong are base to us welcoming the new SCO overlords.
I believe that as consumers find video and audio to be an increasingly important part of their computing experience, increasingly more operations will emmigrate from the main processors, onto auxiliary processors designed and programmed for specific purposes.
The good ol' SGI machines, for example, could have up to 12 "graphics-only" processors, if I remember correctly. Nowadays, as home computers are approaching (and perhaps even surpassing) the raw speed possible with the good ol' SGI machines, perhaps the idea of multiple specific-purpose processors and one (or more) general-purpose processors makes more sense.
Right now, these are available in the form of add-on PCI cards, but I believe that as standards converge and begin to stabilize, you'll find motherboards produced with an assortment of processors. For example, a motherboard might be produced with on-board processors for:
You might even put an implementation of X on a CPU specifically designed for the purpose, thereby offloading even more crap from the main processor. In other words, these specific-purpose processors could offload quite a bit of crap from the main processor, making your applications a lot more responsive, even while you've got a ton of really intensive stuff going on. And if you buy cheaper hardware that doesn't have one or more of these "standard" chips, it'll just happen in software, and slow down you system accordingly.
Oooooooh well.
Doesn't the MPEG-LA alliance expect a user to pay .25 cents per hour of play back?
Don't worry. The decoder card's MAC address is programmed to your credit card number.
Apparently they weren't kidding.
... In my experience, the loop connections between these decoder cards and the video card degrade the image quality so much that it's very hard on the eyes when you're NOT watching movies.
I had a hardware decoder to play DVDs on my old P2 400 years ago - it was great watching a film with only 5-10% CPU usage, but I had to keep switching the plugs around at the back to get real work done. If someone comes out with the same functionality with a card that communicates internally with the normal graphic card then I'm all for it, but I wouldn't buy another "pass through" solution...
Code, Hardware, stuff like that.
I built a dedicated box with a Celeron 333 and a Matrox G400 card with S-Video out just to stream my DiVx's to my TV set. Not so sure if testing this card on a Duron 600 constitutes a TRUE test of this cards capabilities.
Eddy.WriteLinux.Com
More cards, more power, more heat.
Opps, there goes another slot, as well as all the driver issues involved in supporting it.
I'm not the only one to see this, but it deserves to be mentioned in any list: The card will quickly get out of date compared to the newest codecs. Updates, if they come at all, will likely be so slow to appear that most new videos will always be a version ahead of the card. And, of course, if the manufacturer goes out of business, or looses interest, or doesn't support the same favor of M$ software that you want to run, you have a hundred dollars worth of junk.
I'm an American. I love this country and the freedoms that we used to have.
Nobody can sue you (and win) for writing computer programs that uses patents, because a patent must be novel. Mathematics and computer science are natural and therefore not novel.
Got friends?
What I want is a board for video capturing with a decent mpeg encoder chip on it. No a co-processor... a real mpeg encoder that does correct 3-2 pulldown detection, deinterlacing, etc. To get a really good quality video capture on a PC means capturing with any of the many video capture boards using a codec that either doesn't compress at all, or just compresses a little bit... then taking the .avi file and running it through an mpeg compressor that can do the de-interlacing, and whatnot... all to end up with a DVD spec mpeg file. Using a real-time codec usually means crummy quality.
What I want is a board that has an mpeg chip, similar to what might be in a standalone DVD recorder or a TiVo, that will take an incoming signal and spit out a good quality mpeg file in real time. Standalone DVD recorders can do it, so surely a PCI card can too.
-S
--- What parts of "shall make no law", "shall not be infringed", and "shall not be violated" don't you understand?
Sorry, you're wrong. Wish you were right, though.
You can patent software. You can then sue people who make software which violates your patent. You can win.
It used to be the case that computer programs were legally considered similar to mathematical algorithms. Being abstract ideas, they can't be patented. (It's not that you can't patent math because no math ideas are novel; it's just that you can't patent math ideas because they're too abstract to be patented.)
Now, however, most governments (including USA) and other patent-granting authorities grant patents on software. You can't create unauthorized implementations of MPEG codecs without infringing on some valid, enforceable patents.
Sucks, but it's true. It would be *very* hard to argue that an implementation of an MPEG codec is somehow "natural" and not original. It would be slightly less difficult -- though futile, since this question has been decided by the courts already -- to argue that an MPEG implementation is just a reflection of abstract mathematical ideas, and is thus not patentable.
The trolls have been on slashdot far longer than either of us, and get mod points regularly. I suppose it was inevitable.
... is an MPEG4 hardware encoder for $99...
-- iCEBaLM
- The solution involving the loop cable seems slightly antiquated, since conventional TV cards use a Conexant chip, which writes the data directly via the PCI bus into the graphics card's memory. The solution from Sigma Designs does offer one advantage, however - the PCI bus is not overloaded, thus avoiding problems, particularly with older computers. The manufacturer also cuts costs with this solution.
