facebook rss twitter

NVIDIA finalises Tegra 3 design

by Pete Mason on 16 August 2010, 14:34

Tags: NVIDIA Tegra, NVIDIA (NASDAQ:NVDA)

Quick Link: HEXUS.net/qazla

Add to My Vault: x

When NVIDIA's Tegra SoC was released in 2009, it was heralded - at least by the manufacturer - as a revolution in portable media.  Offering low-power usage and the ability to decode 720p video with ease, it lived up to those promises on the few devices that made use of it.  Now a report from SemiAccurate is suggesting that the design of Tegra 3 has been finalised, ready for a mid-2011 launch.

Missing in action

At this point. you may be wondering what exactly happened to Tegra 2.  Well, not a lot.

The new dual-core chip, which doubles up the graphics performance of its predecessor and supports 1080p video decoding, has thus far struggled to make an impact at retail.  While there were a few tablets floating around at Computex powered by the SoC, no major manufacturers have managed to bring one to market.  The closest example might be Notion Ink's Adam and even that's still a few months away.

Having said that, it's not as if the first iteration was particularly successful.  NVIDIA's own website lists only the Zune HD and the Boxee Box as examples of current Tegra powered devices, and we all remember what happened to the world's first Tegra-powered smartphone, don't we?

Catch a Tegra by the toe

Unfortunately, there's not even much information on what Tegra 3, which has been given the product name T30, will bring to the table.  Improved performance and lower power consumption seem obvious, but in terms of new features, it really is anyone's guess.

If Tegra 2 devices manage to make it to retail by the end of this year, we could conceivably see the follow up being available by the end of 2011.  The Tegra family does appear to have plenty to offer, but can NVIDIA find partners that share its enthusiasm?

HEXUS Forums :: 6 Comments

Login with Forum Account

Don't have an account? Register today!
Posted by aidanjt - Mon 16 Aug 2010 16:52
The only way Tegra will get any support is if they support an embedded OS people give a crap about, i.e. Linux. Since they haven't, nobody will waste time and money on an SoC with no feasible platform.
Posted by drunkenmaster - Mon 16 Aug 2010 18:15
I doubt its that much down to os support, as the infamous and usually right Charlie suggests, people who make smartphones and mobile devices have incredibly strict design specs, if a company, Nvidia, Intel, AMD, Broadcom or whoever say use our chip, here are the specs. Companies like Nokia, MS, Apple, everyone else will go and make a design for it, and have an OS that works well enough, when the chips are ready(because designs will start anything from a year to 3 years before that point(somewhere inbetween for phones I'd guess) a sample is sent to the companies and IF it matches the specs, doesn't use too much power, heat, and performs correctly it will probably be used. If the choice is between two chips that work fine, the cheapest would get used.

Either way, the massive number of “design wins” Nvidia goes on about, and went on about at length for Tegra 1 and 2, are just that, they are the early designs that will go ahead ONLY if the chip comes in on spec. THe fact that almost every single design win ended up in a product using someone elses chip should tell you one simple thing. Nvidia couldn't match up the end specs to what they promised. 0.05W power usage over spec and the battery and design thats been made simply won't work, or will result in an uncompetitive product so the “design win” gets shoved in the bin.

Whats worse is, for what 3 years now Nvidia has had design win after design win, which ended up in the bin. Companies remember this, if Nokia have designed 10 phones over the years to use Tegra, then tegra 2, and they all fail, will they really even ask for the specs of Tegra 3 or will they simply give up even answering the phone if Nvidia is calling.

Nvidia products aren't slow, but Apple are moving away from them because they weren't reliable, people won't use a brand they don't trust, not end users(who are largely ignorant) but Apple seem to be refusing to use a company that had many failing products for them, over charge, aren't hugely competitive anymore and can't be trusted to bring in anything on time or on spec.

Tegra 3 might be dead in the water, because everyone who might want to use Tegra 3 in their products, has already seen 1 and 2, and decided Nvidia couldn't provide a worthwhile chip.

Their name might simply already be burnt in that particular market(and several others).

This isn't helped by the firing of the “fall guy” over the failure of the Tegra chips, the problem again being that, he was just fired, the team behind Tegra 1 and 2 made chips that couldn't be built on spec, and aren't being used, they've spent HUGE sums on R&D that failed to make any of it back, and that same team was behind the Tegra 3 design as obviously at this stage Tegra 3 will have been in the works for well over a year.

Tegra 3 has about as much chance as T1 and 2 right now, TEgra 4 might be reworked by an entirely new team and might be competitive, by then, will anyone risk spending money on designs, testing, manpower on Nvidia again….. probably not.
Posted by aidanjt - Mon 16 Aug 2010 20:22
Tegra is actually otherwise a good chip power/w wise. How many complete SoC's are out on the market which requires <1W of juice and includes just about everything bar the kitchen sink?

And while nVidia took some bad reputation hits because of mGPU and chipset failures, Tegra itself hasn't been particularly cited as being unreliable.

Nope.. I'm afraid Tegra's failure is a matter of nVidia shooting themselves in the foot wrt to OS support, or lack thereof.
Posted by kalniel - Mon 16 Aug 2010 21:12
aidanjt
Nope.. I'm afraid Tegra's failure is a matter of nVidia shooting themselves in the foot wrt to OS support, or lack thereof.
I think they were going for a big win off certain companies who like to design their own very proprietary OSen. Only they kind of pissed them off instead.
Posted by drunkenmaster - Tue 17 Aug 2010 00:34
aidanjt
Tegra is actually otherwise a good chip power/w wise. How many complete SoC's are out on the market which requires <1W of juice and includes just about everything bar the kitchen sink?

And while nVidia took some bad reputation hits because of mGPU and chipset failures, Tegra itself hasn't been particularly cited as being unreliable.

Nope.. I'm afraid Tegra's failure is a matter of nVidia shooting themselves in the foot wrt to OS support, or lack thereof.

Thats the problem though, it was DESIGNED to be incredibly low power, I don't think theres a single official spec out of what power they are actually using though. 1W out on a 480gtx wouldn't matter in the slightest, not a single person in the world would care, 1W out, on a chip designed to run at 0.5W, would be a complete disaster.

If these, I think 100's of design wins were, well, good enough in terms of hardware, these companies that spent a LOT of money to make the designs and produce prototypes ready for the final silicon WOULD have sorted out the OS's to make Tegra work fine, dozens of companies and 100's of products don't get canned because no one can be bothered to make it work.

Isn't it a basic Arm core with a gpu added, so shouldn't any OS that supports the arm architecture essentially support Tegra in the first place, the gpu half of the chip is little more than getting drivers to use it surely.

Isn't the whole idea of using an Arm cpu that everyone supports them and time to market is pretty damn quick and cheap because you've got a well supported, compatible cpu to start with?

SEE NEWER »