The second company on that list only has a market cap of 3.5 B.
If Apple wants to join the club, they can do it with pocket change.
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 Reportedly Shifts to In-House Modem Engineering Led by Johny Srouji
- Thread starter MacRumors
- Start date
- Sort by reaction score
The second company on that list only has a market cap of 3.5 B.
If Apple wants to join the club, they can do it with pocket change.
Wait until you've been here for a decade. It's an eye opening experience.
And now I shall go back to nursing my Irish coffee.
"We've been making modems for years. We know this product. These computer guys... they're not just going to walk in here and make a great modem."
Just admit that Intel modems are inferior and make up with Qualcomm so that my Xs will match the performance of my iPhone 6. Really disappointed in range and Personal Hotspot performance(failures) on Verizon when I've got a good to great signal in place that I never had a problem before.
Wasn't it actually AMD who invented x86_64 (aka amd64)? If so, this would be a poor analogy.
Better one would be when Intel design CPUs, they are actually designing them even though they are actually AMD64 cores.
Was there some part of my comment and citation you didn't understand?
You’ve mis-quoted me above, but above I’ve extracted your reply.
No one has forgotten Qualcomm’s SEPs must be licensed. But like I said, they don’t own a huge majority of the necessary patents. They own a fair amount of IP in this space, but so do others, including Nokia, Interdigital, Ericsson, Huawei, Samsung and also Intel themselves.
Of course, Qualcomm must license these other patents in order to make their own modem, just like others must license Qualcomm’s IP.
I’ve posted additional info that might be of interest down thread, at post #92:
https://forums.macrumors.com/posts/27075871/
We are barely reaching 4G LTE sustained max speeds. So what if the 5G is not the fastest. Better be inhouse than rely on other companies. Apple has been i tegrating hw and sw for a while.
Your cite also confirms my understanding.
"But it turns out that the the A4 is a 1GHz custom SoC with a single Cortex A8 core and a PowerVR SGX GPU."
https://arstechnica.com/gadgets/2010/02/meet-the-a4-the-ipads-brain/
I never ever said the SoC wasn't custom. Of course it is. I said that in the A5 (and A4 since you brought it up), the CPU core is an off-the-shelf ARM cpu core.
Is this a joke? Apple uses ARM ISA, their micro architecture is completely custom at this point.
Yes. In fact I designed the first draft of the x86-64 (back then “AMD64”) integer 64-bit instruction set.
Just like A Series processors ?
[doublepost=1549660034][/doublepost]
Look at Intel. I'd say they're the best in the industry (not Apple)
ARM taking 100% of mobile market share, Amazon designing ARM chips for servers, Microsoft working with Qualcom on Windows for ARM, pretty sure Apple has MBP with A series chip in it running Mac OS, I would say future for Intel is bleak, with recent manufacturing issues and loosing the technology edge to other manufacturers make it worst.
Intel is the only vertically integrated semiconductor company. Nobody else has the architecture, ic design, and fab all under one corporate roof. Samsung comes close with custom IC designs and a fab, but they use ARM.
Yes, a combination of ARM, Apple, and TSMC are currently doing very well and their combined future looks bright, but in totality they haven't quite caught up to Intel yet (but likely will soon).
Samsung has held an ARM architecture license just about long as Apple ( 2008 ) has : [ 2013 article so about 2010]
"... It is known that Samsung signed an architecture license with ARM two to three years ago in order to develop its own AP platform. ..."
http://english.etnews.com/news/article.html?id=20130729200006
The Exynos 9810 isn't exactly a simple 'copy and paste" from ARM.
"...At 20.23mm² the Exynos M3 complex is absolutely massive compared to other mobile SoC CPUs. At 3.46mm² for the core and accompanying L2 the Meerkat core is over twice as big as the 1.57mm² of the A75+L2 in the Snapdragon 845, granted that the latter has half the L2 cache. Meerkat indeed almost matches Apple’s Monsoon cores in the A11 which come in at 2.68mm² - but only if one takes into account the L2 cache of the M3 for which I roughly estimate 0.88mm². Apple also has a slight density advantage due to TSMC’s 10FF manufacturing node. ... "
https://www.anandtech.com/show/12520/the-galaxy-s9-review/3
[ NOTE: if Apple , Samsung , and Qualcomm are all implementing something of difference sizes that is significant. ARM has a central clearing house and shared baseline R&D mechanism but individually they are all doing more than just barely substantive work. ]
Whether they are getting 'bang for the buck" out of the extra work is open to debate
https://www.anandtech.com/show/13582/samsung-galaxy-note9-performance-review-snapdragon-vs-exynos
There is also an aspect that goes beyond just hardware. If not putting the right OS kernel support in there are also problems.
https://www.anandtech.com/show/12620/improving-the-exynos-9810-galaxy-s9-part-2
Samsung non match is more so that it is not just one company. The silicon/fab part in separate subsidiary then the other parts. Samsung does a wide range of stuff. Phone . Appliances , Video Panels , etc etc. etc. The parent organization may actually do too much. They run into cooperating with competition ( partners that are competitors of other parts of the business ) and with trying to be everything to everybody ( raising that compete barrier to a large amount and lack of focus ).
