Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Set to Shift From Samsung to TSMC for 'A6' Chip Production Next Year?
- Thread starter MacRumors
- Start date
- Sort by reaction score
I assume by the name and the date you are a known troll and you already realize both of your statements are lies...
I don't know too much about the particulars of Samsung's fabs, but it shouldn't be too different to switch to TSMC. Samsung was just a transitional partner as Apple went with them initially and wanted to gradually take ownership of their design as they went along.
Without question, but I think the particulars of ARM designs on an Intel process is going to be something only people with intimate knowledge of both Intel and ARM IPs and processes is going to know for certain. I don't think anyone can challenge the fact that Intel is years ahead of their competitors when it comes to fab process, but I think the industry is marching inexorably towards a new ISA as people's majority use devices switch to handhelds and tablets. Intel can try and stymy the progress with offerings like Atom and Medfield, but I still think it's going to be an ARM world for quite some time. Given that, I think there is some value to Intel opening up their fabs to ARM designers. The handset market is extremely competitive, and with handset manufacturers getting hundreds in subsidies from carriers, they're going to be willing to pay premium price for higher performance SoCs compared to what TSMC can offer them. The central question is how much of a threat Intel is posing to its own bread-and-butter x86 market and how much it's cannibalizing Atom/Medfield etc. sales. I'm going to wager not much as that is the general direction of the market, but I'm interested to see what happens, regardless.
Without question, but I think the particulars of ARM designs on an Intel process is going to be something only people with intimate knowledge of both Intel and ARM IPs and processes is going to know for certain. I don't think anyone can challenge the fact that Intel is years ahead of their competitors when it comes to fab process, but I think the industry is marching inexorably towards a new ISA as people's majority use devices switch to handhelds and tablets. Intel can try and stymy the progress with offerings like Atom and Medfield, but I still think it's going to be an ARM world for quite some time. Given that, I think there is some value to Intel opening up their fabs to ARM designers. The handset market is extremely competitive, and with handset manufacturers getting hundreds in subsidies from carriers, they're going to be willing to pay premium price for higher performance SoCs compared to what TSMC can offer them. The central question is how much of a threat Intel is posing to its own bread-and-butter x86 market and how much it's cannibalizing Atom/Medfield etc. sales. I'm going to wager not much as that is the general direction of the market, but I'm interested to see what happens, regardless.
There's no benefit to it, so nope.
I designed powerpc and ultrasparc microprocessors. One was fabbed by hitachi, the other by Texas Instruments. What's your point? (BTW, I also designed many x86 processors, albeit for AMD, not Intel).
Yes, IBM has nice fabs. Again, what does that have to do with how Intel's fabs are tuned?
This is gibberish. You don't tune a fab for an architecture. You tune a fab so the transistors are either fast and leaky or slow and power-conserving.
More gibberish. It doesn't take years. And none of the high-end ARM suppliers are using hard cores, so it doesn't matter if the "ARM cores are designed for it." You take your soft core and plug it into synopsys with a cell library based on finfets and you're done. But these transistors (i.e. intel's process) are not tuned for the types of devices apple is interested in, so it's moot.
Yep. It's not like there are a bunch of other companies out there as big as Apple waiting to fill their shoes as soon as they jump ship to TSMC. Even if someone does fill their shoes, that's still lost revenue, because both Apple and those new guys would be buying from Samsung if Apple hadn't left. Now is not the time to buy Samsung stock, that much is sure.
I've never heard of any Intel CPUs being overclocked to 5ghz, and the amount of threads that can run on one core is a design thing, so... The only reason Intel has such a gigantic share of the market is because of its very old Wintel partnership. There are better CPUs out there, but not for x86.
Last edited:
What do you mean not much different from Samsung? I think Apples only reason for the move is because Samsung is a supplier of components as well as manufacturer of end products. If they were only the former like TSMC then Apple wouldn't have a problem with Samsung.
The only way for Samsung to keep Apple is to stop producing smartphones. But I don't know what their profit is from their smartphones. If its significant or they see a potential then they would say bye bye to apple. If not then they would stop producing phones.
If you meant me by Android fanatic, far from it. Apple since 1984 , never looked back. Just not Apple clouded enough to forget about checking reality.
.......but Android is more widely spread because of all the promos etc.
As for the old chicken or egg discussion, everybody copies from everybody IMO. Call it : Improving on existing knowledge/design etc.
That's how humans progressed and inventions were made. You see an idea that gives you a better idea etc.etc.
Only today lawyers get involved more and more.
Agree with you that most consumers are not checking things out.
$ 99 with a free second phone is enticing and doesn't require too much thought if it's what you were looking for..
Clock speed is not everything.
Just look at intel for example look at how their clock speed dropped like a rock compared to the past. They were pushing 4 ghz at one point and my 1.6 ghz computer now would beat the crap out of it.
Or you can look at for example the AMD64 3000+ that at the time was beating the crap out of a intel P4 running at over 3.2 ghz.
Umm so instead of 1 company replacing Apple you have 3 or 4. The semi conductors are in short supply right now and generally there is a shortage. Apple is very replaceable in the market. I would not be surpised in the least that it was more Samsung refused to go as low of a price as Apple demanded and said their is the door. They will have no issue replacing Apple and chances are for higher profit.
Google it. THere are a few out there. I know they did it back in like 05 06 and since then have pushed it higher.
I respect that you designed microprocessors, but it doesn't take an engineer to imagine a MacPad type device designed around an A(x) processor.
I bet that in the next few years there'll be an ARM based Apple product with the word "Mac" in it.
Pick a cutoff date, we'll both add it to our iCal, and we'll see who's right.
