Nvidia CEO Admits Tegra 4i Didn't Pan Out as Hoped

[somebodyspecial]

Go read the cnet aritlce yourselves. Tomshardware is yet again misleading people here. He didn't say they had no interest in phones. What he said was there is no point in chasing CHEAP COMMODITY phones. I agree. They should be going after Apple/Samsung Iphone/Galaxy S5 etc. You wan to be in $400-700 phones where all the cash is made. Apple isn't making $40Billion+ a year on $100-$200 phones. They don't even have one in this range.

Leave the junk to other people and concentrate on getting into apple/samsung/google, or heck, make your own MAXED out feature phone perhaps with two battery sizes and sell it yourself or let others slap a label on it also (phoenix was aiming too low, get a model together that tops S6/Iphone6 coming soon. One for the thin people, one for the people like me who will take a phone 2x thicker for massive battery life. I think many would be buying thicker phones if these people quit giving us ONE option (thin). I don't need a paper thin phone, I want a massive battery. I do not understand the quest to get these things to credit card depth at the expense of battery life. They are so thin now you almost feel weird holding them. My pocket isn't getting smaller yearly. I have a headset anyway, and never really hold a phone today. I'm not talking a 10lb phone here, but they could easily add a few mm to the back for more battery life and may find a ton of people have ALWAYS wanted that rather than thin crap with no battery life.

At any rate the high-end is where NV lives and breathes on desktop and wisely they seem to be aiming there in mobile now. I also hope they put out a 125-175w console soon to compete directly with xbox1/ps4. On top of that I'd like to see them fund say $100mil for android games. IF they fail to sell really well there you move them to PC in an easy port. I don't really see how you can lose money with the size of these two markets (no need to wait for millions of console sales over 7yrs, mobile has 1.2B and PC's have a ton also). I'm talking 1mil-5mil games here that are exclusive to their stuff (unless it fails maybe then open it up to NON NV devices) maximizing perf on THEIR hardware. Who better to make a great game than the people who KNOW their own hardware inside out and have been working with devs for 20yrs (AMD should be making games too)? You already have a store, and kickstarter and small indie teams are showing you can build GREAT games for under 5mil easily if you don't have to pay a big publisher and all the fat that comes with them (EA etc...).

Torchlight (1/2), shroud of the avatar, shadow run returns, legends of grimrock, torment tides of numenera, project eternity, wasteland 2, elite dangerous, star citizen etc etc. Even a AAA game like witcher 1/2 were 7mil for PC. Consoles added 15mil more but I'm only talking android with a 2 week or less port to PC at some point. You want to kill consoles, not help them so keep them exclusive. Intel's payments can fund this for two more years easily and it sells cards and mobile units. Half life 2/portal etc is a good start but you should be working on new IP nobody else has that are great fun but low cost with small teams. Or just pay someone like Gameloft to make games for you on mobile.

One more thing: put out a PC type ARM Denver etc box at some point with NV gpus in them. PCIE/NVlink, pick one and get it done once Denver hits. This is how you invade Intel without x86/x64. Run the soc at 3.5ghz+ with a PC like fan/heatsink and put two inside ($40-60). You can build a PC with no MS/Intel inside. An NV soc+gpu+free OS that is hopefully a triboot of maybe steamos, android, linux and a 60-120GB SSD with a slot for a HD if you want to add it yourself later to keep it cheap and reliable. Or sold as a more expensive unit with 1TB or something as storage SSD for boot).

I really hope they're working on porting steamos to ARM (logical for NV/valve even AMD). That is a great partner who doesn't care about the hardware inside the box, they just want to sell games on their store to kill DirectX at some point. These two can force the market into OpenGL over DirectX just for portability and steam has 75mil users now. This is almost what consoles have in each market (xbox360/ps3 etc) after 7yrs (about 90-100mil last I checked). That's a great place to shift gaming gear with your hardware inside. I hope AMD does the same in this regard. Valve/google need to be making games for android also (or in Valves case, games for steamos on ARM). Both AMD/NV can make a mint stealing from WINTEL and DirectX. Eventually apps will follow the games as we are already starting to see adobe etc coming on mobile. Just as games are improving, so is the apps side albeit a little slower. I think the apps will amp up with 64bit and 4GB+ (shield 2 coming with 4GB, phones too all over soon) and an ARM box that resembles a 200-500w PC. Time for real apps to move over so people who can't afford a PC (with windows+intel fees on top), can buy a box that does everything general users need.