Intel has done some stuff that didn't work out so well. MacAfee . Jumping into the cellular modem business buying Infineon ( about the exact same time Apple dropped them as the modem partner and jumped to Qualcomm). Just as they now wrestled Apple back they run into a process node problem .
[doublepost=1549691067][/doublepost]
it is creditable. Especially, if Apple went OCD with the notion of having Intel and Qualcomm modems "look the same" to the end users. (i.e., looking to minimize the differences). I don't think this was a simple as ripping off the specific binary (or source code) so that is was as much a copyright violation. However, if Apple was collecting all the differences of the modems performance in very fine details ( map all the parts where Qualcomm is 'ahead' of the Intel solution). Then Apple throws operative requirements and specs over the wall to Intel then could be some IP capture. Also could be IP that Apple demand Qualcomm do to integrate but didn't directly pay for. Test cases the Qualcomm had developed to work with quirky country X's custom deviations from the standards ( collecting some of those and throwing them "over the wall" to Intel to speed things up. )
Apple has usually puts the phones on tight deadlines so if Intel fell behind at some point someone at Apple could have tip-toed over the line a bit to get things back on track.
I suspect Qualcomm also knew in the back of the mind that Apple was looking over stuff too. Anything they could collect to help Intel stay on track for phone launch could be archive to help Apple's modem spin up go faster too. [ That all of the other major smartphone SoC implementors had intergrated modems wouldn't have been lost of them. Apple is by far the biggest buyer of 'discrete' modems in headsets. There may be a short blip with 5G but that's a long term trend line. ]
There are some decent ones. Especially if put decent antenna design on them. But the problem with Apple is they are always trying to push the thinner and less power consumption on each iteration. So the complexity is extremely high to stay on the upper bleeding edge of most features while also dropping in power consumption. If the celluar (and wireless) standards constantly change every year then it is a constant race that just requires you have been running the race longer ( more experience). But the time new talent can get to the present Qualcomm is gone because they had already pipelined another team working on next couple year's stuff.
[doublepost=1549692501][/doublepost]
Paraphrased Palm CEO on prospects of Apple jumping into the Phone market. Gruber had this tidbit that turned out to be true
"... If he really believes what he’s saying, it’s probably because he has no clue how Apple would approach this market. An Apple phone wouldn’t do more than a Treo or a BlackBerry or a Razr — it would do less, and what it would do, it would do really well. ..."
https://daringfireball.net/2006/11/colligan_head_stuck
Part of what the Palm CEO was missing point on was making a phone that satisfied all the various cell carriers had an "endless" set of requirements that had to be done. Instead of a phone that work mostly everywhere Apple introduced a phone that worked on one and only one cellular carrier. No apps. Managed to toss off vast majority of 'cruft' that the carriers typically made additional custom requirements.
The problem is that in the current context of selling super premium priced phones is that Apple can't show up with a works optimally with 2-3 carriers modem. It is part of the struggle Intel has had is that to replace the complexity that Qualcomm covers that the depth and all the crufty regional/carrier/equipment/etc variances is a quite complicated task. Jumping in with a highly simplified product isn't really an option.
Apple didn't jump into the Phone SoC by brining in application core and GPU cores and a host of other things in-house on day one. Apple eased its way into the pool.
If Apple is now in the mode of "We rich and double screw Qualcomm and screw Intel too ..... our farts don't sink. We'll just do it ourselves. " this could very well turn into a substantive stumble. Apple has money so it wouldn't "tank the company'. It is far easier to take a smaller working system and turn it into an incrementally larger working system than it is to jump straight to the larger system directly. The latter approach typically fails at significant rate. Especially projects that involves a substantive amount of software ( which software modems have).
Apple isn't necessarily going to fail, but they aren't necessarily going to succeed at this if they have the wrong motivations and plan. (e.g, Apple Car. giant buckets of money and not much to show. )
Why stop there. I bet they'll put it in the secure enclave chip and all your phone calls will be encrypted.