My wife is out for a run right now with a friend. She may stop and talk to people along the way, but the one thing I can tell you for certain is that it's 9:01 AM West Coast time, and I've been up for four hours now and still no coffee! She likes to have fresh, hot coffee waiting for her when she gets home. If there is a device that would clamp onto the heal of her shoe or somewhere on her clothing that would allow the coffee pot to go off when she is within five minutes of home you'd bet she'd own it. Better yet, if she stops off at Starbucks it would cancel her request to make coffee.
You folks may think we're too connected and dependent on technology. I don't think we are anywhere near saturation. Just look how excited about apps people get. Now imagine if the apps had the power to control real things in your life like the lighting in your home.
I read the article by Rocco Pendola this morning, and he's convinced that wearables have no future. If they remotely do have a future, they won't come from Intel.
Consumers have such a profound lust for electronic consumer gadgets that currently isn't being met. The iPad proved that people will buy something and try to think of a use for it. But sooner or later the batteries get old and the technology just gets old which means that it's time for a replacement. How exciting is buying the next tablet, or smartphones when the novelty has worn off?
Consumers need to spend hard earned money on gadget and gizmos to provide a distraction in their lives. Something to admire, discuss, and figure out how it works. The IoT isn't one product or idea, and it's far larger than wearables.
Rocco Pendola should stick to writing about commodities.
The newest ARMH CEO was talking about this issue the other day, and why ARMH had no big announcements at the mobile conference. The big sales opportunity was in ultra cheap phones that are selling in third world countries. They are replacing feature phones. It's a nice model, ARMH gets paid and the foundries fight it out for each order against other foundries.
How will they be pushing the high end phones when they admit that this isn't where the growth is? How can they push new nodes if their isn't a client base to support it?
Oh come on now, Daniel Nenni promised us that 20nm was coming out faster than anticipated. Certainly this is some huge misunderstanding that Daniel will get down to the bottom of fairly quickly.
I've heard many people theorize that Apple really hates Intel and can't wait to get rid of them as a supplier. Intel has been a supplier for Apple since 2006, together they have worked on Thunderbolt and Thunderbolt 2. The newest version is one of the few solutions that can handle the throughput needed for 4K video editing. Tim Cook also seemed thrilled to announce the new Macbook Air's with Haswell processors.
Fab 42 is now empty, the machines that were supposed to fill it have been cancelled. If Apple were to give the green light for 10nm joint project processing how long would it take to fill the factory with equipment, validate it, and start the tape out process? A year or two? Perhaps with the iPhone 6s that would be a reality, but for now I don't see it. Which does beg the question, could Intel fill at least part of a huge new fab without word getting out on the streets?
"*ARM is going to be using 28nm for a long, long time" Part of the dynamic of the ARM economic model is that there are always consumers demanding high performance processors from ARM and are willing to pay the price. What most of us failed to see was that there would be a day of consumer indifference toward tablets and smartphones where their overall cash outlay was of primary concern. Foundries like TSMC now have to fight it's war on two battlegrounds. In order to drive the 20nm node they have to fight for business against Intel who will be pushing 14nm in x86 and fabbing 14nm ARM for non-competitors. Intel's first customer came at the expense of TSMC, more to follow. On the other end of the market they have to fight Mediatek. and GloFo for 28nm business. Expect pricing based on capacity. Those with the least amount of business will drop the price lower and lower to secure orders. It looks serious that Mediatek is out to secure orders against TSMC.
The one thing that Ashraf could do to help Daniel Nenni is explain some basic financial concepts. He keeps referring to crowd sourcing in the wrong context. As long as he's at it he can talk to him about the law of supply and demand, inferior goods, and the law of diminishing returns.
" High end will go to 20nm and then 16nm after. Please don't try to mislead." When? What realistic timeframes are you talking about? Isn't this premium dollar segment going to be exactly where Intel show up in the market? How will the numbers work for ARM?
