This chart says they have Fab in six locations three of which are in the USA.
http://www.intel.com/content/www/us/en/architecture-and-technology/global-manufacturing.html
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.
Intel and Apple Already in Talks Over ARM-Based Chips for Future iOS Devices
- Thread starter MacRumors
- Start date
- Sort by reaction score
http://www.intel.com/content/www/us/en/architecture-and-technology/global-manufacturing.html
So here's what I see from that article:
In January 2006, Intel's newly released Core architecture (Yonah) completely kills all previous Intel and AMD architectures. It's magnificent.
Performance/Watt the Core Solo is almost as good as the two year old 7447/A G4 that was released in April 2004.
The 1.5 GHz G4 consumes ~25 W and is fabbed on a 130 nm process.
The 2 GHz Core Solo consumes ~30 W and is fabbed on a 65 nm process.
By 2005, Freescale had a 90 nm 7448 G4 that consumed ~30 W at 2 GHz that Apple chose not to use, but you could get if you bought processors upgrades from shops like Sonnet or NewerTech.
In mid 2006 Intel went 64-bit with the Core 2 Duo (Merom). An excellent family of devices.
By late 2006, Freescale had a 32-bit 90 nm dual-core MPC8641D that consumed ~30 W at 1.7 GHz. Again, that Apple chose not to use.
In early 2007, P.A. Semi released the 65 nm PA6T at 2 GHz, a 64-bit dual core. A SoC (CPU+Northbridge) that consumed around 25 W and had a performance/Hz that far surpassed the G4 that at this time was still better than the Core 2 that ran at 35W and that's not the Northbridge included.
Intel had better fabs, best in class to this day.
Windows didn't run on PowerPC, a great differentiator.
There wasn't a great variety of PowerPC processors to chose from.
I'd say that Apple would've been in OK shape performance wise if they hadn't jump ship. The G4/e600 core was an _excellent_ CPU core, best of class, years ahead in performance/w and performance/Hz and is probably still competitive in this regard still easily beating most ARM devices too, especially in FP and SIMD. And the e600 evolved into the e5500 and e6400 cores improving this. Now-days (since 2012 actually) you can get 12 core PowerPC SoCs consuming less than 30 W (without GU though). Apple had a PowerPC license, they bought P.A. Semi and could've kept on developing PowerPC CPUs, perhaps licensing a GPU from nVidia or PowerVR and integrated that. They could've done this on their own without Freescale and IBM that didn't really want to make the CPUs Apple needed.
And they used P.A. Semi to concentrate on ARM, witch was an excellent move.
And while we're at it. This new thing Intel's doing, opening up their fabs to other designs.. they don't need an ARM license to make Apple's ARM chips so they could've done this years ago, but I think they pitched ultra low power x86 for Apple's iDevices all those years and missed the boat.
Package deal for sure and if Apple plays it right, priority access to next gen chips for Mac's.
On your link, Fab is in blue. Assy & Test is green and is overseas. Look at the Wiki list (https://goo.gl/HSj6Dt) and that shows that most wafer fabrication, which I was referring to, are in the US and only one, Ireland, is doing fabrication of advance technology (14nm and below). All semiconductor companies do Assy & Test overseas with only limited, specialized facilities for this function in the US.
Agree Fab is in Blue but unless you are seeing something different to me I see six sites in blue including Ireland, israel and
Dalian.
Go to my link on the wiki and you will see that Intel currently has 12 Wafer Fab facilities. 9 of the 12 are in the US. Of the 3 that are outside of the US, only 1, that in Ireland, has the capacity for advance wafer technology (14nm and below) which would be required for an XScale CPU. That is my point.
Mac moving to Arm. I wrote it at the time. This move would not have been by coincidence Apple & Intel would clearly be in lock step in the background.
Closer and closer we get to ARM based macs. This is the pipeline Tim is talking about. I'm calling it again!
Closer and closer we get to ARM based macs. This is the pipeline Tim is talking about. I'm calling it again!
I hope it wont be the sane scenario with intel as on mac on iphone aswell, with intel delateing strategy...