It's not "antiquated", it's "stupid". There has never been a technological reason for this. You don't have to use the PCI bus to write directly to the frame buffer. The "vga feature connector" has existed much longer than any local bus (even predating EISA.) The BS about bus saturation is laughable -- exactly what would be competing with the PCI transfer? As for costs, that is very much a proven lie! (They may claim that it's cheaper only because the technology is already developed and they won't have to devote developer time for a new driver interface. But don't let Sigma's marketing people fool you, the silicon for that hardware overlay is not cheap and certainly not free.)Anyone who has ever dealt with Sigma Designs, Real Magic, et. al. knows very well why the external, analog overlay is there. It's there for one almighty reason: DVD CCA licensing rules. There is zero chance the decoded content can be "stolen" in digital format. Rumor has it, even the external SDRAM on the card doesn't hold the decoded data during playback. Where I live, that's called "paranoid."
They tested this card out on Tomb Raider? My respect for Tom's Hardware just went down a notch or two.
Dyolf Knip
In the article, I mean. There's so much advertising dreck on Tom's Hardware now that it's too tough to read the articles. You get about half a screen worth of text and then have to click on a new page.
Could you provide a link to "Doc Mercury's guide to DVD ripping"? Thanks.
why not use a good one like [...] Real 9
Parse error: You used "good" and "Real 9" in a way that I do not understand.
Y2K Compliant since the late 1890s
I am no expert, but it seems that there are a whole lot of DivX and MPEG4-variants out there. Which ones does the card play, which ones will be developed in the future, and will the card play those? Do they provide specs for how to drive the card or do the users of Free operating systems have to reverse engineer those? Just some questions that popped up in my mind.
---
Thus spake the master programmer:
"Let the programmers be many and the managers few -- then all will
be productive."
-- Geoffrey James, "The Tao of Programming"
Please correct me if I got my facts wrong.
Thanks for the info. I still don't see how software is an actual "thing" since what really is going on is just a microprocessor executing instructions; couldn't someone just say they generated every possible set of CPU instructions and nullify any future patents?
Got friends?
Of course, I'm sure the price will drop to something like $20, and assume the real use of the technology is for living room equipment that is not yet out. I support that use, but a PCI card sound just stupid. Where is their market?
Neither mathematics, nor computer science (nor for that matter physics, chemistry, astronomy, etc...) are "Natural". There are patents in physics and chemistry, for sure, and patents on mathematical algorithms, too.
In particular the 8500 chip. They have a reference DVD player design, which just needs an IDE DVD transport, front panel and PSU to be up and running. And it's tiny. I like reference designs; they make it much easier for small companies to put together interesting kit that the big guys don't see a big enough market for.
As far as I know the MPEG4 standard only specifies how it can be decoded. How to get from uncompressed data to a valid MPEG4 stream, is left as an exercise to the codec programmer. Not to mention how to find the best-looking stream of all the valid ones (for a given size, otherwise bigger is better). I don't think the playback filter has changed mucb, it's the encoding that is changing all the time, adding new features and settings to squeeze the most detail out of the input stream.
Kjella
Live today, because you never know what tomorrow brings
I have an old p2400 (...)
Yeah, I replaced that old crap with a p2533 too, 2.4GHz is just outdated. [TODO: Insert witty pun about a AMD at crawling clockspeeds still being faster]
Kjella
Live today, because you never know what tomorrow brings
Personally I think that would be a lot more interesting... I'd like to have some nice little box running it. Even a mini-pc looks big compared to the DVD/mp3 players. Now a DVD/MP4 (or mpeg4 avi)/MP3 player, and I'm buying. My PC can run this without problems in software, so it really misses my marked. Or combine one of these chips with a DVR solution, would need hardware encoding and/or a network interface, but that'd also be great...
Kjella
Live today, because you never know what tomorrow brings
I'm checking the sites almost daily to see if the 1 ghz Via C3 cpu has come out yet. It's been reviewed favorably, but has not hit the market yet.
Why? Because although it's kind of wimpy, it's cheap, it has the oomph to decode DVD in software, and IT CAN RUN WITHOUT A FAN.
Perfect for my own tv-companion PC project, building a TV PC inside the case of an APEX AD-1200 DVD player (the $65 one that Circuit City sells). The AD-1200 uses a standard IDE DVD drive, so all I have to do is get a C3 in there on a small motherboard along with a power supply and hard drive. Woo Hoo!
Jon Acheson
All opinions expressed herein are my own, and not those of my employers, who are appalled.
Disclaimer: I've only played movies on my work laptop, which is 1.1GHz, and haven't tried it on my 233 MHz PII at home, since I don't currently have a broadband connection and downloading by modem is too annoying :-)
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Thanks for posting anonymously and being a jackass.
And if you think DivX 3 with nandub is better than DivX 4/5, you are a freaking moron.
Not All Who Wander Are Lost
I guess that means that Ogg Tarkin hardware is only 10 years away now!
Calm down, it's *only* ones and zeroes.