Apple's Swift: Visualized

Section by Anand Shimpi

Based on my findings on the previous pages, as well as some additional off-the-record data, this is what I believe Swift looks like at a high level:


Note that most of those blocks are just place holders as I don't know how they've changed from Cortex A9 to Swift, but the general design of the machine is likely what you see above. Swift moves from a 2-wide to a 3-wide machine at the front end. It remains a relatively small out-of-order core, but increases the number of execution ports from 3 in Cortex A9 to 5. Note the dedicated load/store port, which would help explain the tremendous gains in high bandwidth FP performance.

I asked Qualcomm for some additional details on Krait unfortunately they are being quite tight lipped about their architecture. Krait is somewhat similar to Swift in that it has a 3-wide front end, however it only has 4 ports to its 7 execution units. Qualcomm wouldn't give me specifics on what those 7 units were or how they were shared by those 4 ports. It's a shame that Intel will tell me just how big Haswell's integer and FP register files are 9 months before launch, but its competitors in the mobile SoC space are worried about sharing high level details of architectures that have been shipping for half a year.

Apple's Swift core is a wider machine than the Cortex A9, and seemingly on-par with Qualcomm's Krait. How does ARM's Cortex A15 compare? While the front end remans 3-wide, ARM claims a doubling of fetch bandwidth compared to Cortex A9. The A15 is also able to execute more types of instructions out of order, although admittedly we don't know Swift's capabilities in this regard. There's also a loop cache at the front end, something that both AMD and Intel have in their modern architectures (again, it's unclear whether or not Swift features something similar). ARM moves to three dedicated issue pools feeding 8 independent pipelines on the execution side. There are dedicated load and store pipelines, two integer ALU pipes, two FP/NEON pipes, one pipe for branches and one for all multiplies/divides. The Cortex A15 is simply a beast, and it should be more power hungry as a result. It remains to be seen how the first Cortex A15 based smartphone SoCs will compare to Swift/Krait in terms of power. ARM's big.LITTLE configuration was clearly designed to help mitigate the issues that the Cortex A15 architecture could pose from a power consumption standpoint. I suspect we haven't seen the end of NVIDIA's companion core either.

At a high level, it would appear that ARM's Cortex A15 is still a bigger machine than Swift. Swift instead feels like Apple's answer to Krait. The release cadence Apple is on right now almost guarantees that it will be a CPU generation behind in the first half of next year if everyone moves to Cortex A15 based designs.

Custom Code to Understand a Custom Core Apple's Swift: Pipeline Depth & Memory Latency
Comments Locked

276 Comments

View All Comments

  • rarson - Wednesday, October 17, 2012 - link

    Car bumpers are not made of aluminum.

    Aluminum oxidizes. So if you scratch it, then you've removed that oxidation layer to allow it to further oxidize at that spot. Rust is just iron oxidation.
  • Spunjji - Friday, October 19, 2012 - link

    It is not normal for them to scratch so damn easily. Furthermore, you might notice that other manufacturers (say, HTC?) take steps to harden the surfaces of their devices to avoid this kind of problem.
  • name99 - Wednesday, October 17, 2012 - link

    So you're basically
    (a) upset that Apple fans buy products based on how they look
    (b) upset that Apple fans's don't care enough about how products look to care about this
    ???

    The true sign of the demented mind --- that it can happily hold two contradictory opinions at once.
  • steven75 - Wednesday, October 17, 2012 - link

    Would you buy a car that gets nicks and scratches from normal usage?

    Um yes, everyone does. I guess all cars should be recalled!
  • Spunjji - Friday, October 19, 2012 - link

    Would you buy a car that gets nicks and scratches from simply driving down the street? No, you wouldn't. Stop distorting the argument for an easy victory, it makes for extremely aggravating reading.
  • doobydoo - Saturday, October 20, 2012 - link

    And what evidence do you have that the equivalent of 'driving down the street' with an iPhone causes scratching?
  • ltcommanderdata - Tuesday, October 16, 2012 - link

    Any final MHz rating on the GPU? Given Apple tends to use a 4:1 clock speed ratio between the CPU and GPU, the SGX543MP3 being up to 325MHz would make sense. The SGX543MP2 seemed to be clocked at 200MHz in the iPhone 4S and 250MHz in the iPad 2 and Apple said the iPad 2012 has a 2x faster GPU, so the SGX543MP4 in the A5X likely is also at 250MHz. A SGX543MP3 at 325MHz vs a SGX543MP4 at 250MHz would seem to explain the results seen in the benchmarks.

    A few corrections, on page 11 the GLBenchmark 2.5 - Triangle Texture Test - Fragment Lit (Offscreen 180p) is missing the iPad 2012 result.

    In iPhone 5 Device Conclusions on page 22, you write "Going back to the old 4:3 aspect ratio iPhones feels extremely claustrophobic now", but it should be 3:2.
  • daar - Tuesday, October 16, 2012 - link

    The in-depth tech info was nice, but would have preferred it in a second post. As an engineer, while I can appreciate the advances made with the new SoC and the depth of the effort went into researching all the aspects of the phone, I also think for most purposes, the length is counterproductive when the majority of readers are looking for indicators of whether the phone is worth an upgrade. Even without the tech explanation though, the review unnaturally lacked the concise detail I'm used to at AT.

    In some ways, it sort of came across that the tech explanation was a long worded way of making excuses for the iPhone 5's faults and direct comparisons to superior implementations were ignored. Simple example would be the camera, where praise was given about how they cut the size, that it looked good, explanation of the purple tint and so forth. If say, Samsung had released a phone with such issues, I'd expect the review to mention the sloppiness of it, esp with rivals such as the One X having a 2.0f lens (I quite enjoyed the One X/SG3 review comparison from AT). The excuse that the lack of innovation in the new iOS being that the aim of the phone is like that of an appliance whereas Android phones aiming to be PC's is baffling; the concept of a smartphone was a versatile device to aid in our daily lives not reach a point of some ambiguity called an appliance.
  • darwinosx - Tuesday, October 16, 2012 - link

    That was a lot of words to say nothing besides bragging that you are an engineer. Nobody cares.
  • kyuu - Tuesday, October 16, 2012 - link

    What nobody cares about are your rabid attacks on any comment that has even the slightest critique of an Apple product.

Log in

Don't have an account? Sign up now