Last edited:
This article also says that Apple was down a point too, so how did the Verizon iPhone take away from Android sales when Apple lost a percentage point too?
Everyone else except MacRumors has been talking about it.
Apple's probably already mapping out A7, A8, and A9.
I actually doubt they're doing any serious work on A8 or A9.
A6 is likely a 28nm Cortex A9 variant if not a Cortex A15 (eagle) variant and is being designed right now. It can't even tape out until TSMC's 28nm fabs are turned on and mature. A7 is likely a 28nm Cortex A15 chip in the initial planning stages/early development. They may be having initial discussions about the A8, but they don't have any technology to map it to (i.e. they may not even have a process library yet). A9 I doubt is even being talked about yet.
This is all assuming the TSMC shift does happen and Apple keeps a yearly Ax processor release schedule.
Samsung is more than cell phone parts:
The Samsung Group is a multinational conglomerate corporation headquartered in Samsung Town, Seoul, South Korea. The Samsung Group comprises numerous international affiliated businesses, most of them united under the Samsung brand including Samsung Electronics, the world's largest technology company by sales;[3][4] Samsung Heavy Industries, the world's second largest shipbuilder;[5] Samsung Engineering was ranked 35th, Samsung C&T 72nd in a 2009 ranking of 225 global construction firms compiled by the Engineering News-Record, a U.S. construction journal.[6] Samsung Life Insurance was ranked 14th in a 2009 ranking of Fortune Global 500 Industries.[7] Samsung Everland, South Korea's first theme park opened in 1976 as Yongin Farmland. In 2002 it was the fifth most popular theme park in the world, beating out Epcot, Disney MGM and Disney's Animal Kingdom.[8] Cheil Worldwide operates as a subsidiary of Samsung Group[9] and was ranked #19 among the "World's Top 50 Agency Companies" by revenue in 2010.[10] Shilla Hotel, a subsidiary of the Samsung Group, has been ranked #58 among the "2009 World's Best Top 100 Hotels" in the annual reader survey conducted by the prestigious international business magazine, Institutional Investor.[11]
They may or may not do it, but it's not a good idea. If it's a "Mac" it presumably runs Mac OS X with fat binary support for ARM. If it has rosetta-type support for x86 binaries (even if such thing is possible given copyright/patent issues surrounding x86), applications run that way will be slow and battery-sucking.
As for native ARM apps, keep in mind there really isn't anything inherently "better" about ARM. x86 adds a little more die area and a couple pipe stages in order to support addressing modes and complex instructions that make compiled software execute very efficiently. ARM leaves out complex addressing modes and instructions to shrink the die and leakage current, at the cost of less performance efficiency from compiled software.
MOST of the perceived differences between ARM and x86 (e.g. low power) have a lot more to do with design than architecture (for example use of static rather than dynamic logic, types of flops used, tuning of process, clocking schemes, etc.) These power improvements do not come for free - they cost performance.
I can believe there may be an "iBook" someday running a future iOS that looks a lot like Mac OS X and runs on ARM, but it just doesn't make a lot of technical sense to run Mac OS X on ARM - an ARM that runs fast enough to make it practical won't be all that much more power efficient than a properly tuned and designed x86.
In my opinion, what shows Samsung's size best is this excerpt.
"Samsung Group accounts for about a fifth of South Korea's total exports.[14] In many domestic industries, Samsung Group is the sole monopoly dominating a single market[citation needed], its revenue as large as some countries' total GDP. In 2006, Samsung Group would have been the 35th largest economy in the world if ranked, larger than that of Argentina.[15]."
They represent over 12% of South Korea's 1.423 trillion GDP.
Having been to Seoul, Samsung's scope is truly tremendous. They are everywhere. I saw a ton of apartment towers with their logo on it. They seem to own or have sold everything in Korea.
Based on the fact that some of their higher ups have been charged with corruption, I'm sure they own things in many facets of the word.
Of course, but Samsung's businesses continue to expand, including their smartphone share in the US. That's only going to further when the Galaxy S2 is released in the US on every carrier, which is one of the best phones out there without question.
Woah, so it looks like we agree, and this explanation is a long way from "nope vs. yup", thanks.
Before thinking about the technical elements of ARM vs. x86, I thought about the precedents at play. First, at the 2005 WWDC, Steve Jobs proudly demonstrated that OS X had been, in secret, developed for both PowerPC and x86 and that the company would quickly transition to using that architecture for all of its products. Second, Lion appears to be the beginning of a trend towards unifying the mobile and desktop OS X experience as much as the differences in screen size and input method would allow.
While I don't think there'll be an all out transition from x86 to ARM, I think that unless Intel does relatively better than IBM did in getting more performance per watt, in the medium term, especially with Apple pouring development dollars into iOS and the in-house development of the A series, an ARM powered device with a keyboard and a MacBook Air type form factor makes some sense. It also dovetails with Apple's MO of having more and more control over the devices by creating a reasonable premise for locking this hypothetical device down- it's more iOS than OS X.
Also, this could light a fire under Intel to enhance Atom (or to scrap it and build something altogether better) and to lower the already preferential pricing that Apple receives.
This is of course all speculation from someone who knows much more about Apple's product line than the x86 or ARM instruction pipelines, so, thanks for reading.
There's no question that ARM processors are going to start rivaling desktop processors, at the very least Atom processors, and do so with less power consumption.
http://www.itproportal.com/2011/03/14/exclusive-arm-cortex-a15-40-cent-faster-cortex-a9/
Wall street is a combination of real money and BS!
If wall street was told to trade its current value into cash, it couldn't.