ARM will NEVER gain any serious market share in the real world of the Data/Storage Center businesses because their chips are not able to perform or measure up to the standards needed in the mission critical demands of the "actual" Data Center World.
FYI, even the lowest of low ARM CPU's can saturate an IO interface.
I think I'd like to hear BK address how Intel is going to deal with people like Ashraf, Rasgon, etc. who are clearly agendists trying to manipulate the stock. After a simple "cease and desist letter" there's a lot of options. Ashraf is constantly attacking "we the people" who hold Intel stock and see it, overall, as a balanced and noble investment. I would be delighted to see a defamation damage suit for some nominal amount, say $10 M for this misinformation promulgated by this guy...such as the "product delayed by a year" based upon a mis-construal of an outdated job ad, etc. I am sick of this guy yelling "fire" in the theatre every time I want to watch the show.
How many shares did you sell when you listened to Ashraf. Was it 40K? Is your issue with Ashraf because of this?
I doubt that anyone could go after Ashraf, but for giggles, why don't you try?
So brm, a lot of hoping and wishing on your part when it comes to ARM servers. If Qualcomm comes crashing down losing a significant part of its modem business with Apple, Samsung will be the only one that can continue to invest in ARM server efforts. If 3DXP takes off and puts a dent in the DRAM/NAND business, Samsung will have bigger problems to worry about. Of course, we'll see what happens.
No, not at all. I think you are misunderstanding what I was suggesting. The article is interesting because it's talking about CPU dominated workloads (ie, running facebook's php environment). The ARMy can't compete in this space, at least with known CPU's. While I know nothing about Qualcomm's server CPU, I doubt that it's that much more powerful that Kryo, so it can't compete either. The ARMy, in it's existing form can only compete in IO dominated workloads.
I think you are overplaying the loss of business from Qualcomm to Intel regarding the modem slot but lets imagine, you are right, and QCOM throws in the towel. I don't think the threat to Intel will come from QCOM , but from the likes of HiSilicon (Huawei) and firms like Amazon using their own silicon but again, it would be for IO workloads. Things could change however, if ARM (the company) starts selling high end CPU cores that can get near (ie, to 80%) to Intel's single threaded performance.
and optimized for TSMC 7nm FinFET -
you're a dope head - I love it
Do you have memory issues?:) We talked about this a few days ago. TSMC's 7nm is the 'half step' from 10nm, so using the same back end, just 'better' transistors, no EUV.
The really interesting part of this news release is the focus on performance. The ARMy's server ambitions currently can't address CPU performance dominated workloads. This move by TSMC/ARM is significant. Imagine an Apple style Twister CPU core, high clocked, sold for 10 million a pop...
I don't know much about the ins and outs of chips, and therefore am in no position to evaluate the pros and cons. But I can spot an obnoxious person a mile away. And you're one.
Very honest of you. Just put me on ignore:)
I think you should use your considerable abilities to post Intel news on the ARM board. You seem to be very one sided.
Well, this is Intel related. Better than the politics talk on this board, correct? We might end up having a conversation about tech. I've been quite open regarding my ARM v Intel biases.
“Future ARM technology designed specifically for data centers and network infrastructure and optimized for TSMC 7nm FinFET will enable our mutual customers to scale the industry’s lowest-power architecture across all performance points.”
Perhaps ARM is gearing up for higher performing CPU's.
"Intel has been perfectly honest about the fact that certain technologies it is putting forward are being driven by hyperscalers and cloud builders. The company’s $16.7 billion acquisition is being driven by these customers, who expect to deploy these hardware-software hybrids to accelerate their workloads. Ditto for the Xeon D processor, which Intel launched last March after substantial input from social network juggernaut Facebook."
Interesting read, not sure I agree with the article as the ARMy don't have the compute side yet (only the IO side).
"VELDHOVEN, Netherlands, 24 February 2015 - ASML Holding N.V. (ASML) today confirms that its customer, Taiwan Semiconductor Manufacturing Company Limited (TSMC) has successfully exposed more than 1000 wafers on an NXE:3300B EUV system in a single day, an important step towards insertion of EUV lithography in volume production of semiconductors."
"The NXE:3300B system is one of two such systems currently at TSMC, and will be joined by two additional to-be-shipped new NXE:3350B systems. TSMC has stated that it intends to use EUV in production."