You might be interested to know that ARM which has always been a British company based in Cambridge England has just been purchased by a Chinese company. The deal will be finalised in the second week of September. As a stock holder of ARM I am delighted because I made out like a bandit. However, you have to wonder what the future holds for the technology. They claim they will keep the Research in the UK and provide additional resources. However, you have to wonder how the company and its technology will be viewed in the future.Mac moving to Arm. I wrote it at the time. This move would not have been by coincidence Apple & Intel would clearly be in lock step in the background.
Closer and closer we get to ARM based macs. This is the pipeline Tim is talking about. I'm calling it again!
It's inevitable.
I think a variation of Macbook Air and the Mac Mini will be the first to switch.
Dalian Fab is considered to be memory-only fab for 3D NAND according to the trades. Only Intel truly knows what the long term plan is, but if they keep the current footprint, the ARM chips would be Made in AZ or OR, outside chance Ireland. I think that Israel is not uP or advanced node based on Intel's press releases. Assembly and Test is 99.99% Asia for all integrated circuits (Micron, Boise might be the last remaining high volume assembly and test in USA).
I know x86-64 =/= ARM but again time might come mobile devices might be capable of running full desktop apps in the future.
The lack of progress was because there was no interest by either Motorola or IBM to cater for that market without having to do a ground up redesign of the POWER processor architecture or major surgery to their PowerPC 7XX series to not only make it more efficient but also upgrade it to 64bitness so then Apple would have uniformity when they eventually transition to 'all 64bit'. End of the day Apple was a minor customer who was more of a PITA for those two companies than what it was worth then combine that with the industry movement towards x86 with Playstation and Xbox not to mention the failure of alternative architectures to take off, I think that Apple made the best decision to move away from PowerPC.
The big question now is whether Apple see that long term they can re-use the same basic core architecture of their A series processor and then scale it up to move beyond portable devices to eventually replace x86 which would enable them a lot more control then combine that with maybe keeping AMD around for discrete GPU's and PowerVR for their portable devices but I think long term I could imagine Apple buying out PowerVR then bring it in house then along with the rumours of their own Wifi chip that eventually Mac and iOS devices will share a common hardware core that is scaled up according to need and being able to make those choices based on their on schedule rather than on Intel's (which explains a lot of the focus recently by Apple regarding USB 3.1 etc where as thunderbolt got a lot of attention before). Interesting times a head but then again this could all go up in a pile of smoke if Apple ultimately decides that the benefits of x86 compatibility far outweigh and possible benefits from uniformity of architecture across the product line.
I assume that eventually that will be the place where XPoint will be tabbed along side their Micron partners. It'll be interesting to see what Intel does because although Apple using Intel capacity for fabrication enables Intel to fully utilise their fabrication investment the margins are very low which makes me wonder whether the obtaining of a ARM licence is actually pointing to something bigger in the long run given that Intel really needs to have a plan B when one considers the shrinking shipments of PC's combined with their exiting of the mobile handset market etc. whether long term they're looking at reusing their architecture designs then retrofitting the ARM ISA on top whilst discarding the remnants of x86 within the silicon that are no longer needed.
I also wonder how much of the x86 itself is slowing down Intel's ability to shrink the die and scale up manufacturing but then again that would require insider knowledge.
Last edited:
Just speaking to this part of the response. I did IC design here in Phoenix for some 30 years and worked for every chip company in the valley with the exception of Intel. That being said, physics and tools work the same on the east side of Phoenix as they do on the south. With today's integration levels, I doubt very seriously that the x86 architecture has squat to do with it's ability to be shrunk or not. Way back in the day when chips were truly a planer horizontal process and advanced technology meant you did poly plus 3 layers of metal ..... yea your architecture would impact +/- a core's integration function. Today's 14nm fin-FET vertical tech that is using upwards of 12 layers of metal interconnect.... gates are truly free. Yea... you still have a packing density factor and you will do very tight custom packing on memory and the main ALU and highly structured arrays but the rest, which now is the lions share of the chip is as tight as it can be. It would be interesting to see a side by side analysis of performance/area of an ARM vs X86 CPU. Not sure if anyone has ever done that. I'm guessing I'm sure they have within the wall of the I-Borg but it's probably never been published.