I am shopping for a good CS school right now for my daughter. San Jose State University is very well respected in Silicon Valley. A BS in Computer Science from San Jose State is actually something very worthy. They do have an excellent program. Being a National Merit Scholar (Ashraf) is something most people don't get. It's very prestigious. I am a graduate of a California State University. My family was nearly bankrupt after sending my brother to a private college and thus I went to the local State University literally down the street. Now I'm looking at private colleges that are offering generous scholarships to my daughter and yet she'll rack up something like $160K of debt over four years. Or there is the state university route. So yes, I do know where Ashraf is coming from.
I also know where Wallis is coming from. Nenni comes off a bit smug and talks down to those who aren't in Silicone Valley. I know someone just like him, constantly posting on Facebook where he's eating dinner with his so called beautiful wife like we're supposed to envy his lifestyle. After a while you just learn to ignore it.
Two parts to my answer, the first is advertising. Google will be able to make it's ad budget go further fro the Nexus 8 with Intel. Then Amazon may jump on board as well. The second part comes from the article that got Ashraf excited in the first place.
"ARM expects many early Cortex A17 designs to be built on a 28nm process, with an eventual shift over to 20nm once the cost of that process drops."
Who is going to fund the 20nm nodes a fair market value when Intel is the market at 14nm.? Though this won't stop ARM from eventually going to 20nm planar, it's certainly going to slow it down, extending Intel's lead. In 2015 if you're happy with a 28nm ARM processor, then more power to you. But if you want the best technology in the market that you can realistically get your hands on, it's not going to be at 28nm.
Alright Ashraf, explain to us the future growth trajectory of ARMH that makes it worth a P/E of 100? You said yourself as the sales growth curve becomes more predictable the P/E will fall. I went to the Apple Store last night and actually played with the iPhone 5S. It was very nice. It was a bit more zippy, I did notice that . I don't think the average consumer cares. From best to worst in ARM the differences aren't big enough to make a difference in the minds of most consumers. Even with a so called 60% performance increase (which I feel is hyperbole) it's a tad bit faster version of been there and done that. These new ARM processors will once again fails to ignite consumer demand for higher end phones as ARM model creates too much hungry competition that turns it's products into a price driven market.
How is ARMH going to spread the news on out of order processing? Just for haha's ask the next non-techy person you see with an iPhone 5S about the processor in their phone. They will have NO CLUE about ARM technology. Ashraf is once again posting like the fastest processor wins. There are far more factors than speed. And for the record I want to see proof that there is a 60% performance increase from ARM with independent tests. I smell a load of BS here.
I read this article. It was about as Intel friendly as Nenni can stand. It casts Intel as a large, slowly failing company and the unopened FAB as a testament of it's failure. The new Quark is something far off and obscure that may ultimately save Intel.
Aside from the positioning of Intel I'm curious to see where this all goes. Can Daniel actually let a series of article that points to the future where Intel has the upperhand go on his website? This contradicts the contrarian POV of Intel he's been offering on his website. I smell an ambush being set up.
Our friend Daniel Nenni mentioned that Intel would be better off buying a foundry instead of opening it's own to get into the foundry business. I kinda have the feeling that TSMC is being positioned for sale. Intel doesn't have the money or desire to take on one of the larger ARM fabs at this current time. Intel still has the unopened building for a FAB in Arizona that would most likely be a more logical choice to populate than to take over a large and functioning FAB in Taiwan. Besides, at the current time the books look to good to touch. Perhaps in a few years with Intel beating it from the top end and GloFo eating it from the bottom it will be more value priced.
I don't think that TSMC will sell in the current market. But there will be consolidation as Intel dumps more capacity into the foundry mix which is currently seeing softening demand and price erosion. My question is, which fab will be first to call sell out? Or will there be a good old fashion shutdown with the FAB declaring bankruptcy?
"Complete nonsense. Intel will do fine on 14nm but not in foundry or mobile."
And why do you say this? How can you tell? What do you have to backup these claims?
AMD is pushing a new filter in Photoshop as a reason to build a system with AMD. Photoshop really doesn't use a lot of GPU resources. It's not big on parallel processing though it's seeming to add more with each version. When you do RAW file conversion for example it will beat up your processor for however many images you're processing. Sometimes the system is running at full tilt for an hour. I really don't know of many applications that do this. I've noticed that the Ivy Bridge i5 processor doesn't quiet go up to using the full processor, it hangs around 90%. The temperature in the processor is around 50C.