As an owner of ASML I'm sure you knew all this.
Over on anandtech, worth a read.
"At the recent annual SPIE Advanced Lithography conference, Intel, TSMC and other leading semiconductor companies said that significant strides have been made in extreme ultraviolet lithography (EUVL) over the past year or so. Intel did not reveal when exactly it plans to start using EUV technology, but indicated that it will clearly utilize it once it is suitable for high-volume production of semiconductors and is sufficiently cost effective . Samsung and TSMC also intend to start inserting EUVL tools into production in the coming years."
I'm really not sure what you are trying say?
opposite to Intel the foundries are progressing at half node while clowns like A.E. create the perception that Intel lost its process/manufacturing lead -that was the point I made you can go and go and argue with ASML on this - I won't
I am not arguing anything. You are confusing things.
From AE's perspective (mobility), Intel clearly has problem and has lost ground against the fabless ecosystem. Only a fool would argue against this. Where are Intel's 14nm parts? Why is Intel still using TSMC's process for it's modems? Look at the SoC design, just from a CPU perspective alone, Atom V A57, A72, M1, Kryo, Twister, etc etc.
Why do they do it ?
They were probably pressured to keep up with Intel but were not capable to obtain the cost reduction to full amount - hence it's a "half node step" - the term was used by ASML NOT by me
if (shrinking the backend) would be a cake walk the foundries would have done it to obtain the cost reduction
ASML is saying that TSMC and Samsung 14/16nm was really a dog because it's still more or less a die based on 20nm
I'm really not sure what you are trying say? Yes, the foundries have done a half step transition getting from 28nm to 16/14nm, and it looks like they are going to do another half step transition to get to 7nm.
As for why? It's about risk management. The step to 16/14 was too far for the foundries to jump in one go, so they separated out the backend and frontend into these half steps (it's more complicated than this).
Remember the foundries have to bring along the entire ecosystem and this isn't an easy thing...You have to have the tool vendors, IP firms, etc all on board. Design cost is a major hurdle for the fabless firms at these nodes (it's not *just* about cost per transistor!) and you are starting to see TSMC bringing 'new' 16nm nodes online that significantly reduce design costs (for example).
No one is disputing that Intel's process is 'better' or that Intel did a 'fullstep' to get to 14nm, but you are shockingly naive when it comes to the details (ie, the reasons why).
The reality is that there are 2 foundries churning our hundreds of millions of SoC's a year on their 'latest' nodes. Given the comments from the tool vendors and IP firms, it looks like they will transition to 10/7nm quite quickly.
the difference between Intel's 14nm FinFet and TSMC / Samsung 16/14nm FinFet.
AS ASML put it the foundries are progressing @ half node steps - too complicated for A.E.
To be blunt, neither do you. What does "half node steps" mean, and why do the foundries "do it"?
[Yeah, this post of yours is a great example of how "technical" you are these days, marketing boy.]
:) So, is Apple going to drop it's own SoC for an Intel designed one?
[Wow, talk about a mindless non sequitur!!! ]
It's been discussed to death on more technical forums...but we know you run away from the technical.
I agree with that. I'm hoping that BK and Mr. Cook are becoming 'friends'. And that the interactions of the past, are in fact, past.
Even if BK and Cook become best buddies, why would it be in Apples interest to dump it's own SoC for a 3rd party? Why give up all that control and ability to differentiate?
Apple could drop Intel, it works both ways. Extremely unlikely, yes, never happen, I don't know. And it's non-starter, not none-starter.
Thank for your correcting my English:) Lets put it this way, it's much more likely that Apple dumps Intel (macbooks), than Apple dumps Apple (ipdad) :)
You never, ever know for sure. Things change rapidly in this business. For Intel, ARMH, and the rest of the competition. I certainly have no idea what goes on behind the scenes at these companies.
Well, if we're going down the route of "never, ever know for sure", we might as well rehash the idea that Apple is dropping Intel in it's Macbooks for it's own processors:)
Seriously, it's extremely unlikely that Apple would use Intel's App processor SoC's in it's ipads for a whole set of technical reasons. Then you the various business reasons. It's a none starter, certainly for the foreseeable future.
The next question is will AAPL go to 86 processors for their iPads too
Not going to happen.