Nvidia CEO: We Have A CPU Strategy

[HardOCP News]

I read the headline and I thought the same thing you did…a CPU strategy? Yeah, it’s not what you think.

"Our CPU strategy is ARM," Huang said, referring to the fact that Nvidia was, unit last year, only a supplier of GPUs. "ARM is the fastest growing processor architecture in the world today. ARM supports (Google's) Android best. And Android is the fastest growing OS in the world today," Huang said.

 

[anthrex]

They need to release that CPU strategy. Tegra 2 is taking too long to come to market as I thought it was supposed to be released at the beginning of the year.

 

[SirKronan]

They need to release that CPU strategy. Tegra 2 is taking too long to come to market as I thought it was supposed to be released at the beginning of the year.

I thought I remembered reading pretty much only positive about the Tegra in the first place ... wouldn't a good CPU strategy just be to get that tech in more devices? And of course upgrade to Tegra 2. I was under the impression that Tegra got some pretty good responses and packs quite a little punch for the power it uses.

 

[Peteman100]

I thought I remembered reading pretty much only positive about the Tegra in the first place ... wouldn't a good CPU strategy just be to get that tech in more devices? And of course upgrade to Tegra 2. I was under the impression that Tegra got some pretty good responses and packs quite a little punch for the power it uses.

The CPU part of Tegra was pretty tame, which is fixed in Tegra 2 with a shift to Cortex A9

 

[niconx]

ARM is for girly men sporting smartphones and matching handbags.

It would be hilarious if Nvidia released an upscaled ARM on par with quad core x86 chips.

 

[Stoly]

Actually it's exactly what I thought.

ARM is the easy choice for nvidia. There's a huge potential market with mobile devices, be it phones, media players, slates, netbooks and laptops. And there's the potential por nettops and even htpcs.

Unfortunately Tegra is late.

Back when It was announced, it was going to revolutionize the industry, they even had working prototypes for many products ready to market, but none of them ever did. maybe it was too much too soon, or too expensive, or OEMs just didn't like it.

Fact is that to date only 3 devices use tegra chips and 2 of them are utter failures (the kin one and 2), that makes the Zune the only moderate successfull tegra based device.

To make matters worse there will be no windows 7 phones based on tegra, and the boxee box is taking so long that its starting to sound like vaporware.

Still there's hope. Android may very well be tegra's salvation. Rumors says Tegra 3 has just taped out, and spec wise it seems it could take on anything out now and in the near future.

But the problem for nvidia is not the hardware, the problem is no one is buying.

 

[Saist]

Stoly: there is one other successful Tegra implementation.

Volkswagen.

More preciously, the Audi division of Volkswagen. http://gizmodo.com/5448170/audi-turning-to-nvidia-tegra-chipset-to-make-their-dashboards-pop

 

[Stoly]

Stoly: there is one other successful Tegra implementation.

Volkswagen.

More preciously, the Audi division of Volkswagen. http://gizmodo.com/5448170/audi-turning-to-nvidia-tegra-chipset-to-make-their-dashboards-pop

I've read that tegra is being tested for integrating into cars from Audi, Citroen and Maserati as entertainment centers, GPS, driving assistance and navigation, but there's none out yet. And since there have been hundreds of products announcements that never came to market, I rather not mention it.

But from the link you posted, at least Audi is serious about it.

 

[Sikkyu]

why do i get the feeling nothing will happen and I wont care at all ina bout 30 sec?

 

[RumpleForeSkin]

Stoly: there is one other successful Tegra implementation.

Volkswagen.

More preciously, the Audi division of Volkswagen. http://gizmodo.com/5448170/audi-turning-to-nvidia-tegra-chipset-to-make-their-dashboards-pop

They've been putting steering wheels like this in concepts for decades now, other than some Italian sports cars, when has anything other than the basic wheel been actually realized?
sweet dash though.. I can only imagine how [H]'ers will mod it

 

[Lorien]

But the problem for nvidia is not the hardware, the problem is no one is buying.

The problem IS the hardware.
They promised specs and got many design wins based on those promised specs. Fast forward to delivery of final silicon and all those promised paper specs were nowhere to be found. You know like the initial Fermi paper specs?
You are right on one thing, OEM's didn't like it and rightfully so. In the mobile market you can't pull off a Fermi (late, too hot, too power hungry) because the OEM has the final word and you get chosen by delivering efficient and powerful chips on time. Tegra definitely fails in 2 out of 3 of those metrics.

 

[bexamous]

Tegra2 is turning into something pretty good with Android. Its hard to say Tegra1 was bad hardware as Zune is awesome, Kin is a shit product but hardware is fine.

 

[Stoly]

The problem IS the hardware.
They promised specs and got many design wins based on those promised specs. Fast forward to delivery of final silicon and all those promised paper specs were nowhere to be found. You know like the initial Fermi paper specs?
You are right on one thing, OEM's didn't like it and rightfully so. In the mobile market you can't pull off a Fermi (late, too hot, too power hungry) because the OEM has the final word and you get chosen by delivering efficient and powerful chips on time. Tegra definitely fails in 2 out of 3 of those metrics.

What promised specs? the ones at the nvidia page are impressive to say the least even for 1st gen Tegra. Keep in mind that Tegra was announced in early 2008, with working devices (nvidia prototypes), but only made it to the ZuneHD in late 2009.