The HD4000 graphics built into the Ivy Bridge i5 and the HD4400 built into the Haswell i5 are both very good. I'd advise running a new system build without a discrete graphics card. If you truly need a discete video card it will be rather apparent. But to get one better than the surpasses the performance of the onboard video of the AMD Kaveri isn't an expensive proposition. The thing about buying Intel is that you can get yourself a lower end i5, then wait a couple of years and pick up n i7 when they get cleared out. I see no advantage buying AMD. If graphics are important to you, you're much better off buying a discrete video card.
This is the new processor AMD is challenging Intel with in it's press releases. The two areas AMD touts it strengths are PC gaming and Photoshop. AMD is promoting its strengths of a respectable CPU working in conjunction to onboard GPU. Hardware(info) did a very extensive review on the new Kavari. Here's what they had to say.
"The downside of new technology is that very little compatible software currently exists, at least not mainstream software.The performance gain is therefore largely a theoretical one, "
"the strong point of AMD's new APU is its graphics card. Steamroller CPU cores may be faster than the Piledriver cores, but their lower clock frequency largely undo that performance gain."
Say it ain't so.
"AMD compares its new processor to the Intel Core i5 4670K (even if the Core i5 4570 is a fairer match), and if you want to assemble an affordable PC without dedicated graphics card, then the Kaveri is indeed the best choice If you plan on buying a separate graphics card, we would recommend the Core i5.
The price on Newegg The AMD is on sale for $185, it has 33 reviews and four stars. The i5 4570 has 63 reviews, 5 stars and is on sale for $189, The 4670K is not onl sale, it's $239 and has 317 reviews all five stars.
I think the numbers speak for themselves.
They don't let me out of the house much anymore, but I may just have to go to the industry meetings you talk about. Where do they hold the TSMC annual meeting, Taiwan? My Cantonese isn't so great. Will your book be available in Simplified Chinese?
I'll be looking forward to meeting the other paid bloggers from TSMC. By May all those large scale 20nm planar orders should be in hand and the FAB should be shooting out high yields of 20nm without any problems. I want to hear all about that Because i would hate to be at at the meeting when things aren't going so well for the fabless model. I don't want to hear about delays, problems, and low yields. Actually, I take that back. i really want to hear about these things. I would really enjoy seeing the fear in the room when you finally realize that Intel has a process lead and the fabless model never catches up. That will be the last chapter in the book, next revision, right?
BTW, the first 50 pages were up the other night. I got a look at it. There was a story about the first fabless customer.
Oh my gosh, now I'm really laughing. The AMD demo was based on AMD versus Intel, with an optimized sharpening program written for AMD GPU's. What they essentially did was challenge Intel to a footrace with the condition that AMD could use a bike, and by the way, the course is all downhill.
These sharpening filters are kinda a dime a dozen. I've seen ones that work great when written specifically for Nvidia GPU's. Do you really think people are going to drop their Intel based PC's for a second rate AMD computer when they can do essentially the same thing with one of many after market filters?
The Adobe Creative Cloud is subscription based software. They don't really want you to buy a copy of Photoshop, but rent it for a monthly fee. Adobe hasn't improved Photoshop much since version CS 4 and all these worthless little add ons aren't worth the price of buying a subscription. Looking at the stock price the last last year the real winners are the Adobe investors. For now. There are competitors in the wings making vast improvements in their photo editing software. I'm keeping my eyes on GIMP. In another year to 18 months they'll have a pretty good CS4 knockoff.
AMD made a press release about a freaking Photoshop filter! Wake up and smell the desperation.
The lights are out at FAB 42 for now because the yields on 14nm are much better than they expected. This means that all that talk of Intel not making 65% margin on mobile are correct, it will probably be more!
Here's my new formula, I call it the IFB constant
That means the rate of the fabless model destruction is equal to the amount of PR ARM is paying for.
If Kofi and Daniel need a play to stay once their high paid PR position with ARMH, TSMC, and NVDA ends, they can stay with me. I've got room in the garage, next to the rats. They should feel right at home.