Speaking of ARM vs X86 I'm curious. I'm not an architect person at all and I've not looked into the details of CPU's in ages, but do the newer 64bit ARM cores have FPU's in them? Way back in the day when I worked with ARM (I was at VLSI Technology at the time. An ASIC company and ARM licensed semiconductor manufacturer) ARM cores were very efficient but strictly integer machines. They had a SANE module for their compiler but the hardware was strictly integer. I ask only on the speculation of a ARM based Mac. I can't see that happening in the absence of a FPU in the ARM core.
Last edited:
From what it appears, it is an optional component but I'd hazard to guess that the vast majority include an FPU with niche scenarios leaving it out for I'm sure a technical reason: https://www.arm.com/products/processors/technologies/vector-floating-point.php
I could see Apple easily scaling up an ARM SoC design they have but the big question is whether the pain is worthy the long term gain or whether there is ultimately any gain at the end of all of it other than a smidgen more power savings and development only made slighter easier. That being said, Apple is putting a heck of a lot of work into LLVM/Clang ARM optimisation which makes me wonder whether part of that is about ensuring that LLVM/Clang is mature enough so if there is a transition in the future there will be the ability to recompile and ago without having a major performance hit because the compiler toolchain is immature.
I was aware of this claim. However it was a Japanese company, but I couldn't' see the links to China made clearly. However I stand to be corrected. It just wasn't clear at the time when the claim was made that it was a front for Chinese company.
Last edited:
Chipgate wasn't Apple's fault. I blame all the samsung chip owners being in denial about their inferior chip's.
It wasn't the TSMC owners fault we got the better chip.
There will be arm based macs. The regular macs will use arm, the pro models will use x-86.
The iPad Pro is the first step.
So iMac and iMac pro(no more macpro ?)
MacBook and MacBook Pro
And perhaps a Mac mini, Mac mini pro.
In price range from lowest to highest price:
IPad Pro 9.7 and 12.9 just running iOS . Without smart/keyboard cover
MacBook with arm based macOS without touchscreen support, no iOS mode.
IPad Pro with keyboard cover, etc, and will be able to run macOS (when keyboard /smartcover is connected) , and iOS .
MacBook Pro for the professional user, which requires dedicated software and hw.
The iPad Pro is the first step.
So iMac and iMac pro(no more macpro ?)
MacBook and MacBook Pro
And perhaps a Mac mini, Mac mini pro.
In price range from lowest to highest price:
IPad Pro 9.7 and 12.9 just running iOS . Without smart/keyboard cover
MacBook with arm based macOS without touchscreen support, no iOS mode.
IPad Pro with keyboard cover, etc, and will be able to run macOS (when keyboard /smartcover is connected) , and iOS .
MacBook Pro for the professional user, which requires dedicated software and hw.
Last edited:
Not quite. Intel is already slated to be in the next iPhone
http://tinyurl.com/z5nxul2
http://tinyurl.com/j8uhh35
Granted it's not the CPU but it's the first step.
I agree, that's sums up my recollection as well. The switch to Intel has also not been all rainbows and unicorns as anyone waiting for new Macs can testify to: Intel's "Tic Tock" schedule has landed with a bit of a thud.
Ironic, given this discussion is about iPhones, that we are recalling the days when people would look at straight processor specs and assume the faster chip was, well, actually faster given recent speed comparisons between iPhones and Android devices.
I am pretty positive SoftBank is a Japanese company. I don't think a Chinese buyer would have gotten the takeover approved by either the EU or the US, on national security grounds.
[doublepost=1472424299][/doublepost]
I wonder how Intel has managed to keep a large portion of their manufacturing in a high cost location like the US. Given the way the rest of the US manufacturing sector has gone - I wonder how Wall Street has let Intel get away with this, almost a breach of fiduciary duty. Maybe because they just can't ship their tooling to China with all the export control restrictions.
Last edited: