Saturday, September 14, 2013

Apple's Profit Addiction Leaves Little Hope for Future

Well, we've all had to time to absorb Apple's latest "event" and the necessary time for the famous "unreality field" to wear off. (Did it seem to you that the field's effect was weaker than it ever has been before?)

Frankly, the news is grim. Apple has chosen profit margins over building solid products that "just work". Apple has always had this odd notion that "good products" and "expensive products" go hand in hand. This is largely true when products are new; when infrastructure and supply chains barely exist. But once "consumerization" of a product occurs, trade offs have to be made, different levels of products have to be aimed at different groups. Apple seems unable to make reasonable trade offs, or live with reasonable profit margins.

But you say... Apple launched TWO phones. (Wow, 2 -- with the same "perfect" size of screen... I bet Samsung's scared.)

Well, the "lesser" iPhone 5C costs $733 in China and $549 in the US (without contract). It has a gross profit margin of around 40%, compared with Apple's GPM of 35.6% last quarter. The iPhone 5C is simply a ploy to continue Apple's absurd profit margins, and we've already seen what that does:

In case it is not clear to you. An iPhone 5C is, quite simply, an iPhone 5 in a colored plastic case. The plastic case greatly increases Apple's margins and I'm sure they hope it will drive some buyers to the iPhone 5S.

The iPhone 5S

Now, if Apple wants to become the Ferrari of smartphones, fine by me. But as that would mean significantly fewer potential buyers, Apple shareholders would indeed not like that. Sure enough, after the launch event, Apple's stock has dropped significantly and its outlook has been downgraded by a number of investment analysts.

My problem with the iPhone 5S is they are not even doing a good job making a "Ferrari" phone:

Here's some overviews of the problems with the iPhone 5S:

Five disappointing things about the iPhone 5S - CNET Reviews

Google Laughs at the New iPhones - TheStreet

The Joy of Tech comic... Reaction to Apple's MehPhones

Now let's consider the "innovative" features of the iPhone 5S, one by one:

64-bit Processor

The 64-bit processor in a phone at this point is so silly, I considered putting this Apple Insider article defending 64-bit usage in my Strange, Possibly Humorous News. 64-bits becomes important only when you need to have more that 4Gb of memory. As the iPhone 5 had only 1Gb of memory (one assumes the iPhone 5S will have 2Gb), the only significant effect that the 64-bit chip will have is to waste a great deal of that extra 1Gb.

In fact the "2x" speed up that Apple is bragging about is likely from adding two more cores (to make the phone quad-core). As most high end Android phones are already quad-core, you may hear much laughter when this is confirmed. (Apple is annoying in that they never really tell you what's inside their phones... one has to wait until they are released and various engineers around the world tear them apart and report back.) If this is confirmed, then Apple's new processor will in fact be significantly slower than what is in many high-end Android phones. Apple routinely does this sort of misdirection in their launch events.

iOS 7

Well, it's old news. It doesn't really catch up with the current Android (and a new version of Android launches next month). I believe that when this is inflicted on users on September 18th, there will be much screaming. User's are used to upgrading to a new iOS release as soon as it's available. This new release changes (often gratuitously) "everything" and most reviewers do not believe it changes things in a positive way. To understand the screaming level, imagine if Microsoft launched Windows 8 by sending out a Windows Update that pretty much every user installed as it released. Now realize that most members of the mainstream press are iPhone users and will upgrade as soon as it comes in.  Am I amiss in thinking one should wear hearing protection on launch day? Another comparison... this will make the Apple Maps fiasco look small. Google will be unable to bail them out of this new mess, like they did with Maps.

iOS 7 is also putting a strain on developers (one survey showed most were making new app versions be iOS 7 only, meaning older customers may have lost support and that Apple made it difficult to maintain support for earlier OS versions). One developer is in full rebellion:

Fingerprint Sensor

Ah, the finger print sensor. It really is pretty much the old technology from laptops. It will fail big, even Apple is hedging about its quality:

Apple Insider refers to the problems as those of "cutting-edge" technology, but the system doesn't seem to be much better than when it was used on laptops in 2007.

The amount of traffic on this particular "feature" is huge... much of it serious:

Oh, but the crazies are there as well:

The finger print sensor seem also to be causing production problems:

M7 Motion Processor

This is basically a batter saver. Allows monitoring of certain sensor with less power, much like the special processors for the always on speech recognition in the Moto X. It's not a bad thing, but alas, as the iPhone 5S does not increase battery life (something the iPhone needs badly)... it is an essential thing as battery life would have decreased otherwise. A specialized processor like this is not a new idea. In fact, most mobile ARM SoCs have specialized processors for things like video playback.


Again, it's not a bad thing, but it's the sort of thing you would expect on any higher-end smartphone.
Think Volvo, not Ferarri. Here's a good review of the camera:


I'm sorry, but this set of new things simply doesn't constitute "innovation" and certainly does not justify Apple's premium pricing. I'm far from alone on this (these stories are all AFTER the iPhone 5S launch):

Apple's innovation problem is real

Why Apple needs to innovate faster

Apple is no longer an innovative company, says the man who helped Steve Jobs design the Mac

To add insult to injury, it appears that Apple's days of protected pricing has ended:

There is little reason to believe that coming iPad and laptop announcements will prove any more innovative or wise (with respect to pricing and positioning). So expect Apple's market share to continue to drop. It's the 1990s all over again, except this time Steve Jobs isn't available for a come back.


