Toshiba's Cell TV
- Thread starter patsu
- Start date
The original Cell TV used a 1:8 Cell BE. Spurs doesn't have a PPE.
SUPRSEngine is 4 SPEs, no CPU, allowing whatever other CPU to be connected. They are (were?)using SPURESEngine in some TVs, but the original uber-TV was a full Cell BBE.
Gosh, so Cell only in name regards concept of cellular processors. I'm surprised, as Toshiba's idea for Cell originally was just SPEs, like SPURSEngine. To go with complex multipurpose multicores is a deviation from their path. One also wonders what is it about Cell that generates so much heat versus this solution of theirs? It sounds rather as though they are forgoing programmability for specialisation, with the custom processing chip in the processing-sandwich's centre. This was always a concern for Cell's future, whether its performance and programmability that could drive media devices would ever be competitive with ASICs. History is telling us 'no'.
CE space will evolve with new devices like iPad and GoogleTV (App Stores !). ASICs alone may not cut it. OTOH, GoogleTV's Atom CPU is too d*mn slow. So we may still need general purpose power, programmability, plus specialized hardware. Toshiba TV alone may not foretell the complete CE future (e.g., They may have executed badly).
Which are stupid. At least, integrating them is. The display should just be a display. The gubbins to play stuff on it should come from a separate box, that can be upgrade as needed, or replaced when it busts. I can see a need to replace a digibox every three years or so, but a TV can last much longer.
More on topic though, I don't see how this evolution is good for Cell versus other systems. It's probably more likely TVs will go with ARM and SGX or something - some architecture commonplace and understood and with current software. I can't see how Cell could find its way into TVs now. Even if it's the better architecture overall due to flexibility, the Cell ecosystem has grown beyond console games and a bit of supercomputing.
Even IBM shuttered the Cell resource center site recently. Seems pretty dead an end.
I suspect displays will find its way into more devices... integrated rather than staying "standalone".
I think the point was we still need programmability in CE space. ASICs alone may be insufficient.
Yap !
ASICs evolve too...just because it's "application specific" doesn't mean it won't be able handle future applications. Looking at today's low cost "media processors" they could handle all kinds of apps. I mean just take a look at Broadcom's portfolio. They've got ASICs for everything. Want to support new apps? Update the firmware.
That's the point though. You will always want a CPU, but it can be a lightweight component with ASICs to drive the heavy workloads. Is there room for powerhouse CPUs in CE devices? Don't think so. Maybe something like IMG's GPGPU focussed SGX 554 will be the platform of choice, rather than FPU heavy CPUs?
Whys that? A TV nowadays needs to be able to decode Mpeg2 and H264 anyways. Why not add simply an USB port and mediaplayer, the TV should be able to provide the best possible picture if evering from the source to the display is known.
Certainly better than CE devices that upscale their output (or worse converts framerate) and then the TV needs to work with mangled data and lacks information of the source format which it could process much better without eg. duplicate frames added by the CE Box. Unless HDMI gets replaced with something better, more "networky" solutions by removing all those imposed restrictions of decade old hardware that couldnt buffer input I cant think of a better way to play back media.
I certainly wouldnt want a seperate box (maybe with a small whiney fan bc of space and heat concerns) for media playback via USB/LAN, a TV already has the necessary components included so why dont use them.
On a similar note, I dont understand why no one integrated a webcam to a TV, Skype seems to be a perfect match aswell.
edit: Unless you meant adding something like a webbrowser/office suit to a TV. Im not really interested in that.. but it still makes more sense to me than any Smartphone =)
It depends on what you want to do. None of the ASICs based BR player could transit to 3D Blu-ray for example. That is why PS3 is an exceptionally good platform for Sony to spearhead unproven media heavy/power intensive CE markets. The 40-50 million install base makes a very strong business case to persuade people to come along.
What went wrong is the development and business plans were abandoned half way, and not updated.
A 5-year old Cell can't compete with the latest ARMS, IMG and Tegra. So vendors should choose whatever make sense for them.
Many ASICs have some sort of CPU core...
Another big factor is that CE companies come out with newer products every few months and they want to sell them, they do not want to keep releasing new software to run on old discontinued models which do not continue to bring in profits. That is why you don't compare CE devices with "game consoles"...eg consoles are sold for years so it makes sense to use the CPU that's already there to do CE functions using software. Doesn't make sense for CE devices of which new models come out every year to replace old models.
Last edited by a moderator:
