You must be the chief village idiot here, there is nothing 'unofficial' about it ...
Intel android tablets and phones are sold 'officially' all over the world e.g. Lenovo K900 in China is a top selling android phone.
and AMD really need 20nm too to rescue their inferior designs from the margin hell they are in especially the big console chips. The XboxOne chip is 363 mm2 which must make it a ~$75 die ($40 raw cost + $35 cut to TSMC) so it is obvious why they are getting poor profit from each one, $10-15. Shrinking it to 20nm would make it a $40 die and give AMD another $35 profit from each chip so AMD really need 20nm and if they are not using it straightaway in 2014 that should tell you how ready it really is.
'I would not want to hold Intel shares in 2015.'
Yeah a 50-100% gain form here would be hard to swallow when you are used to a loser stock like AMD ;-).
'Android-X86 runs on X86.'
Yeah so, just like Android-ARM runs on ARM or Android-MIPS runs on MIPS. It's still the same friggin o/s idiot ! Every piece of software written has a specific variant for each ISA it is run on.
'Just a post you ago you claimed it would be a cross between bulldozer & A57'
Just some deductive thinking based on the few public sentences Nvidia have uttered about it. It's planned to be a HPC processor primarily and you do that best with a speed racer design to get the FLOP count up so it might clock up to say 4 GHz in desktop but burn a lot of power doing so. In mobile it will have be down-clocked to 2 GHz but even then a long pipeline and hefty FPU units might preclude quad-cores in mobile devices. Considering Nvidia have given the world so little information such speculation is what we are reduced to. The stexar ex-Intel architects are also from the Pentium 4 generation so they would be familiar with doing such a speed racer design.
Well so far it all seems beyond them as the product keeps getting pushed back otherwise it would be appearing next year in Logan as it was originally planned to be. ARM A15 was only meant for one Tegra generation not two !!! Why are you attempting to pump a paper processor about which we know exactly squat ?! You have been doing it for 2-3 years now and it's got you precisely nowhere. Be more skeptical, wait for the proof first before proclaiming the second coming of processor technology. It's pathetic.
From 2005 until 2013 Stexar has not designed or released a single product so my comment still stands. All you have got with PD is HOPE !!!!!
'It's advertised for 2015... and you know that, but remain conveniently oblivious to it.'
It was also advertised for 2012, 2013, 2104 before then so how exactly did that turn out ? Why should we believe Nvidia now after such a dismal prediction record and on official investor slides too ?!!!
'OK, please tell us any info at all about PD... or rather you don't know squat about it !!!!!!!'
Neither does anyone else because Nvidia is either too frightened to release any specifications or even does not have any. I am not going to give Nvidia any credit for an unknown design especially as Nvidia has precisely zero experience of successfully designing its own cpu. It will probably turn out like a cross between Bulldozer and A57 judging from the few inferences that have been given, too hot to even use as a quad-core in mobile.
Apple is stopping itself by having expensive products in a proprietary ecosystem. The free Android is kicking it into irrelevance. The defensive comments of Anand C should tell you all you need to know about the lack of a 64-bit Krait. Nvidia's PD is always just around the corner yet still MIA even in specification ! Samsung is just going to use A57 just like the majority of all the fabless companies, good for them they can all lose money together as Intel out prices them and beats them on performance/watt. I am more aware than you of all the upcoming ARM64 products but unlike you I don't think there is anything substantial to see.
Everybody is confused on your site including you, reality is definitely not required for your tabloid site preferably suspension of reality as long as a shock headline and site hits can be generated.
Trending on Yahoo:
Dnenni does some more unsolicited spamming of his site.
p.s. all those so called 'servers' are network packet router chips, hardly general purpose servers but then again inaccuracy is a necessary element of your sensationalist tabloid site because the truth is too mundane and ordinary.
Just because you put your hands over your ears and reply like a stupid truculent 5 year old child does not mean I did not tell you ...
'"A57 has inferior performance/watt "
ARMH is a rubber stamp !!!!!!!!!!!!!!!!!!!!!'
AMD will die long before x86 ever does. Hey in case you have not noticed Intel's x86 chips run Android since last year. You have pumped AMD all the way from $42 right down to $3 today so your opinions and predictions have already been proven to be completely useless and in fact outright dangerous to anyone who wants to keep hold of their money never mind increase it..
Memory capacity is something that can easily fixed in next year's model unlike poor inherent processor performance. All of AMD's products are on the ragged edge of their development curve and there's very little more juice that can be squeezed out of them and AMD is not even going to try keeping them on 28nm way into 2015.
'The key point it that AMD has leveraged the one corner of the X86 market that is about to explode; the custom APUs for MSFT and Sony. That is huge. '
Yes the 363 mm2 monolithic die is huge and after TSMC has taken their $35 cut producing a $40 die it only leaves a few dollars left for AMD after it sells it on to MSFT for $90 like about $15. Until AMD shrinks these big console dies to 20nm and reduces them to about 200 mm2 they will never be a big earner for AMD.
2 Ghz AMD A57 30W octo-core in 2014 is DOA.
2.4 (2.6 turbo) GHz Avoton octo-core available *now* is only 20W.
1.8 (2.8 turbo) GHz Xeon E3-1230L v3 quad-core octo-thread available *now* is only 25W.
Intel will eventually make Nvidia extinct. Moore's law will make discrete gpus increasingly unnecessary and in cpu x86 will crush its ARM's efforts. Give it 5-10 years and NVDA will start looking like AMD, a slow painful descent to 0.
'If you have two companies (INtel & AMD) building the same thing,
then no one should be surprised better manufacturing wins'
Haswell and Bulldozer/Piledriver are not the same thing just as Jaguar is not the same thing as Silvermont even if they were all built on the same Intel process size. Intel is out-designing AMD as well as out-manufacturing it and it is doing the same to ARM from 22nm onwards.
'You're declaring game over for ARM64 before the players even take the field.
ARM64 will be ubiquitous and will have world support. '
I told you all the specific reasons why every ARM64 implementation is flawed. Consumers will *always* buy the better product no matter how many OEMs try to enforce an inferior 'ubiquitous world supported' standard upon them. You have fundamentally failed to learn this crucial lesson of computing history and which is why the Mainframe standard was replaced by Risc and the Risc standard was replaced by x86. ARM implementations are just not good enough to replace the x86 standard no matter how many OEMs waste their money trying. Itanium was a 'ubiquitous world supported' standard too, yet little old AMD tore this new standard up with AMD64 with mass consumer buying.