AMD might want to build a box with a SOC and APU inside so it can swing a quad OS adding windows maybe. They are the only ones that can do this unless Intel starts pumping out ARM socs, so a good market only they'd have. Socs are $15-40 so not too expensive for AMD to make a dual purpose box on that side, but not sure how much that raises a board price with two sockets. Maybe they can just integrate an arm core easily instead (on the apu). They could both share the GPU maybe, so you're only integrating the ARM CPU side not a 2nd gpu also. Not sure but I see a market here only AMD can have for now.

 

[InvalidError]

You wan to be in $400-700 phones where all the cash is made. Apple isn't making $40Billion+ a year on $100-$200 phones. They don't even have one in this range.

There is one major problem with that: most of the features on $700 phones are available in ~$300 ones and the low-end bar is rising much faster than the high-end one does so the justification to continue paying huge premiums gets considerable erosion every year - even Apple got forced into lowering their entry-level prices not so long ago due to much cheaper Android devices with comparable feature sets proliferating like wildfire; Apple did not want to produce the mini (originally sworn never to produce anything in the neighborhood of 8"), market forces forced them to reconsider and the mini turned into one of Apple's most successful iPads ever.

The downward pressure is great and is only going to become greater as more device parameters hit practical limits that level the playing field across the board.

 

[somebodyspecial]

Fun fact: At the current rate of Intel's profits shrinking yearly, they'll hit NO profits in ~6yrs. That can easily happen if they start taking desktops shortly like they did the 21% of notebooks. You're not getting into their world, but they are clearly getting into Intel's already.

Price target on NV just got raised to $26 from 21 the other day. Looking at the numbers I agree. My shares are all $13-16.67 (with the avg in there somewhere). 😉 I wouldn't touch Intel's stock with your 10ft pole let alone mine until they change course

 

[somebodyspecial]

You wan to be in $400-700 phones where all the cash is made. Apple isn't making $40Billion+ a year on $100-$200 phones. They don't even have one in this range.

There is one major problem with that: most of the features on $700 phones are available in ~$300 ones and the low-end bar is rising much faster than the high-end one does so the justification to continue paying huge premiums gets considerable erosion every year - even Apple got forced into lowering their entry-level prices not so long ago due to much cheaper Android devices with comparable feature sets proliferating like wildfire; Apple did not want to produce the mini (originally sworn never to produce anything in the neighborhood of 8"), market forces forced them to reconsider and the mini turned into one of Apple's most successful iPads ever.

The downward pressure is great and is only going to become greater as more device parameters hit practical limits that level the playing field across the board.

The junk ones don't sell well in usa though (ie only apples top one sells here, they had to reduce 5c orders etc). I get your point though, but there's enough USA sales etc to make anyone money in the top end with the right models. Some people buy the best no matter what over and over. IE Jen commented most of the people buying titan buy it again every year because they can and want it. The same is true of phones/tablets. The money is made on the top end, and a lot of people buy each model in their fav brand or ecosystem (ios or andriod, apple or samsung etc).

We pay here to fund that downward pressure in developing countries, while they up the ante on us here constantly. Note you can buy a phone over $700+ off contract now. I can build a PC for less. They are not coming down in our country. That was $600-650 before. Now they have some going for $800+
http://www.newegg.com/Product/Product.aspx?Item=N82E16875176733&ignorebbr=1&cm_re=galaxy_s5-_-75-176-733-_-Product
Jump from 3g to 4g is almost $125. Ouch. $815! If they were under pressure they'd be $550-600. Every year I look at a new phone, I see another $50 on top these days. Also we buy status symbols a lot here, so taking a $300 one to class isn't as cool as carrying the latest and greatest. Nexus is the only cheap phone that competes with the high-end at a low price. That mediatek etc junk doesn't count at $200-300. We wouldn't buy that and don't. Everyone I know has a samsung, apple, Lg or HTC and all top of the line, except for me...ROFL. I don't buy Titan's either 😉

The keyword in your sentence is "MOST" features. We buy ALL the features for some reason (ego, bragging etc...slap any name you want on it). Most of my generation will forget about a house and buy a car way out of their range, toys they shouldn't have etc and live in a shack IMHO. It's comic to me when I see a 40-80K beamer sitting in the driveway of a mobile home...LOL. We Americans are a funky bunch. I see more people in AZ in rags (look like homeless people) that are wielding an iphone than no phone. Even the poor have top end.