 

[sbuckler]

The current cpu of choice - snapdragon has been around (working) since late 2008, and is based on the ARM A8 I think.

Tegra 2 is newer then that with dual A9's- not a lot uses A9's yet.

I do think nvidia got it wrong - focusing on windows mobile too much - but even if they had got it completely right the tegra 2 soc wouldn't be shipping yet anyway. I expect it'll start to appear in more stuff (e.g. android tablets) towards the end of the year.

 

[housecat]

This seals the fate of Nvidia. Here's to AMD and Intel pushing them out of THEIR (PC) market completely. Good riddance, hopefully a little more professional outfit will take their place.

 

[Stoly]

The current cpu of choice - snapdragon has been around (working) since late 2008, and is based on the ARM A8 I think.

Tegra 2 is newer then that with dual A9's- not a lot uses A9's yet.

I do think nvidia got it wrong - focusing on windows mobile too much - but even if they had got it completely right the tegra 2 soc wouldn't be shipping yet anyway. I expect it'll start to appear in more stuff (e.g. android tablets) towards the end of the year.

I think nvidia, pitched tegra to MS, and when they got the ZuneHD thought they had secured a major design win with winmo7 phones, there was a lot of buzz late last year about an upcoming nphone/zune phone with tegra and winmo7.

Something happened since then and AFAIK Winmo7 will not support tegra at all. So nvidia had to quickly move and embrace android.

 

[SirKronan]

This seals the fate of Nvidia. Here's to AMD and Intel pushing them out of THEIR (PC) market completely. Good riddance, hopefully a little more professional outfit will take their place.

You had better hope you're wrong, for all our sakes. I certainly pray that you're wrong.

 

[Zellio2010]

Combine tegra with fermi for a nice portable bomb

 

[lightp2]

Here's something probably already well-known

1. nVidia is cornered in the chipset market. Due to Intel/AMD both firmly establish full-scale platform strategy, (not only establish, but enforce with multiple integration). nVidia does not appear willing to pursue new AMD chipset business, and Intel does not grant access to new platforms.

2. nVidia has reached the top tier in its segment. In this market, Success is 99% hardwork and very little miracle. Competitors are always aiming. nVidia also need volume to sustain itself. In this top-tier scenario, un-fortunately that implies its fortune is tied to bleeding-edge electronic manufacturing, which again goes back to 99% hardwork, plus the Fab race.

2a. In the old days, nVidia can count on slower pace of tech advance, so every dollar of investment can sustain, example multiplier of X years. However, in this edge of hasten society, there's very little patience, the multiplier is now (X-Y) years, that means you have little time to sit on your good fortune.

3. Since nVidia original fortune is built on x86 chipset, consumer GPU, Professional GPU, and since Unix workstation market is now percentage-wise relatively small, you can generally say nVidia fortune is built on x86 ecosystem.

4. Tesla GPGPU computing is attempt to build sustainable new high-margin revenue, however the market is smaller. Margin is very high, but you need volume to sustain a big size company.

5. VIA is now viable, but surely nVidia+Via will have so many potential issues it may not worth the trouble unless you have extremely co-operative environment. As someone else said, just ask AMD how difficult to compete in this business.

6. Due to FTC PCI-Express settlement, the consumer PCI-Express GPU business still has several useful years of revenue. nVidia has a buffer to build something concrete now.

7. Tegra series are the alternate strategy in case x86 dependency collapse. Due to dependency on x86 ecosystem for all these years, ARM means you need to change gear for new ecosystem.

8. In the x86 fields, Microsoft, Apple provides the main software options. Linux handles the OSS alternative with Redhat/Novell/Ubuntu pushing the boundary, so for nVidia, as long as it has great driver support, everything good.

9. Switch focus to ARM, and you find the following

9a. Apple has its own ARM plan
9b. Microsoft seems to have some plans by signing with ARM architectural license.
9c. HP has its own plan of WebOS
9d. Dell has plan on Android tablets
9e. many more

observe that each one has some kind of alternative in place. None of them are the kind of business that you sell the hardware, give the driver and forget.

Why, ARM currently has no coherent ecosystem support. You cannot write once and expect it runs everywhere. Android ARM is the common ground for 3rd parties and no-name imports. They could be very good but nVidia needs volume fast. But Oracle does not intend to make it easy for Google Android.

And then you have Intel/AMD gate crashing with incoming low-power products.

So you see, irrespective of AMD or nVidia, eventually both realize one thing : they are missing/shortcoming on the software-side.

One of my unknown reading friends can make a salvation plan for nVidia. It does take determination

Do not ask me anything. For the record, I guarantee have no idea. Cheers

 

[Stoly]

#2 nvidia is at fault for that. They pushed their 6 month product cycle, at first it was great for them, they remained the fastest solution out there and brought out many products very fast. ATI (back then) just couldn't keep up. But now even 12 month cycles are hard to keep up with. GPUs are getting more complicated than ever, factor in that TSMC won't be able to build 28nm products till mid/late 2011 and it just makes things worse.

#9 Its clear that nvidia has no place in Apple's IOS, HP's WebOS or MS's WinMo7. The only choice they have is Android. Actually I think its the best choice even if they could somehow get into one of the others.