  1. 64-bit might be useful long before 4gb:

    1. The issues they are discussing are Intel only... it does not apply to ARM.

    2. No, they mention embedded systems as well. Mapping scheme is important with any hardware. Intel is just an example they chose to dig.

    3. Could you point to a specific post or explain what you are talking about. The numerous posts in the thread you were talking about referred to an oddity of Intel architecture and about high-speed I/O... not exactly a topic for a phone or tablet.

    4. "Another one was that on some embedded systems, you map the same physical memory to multiple, different virtual address regions with different permissions/characteristics (cacheability as one example)"

      "Not just that. Even if you only have a single mapping, some IO mappings can be quite big."

      "There are also good reasons to support large virtual address spaces independently of amount of physical RAM you want to map. For example, when you want to emulate something else."

      ...etc. I see nothing intel-specific here (nothing too hardware-specific even). Also Apple seems aiming further than just smartphones - "desktop-class architecture" gives a hint.

  2. Yes, and not one of them have anything to do with usage on a phone, which is the point... Apple is bragging about doing something that doesn't make sense (at this time).
    Eventually it will make sense, certainly once more than 4gb becomes desirable. But for now, it is actually wasteful.
    A 64-bit executable generally approaches twice the size of a 32-bit executable, meaning the limited cache on the chip is less well utilized.
    In application usage, 64-bit apps tend to need about 50% more memory, wasting precious memory resources.
    Plainly and simply, Apple's use of a 64-bit processor on a phone at this time is unwise.

    The discussion you linked to started from this:
    (See the post 4th from the bottom.)
    It has nothing to do with whatsoever to do with what is desirable in an iPhone 5S.

    If you want to make a point here, please make one relevant to Apple's iPhone 5S.

    If Apple is aiming further, fine... _then_ they should use a 64-bit chip, for the iPhone 5S it is a bad design decision.

    1. Ah, come on, phones or not, all next generation ARMs are going to be ARMv8 compatibles, and that means 64-bit, Apple's just being first, thats about it.

    2. I agree that many of the next round of ARM chips will be 64-bit. But many will not be. For example the ones headed for watches and other wearables will not be 64-bit.

      Yes, Apple was first, but that does not mean it was a wise design decision. Essentially, Apple is penalizing their customers with less performance in order to score a marketing point. It may actually have made sense from Apple's point of view to skip to the 64-bit chip if they do intend to use it soon in a more appropriate settings. (It would save them the cost of developing two chips.) Again, penalizing their customer with less performance for their own ends.

  3. Those won't belong to the *next-generation* architecture then (and in many cases won't be "application" cores even). Also, ARM promised better 32-bit legacy performance with ARMv8, do you think they lied about it?

  4. My point on the "next generation" is that they will be the appropriate size needed for the application.

    The performance of the 32-bit code has nothing to do with this.
    The performance Apple is denying their customers is from the unnecessary use of 64-bit code.
    Because 64-bit code is bigger, the cache is less effective and memory utilization is higher.
    Apple said that the OS and all standard applications will be 64-bit.

    1. Now you're thinking x86 when discussing arm. But unlike x86-64, arm's 64-bit instruction set is much more different from its predecessor - easier to pipeline, tailored to modern hardware capabilities, with much less dependencies to watch. That could mean less misprediction and pipeline stalls, less cache misses and less cache utilization overall. So it's more like 64-bit arm OS could be faster AND less power-hungry at the same time (that's beside improved SIMD performance gains)

    2. I have no doubt that ARM will do a better job of 64-bit than Intel did as they get to see where Intel (really AMD) came up short.

      The points about the wasted cache and wasted memory when you unnecessarily go to 64-bit still hold true.

    3. Ok, the question is: will the pros outweigh the cons?

      Well, methinks they will! ;)

    4. ...and looks like they do:

      Geekbench3 iPhone5s single/multi score:
      iOS7 (32-bit) =1078/1938
      iOS7 (64-bit) =1414/2564

    5. Thank you for the link, I've added the results to the ARM vs. Intel Benchmarks Page (you can get to the link at the top right of this page, under Resources).

      It has wonderful single core numbers, but alas it is only a dual core. It loses in the multithreaded numbers against all the leading 32-bit ARM chips and also to Bay Trail (which are all quad core).

      My first thought was this is great news for ARM on servers. However, AnandTech is saying that it is NOT based on the Cortex A57 (it is being built by Samsung by the way... AppleInsider keeps claiming its from TMSC). Wow, Apple's being really strange here... they made a custom 64-bit chip for a phone... that's just crazy.

      However, w.r.t. the current discussion. The speed here has nothing to do with it being 64-bit. The speed comes from a combination of smaller die and better internals. To be clear, if they had used the same technology in a 32-bit chip, it would perform even better.

      I'm a bit shocked that the 32-bit benchmarks (ran on a 64-bit OS and chip) are that bad. No wonder Apple is pushing developers to do 64-bit versions of their apps.

    6. Well, maybe, but that means different ARM-incompatible 32-bit ISA, doing that could be *really* crazy move right now.

      Also 64/32-bit geekbench difference should be not so big w/o the crypto parts, i suppose.


Now allowing anonymous comments (but they are moderated).