 

[Vlad Rose]

Fun fact: At the current rate of Intel's profits shrinking yearly, they'll hit NO profits in ~6yrs. That can easily happen if they start taking desktops shortly like they did the 21% of notebooks. You're not getting into their world, but they are clearly getting into Intel's already.

Price target on NV just got raised to $26 from 21 the other day. Looking at the numbers I agree. My shares are all $13-16.67 (with the avg in there somewhere). 😉 I wouldn't touch Intel's stock with your 10ft pole let alone mine until they change course

And only at 18.5 yesterday and today so far for Nvidia. Intel is not a dumb company at all and wouldn't be around for so long if they were. I remember when everyone thought Intel was done for when the Athlon CPU trounced the Pentium 4s. And again when AMD was the first with a practical 64-bit x86 compatible CPU. Look at the situation now. Intel is aware of the threat of ARM cpu's in the market and have been preparing against them. Look how much the Atom line has improved each generation (still weak by x86 standards). Intel does also make ARM chips (as well as other architectures), which most people don't seem to be aware of. I really don't think Intel is going away from it's current position any time soon. Nvidia on the other hand has a tendency to piss off other corporations, not a smart thing to do.

 

[InvalidError]

Fun fact: At the current rate of Intel's profits shrinking yearly, they'll hit NO profits in ~6yrs. That can easily happen if they start taking desktops shortly like they did the 21% of notebooks. You're not getting into their world, but they are clearly getting into Intel's already.

Most of this year's consumer ARM-based devices so far are almost carbon-copies of devices launched over the past two years with slightly improved specs and often much lower prices. Stagnation is starting to set in just like it did with PCs a few years ago but the market has not reached saturation yet so there is still plenty of room left to sink new products in.

Even enthusiasts have limits over how little incremental improvements can be before they quit bothering with them. SoCs are starting to enter that zone and products using them will follow. I give that about three years before the PC history repeats on mobile computing.

 

[somebodyspecial]

Fun fact: At the current rate of Intel's profits shrinking yearly, they'll hit NO profits in ~6yrs. That can easily happen if they start taking desktops shortly like they did the 21% of notebooks. You're not getting into their world, but they are clearly getting into Intel's already.

Price target on NV just got raised to $26 from 21 the other day. Looking at the numbers I agree. My shares are all $13-16.67 (with the avg in there somewhere). 😉 I wouldn't touch Intel's stock with your 10ft pole let alone mine until they change course

And only at 18.5 yesterday and today so far for Nvidia. Intel is not a dumb company at all and wouldn't be around for so long if they were. I remember when everyone thought Intel was done for when the Athlon CPU trounced the Pentium 4s. And again when AMD was the first with a practical 64-bit x86 compatible CPU. Look at the situation now. Intel is aware of the threat of ARM cpu's in the market and have been preparing against them. Look how much the Atom line has improved each generation (still weak by x86 standards). Intel does also make ARM chips (as well as other architectures), which most people don't seem to be aware of. I really don't think Intel is going away from it's current position any time soon. Nvidia on the other hand has a tendency to piss off other corporations, not a smart thing to do.

The target was 21 for the $18 stock (actually below $18 as the $21 target was put out before that). Moving it to 26 says where they believe it is going, not that you magically wake up tomorrow and it's $26. They mostly talk in 6,12, or 18 months for price targets and if they go smaller they'll tell you this is a short term play (or something like that).

RE: the intel makes arm comment:
http://www.cnet.com/news/intel-used-arm-chips-in-wearables-demos-at-ces/
"However, Calder said that didn't mean Intel would help partners build products that use ARM chips."
They have an Architecture Arm lic but don't plan to use it for their own chips. What they are making so far is communication stuff not a SOC for mobile. IE, not directly competing with their cpus/socs.
Altera and Netronome so far & Marvell will be the 3rd they fab (if true).

 

[somebodyspecial]

Fun fact: At the current rate of Intel's profits shrinking yearly, they'll hit NO profits in ~6yrs. That can easily happen if they start taking desktops shortly like they did the 21% of notebooks. You're not getting into their world, but they are clearly getting into Intel's already.

Most of this year's consumer ARM-based devices so far are almost carbon-copies of devices launched over the past two years with slightly improved specs and often much lower prices. Stagnation is starting to set in just like it did with PCs a few years ago but the market has not reached saturation yet so there is still plenty of room left to sink new products in.

Even enthusiasts have limits over how little incremental improvements can be before they quit bothering with them. SoCs are starting to enter that zone and products using them will follow. I give that about three years before the PC history repeats on mobile computing.

The perf of these socs are growing massively and we're about to hit a die shrink which will vault them ahead massively again. Desktop gpus in mobile will sell some gamers for sure (all the ones NOT buying xbox1/ps4 due to cost and expensive games and not that many of them yet). When your phone or tablet is xbox360/ps3 level or above you'll see more sales. Gartner/IDC predictions say 2-2.5B units shipping by end 2016. Considering we are at 1.2B now, we have a long way to go before saturation. Even then the march into Intel's turf will take some years and it will be a while before developing countries can even purchase a higher end model of anything.

Sure there's a point where the market gets saturated....obviously. But there is a about $80B+ in mobile/pc (not including apple) for AMD/NV to feast on since neither has much share here. AMD can't compete in x86/x64, so start using ARM who keeps doing much of the work for you (which they are starting to do more now). AMD/NV both coming with custom cores and the two top gpus can do a lot of damage to Wintel using ARM with a free OS. Don't forget a good portion of purchases each year are just due to drops, cracked screens, getting soaked, losing your phone/tablet (stolen, whatever) etc. I submit the war won't stop at saturation, as we'll see quite a few years of ARM bleeding Intel's world unless something changes (like Intel buying NV). NV will certainly have a ton of land to grab, and AMD to if they can last long enough to get into the game.

 

[InvalidError]

The perf of these socs are growing massively

There is no massive improvement on anyone's roadmaps for this year aside from 64bits ARM which may not necessarily be that much faster than 32bits so on the CPU side; there may be relatively little change of consequence at the high-end beyond the IGPs.

As far as die shrinks go, mobile devices will be even more heavily biased towards lower costs and lower power than desktop chips were and we all know how little good the past 2-3 die shrinks have been on the PC side as far as raw performance goes... ~30% improvement in about five years for process and architecture improvements combined. I see no reason why ARM-based designs would magically fare significantly better unless ARM's in-house IP-core was amateur crap and third-party re-designs ended up running circles around it. More CPU cores would be of little use due to lack of software that makes meaningful use of them and more IGP cores would get hamstrung by the single-channel 64bits DDR3 controller most SoC manufacturers are sticking to due to cost, complexity and power. (Doing 128bits would roughly double the amount of power consumed by RAM and the memory controller.)

And power is not only a problem for battery life: many people often overlook that you still have to hold the damned things so the SoC's power has to be low enough that the device does not get uncomfortably warm. Pumping more power in a mobile SoC and its support components is often not an option even if the device had infinite battery power.

AMD/NV both coming with custom cores and the two top gpus can do a lot of damage to Wintel using ARM with a free OS.

Android and most other popular mobile OSes also run on x86 too so there is no shortage of "free" OSes for Intel's SoCs and most mobile apps are written in Java (or C# for Microsoft's mobile OSes) that gets compiled into machine-agnostic intermediate bytecode that can run on any CPU as long as the OS has the necessary interpreter or native recompiler. If AMD had wanted to, they could have gotten in the mobile space using their x86 and GPU cores too; no need for ARM... but most of AMD and Nividia's discrete GPU might goes down the drain when it all has to go through a 64bits DDR3 interface shared with the CPU.

From the looks of it, Google's Nexus 8 will be using an Intel Moorefield Atom SoC.