The GPU

3D rendering is a massively parallel problem. Your GPU ultimately has to determine the color value of each pixel which may not remain constant between frames, at a rate of dozens of times per second. The iPad 2 had 786,432 pixels in its display, and by all available measures its GPU was more than sufficient to drive that resolution. The new iPad has 3.14 million pixels to drive. The iPad 2's GPU would not be sufficient.

When we first heard Apple using the term A5X to refer to the new iPad's SoC, I assumed we were looking at a die shrunk, higher clock version of the A5. As soon as it became evident that Apple remained on Samsung's 45nm LP process, higher clocks were out of the question. The only room for improving performance was to go wider. Thankfully, as 3D rendering is a massively parallel problem, simply adding more GPU execution resources tends to be a great way of dealing with a more complex workload. The iPad 2 shocked the world with its dual-core PowerVR SGX 543MP2 GPU, and the 3rd generation iPad doubled the amount of execution hardware with its quad-core PowerVR SGX 543MP4.

Mobile SoC GPU Comparison
  Adreno 225 PowerVR SGX 540 PowerVR SGX 543MP2 PowerVR SGX 543MP4 Mali-400 MP4 Tegra 2 Tegra 3
SIMD Name - USSE USSE2 USSE2 Core Core Core
# of SIMDs 8 4 8 16 4 + 1 8 12
MADs per SIMD 4 2 4 4 4 / 2 1 1
Total MADs 32 8 32 64 18 8 12
GFLOPS @ 200MHz 12.8 GFLOPS 3.2 GFLOPS 12.8 GFLOPS 25.6 GFLOPS 7.2 GFLOPS 3.2 GFLOPS 4.8 GFLOPS
GFLOPS @ 300MHz 19.2  GFLOPS 4.8 GFLOPS 19.2 GFLOPS 38.4
GFLOPS
10.8 GFLOPS 4.8 GFLOPS 7.2 GFLOPS
GFLOPS As Shipped by Apple/ASUS - - 16 GFLOPS 32 GFLOPS - - 12
GFLOPS

We see this approach all of the time in desktop and notebook GPUs. To allow games to run at higher resolutions, companies like AMD and NVIDIA simply build bigger GPUs. These bigger GPUs have more execution resources and typically more memory bandwidth, which allows them to handle rendering to higher resolution displays.

Apple acted no differently than a GPU company would in this case. When faced with the challenge of rendering to a 3.14MP display, Apple increased compute horsepower and memory bandwidth. What's surprising about Apple's move is that the A5X isn't a $600 desktop GPU, it's a sub 4W mobile SoC. And did I mention that Apple isn't a GPU company?

That's quite possibly the most impressive part of all of this. Apple isn't a GPU company. It's a customer of GPU companies like AMD and NVIDIA, yet Apple has done what even NVIDIA would not do: commit to building an SoC with an insanely powerful GPU.

I whipped up an image to help illustrate. Below is a representation, to-scale, of Apple and NVIDIA SoCs, their die size, and time of first product introduction:

If we look back to NVIDIA's Tegra 2, it wasn't a bad SoC—it was basically identical in size to Apple's A4. The problem was that the Tegra 2 made its debut a full year after Apple's A4 did. The more appropriate comparison would be between the Tegra 2 and the A5, both of which were in products in the first half of 2011. Apple's A5 was nearly 2.5x the size of NVIDIA's Tegra 2. A good hunk of that added die area came from the A5's GPU. Tegra 3 took a step in the right direction but once again, at 80mm^2 the A5 was still over 50% larger.

The A5X obviously dwarfs everything, at around twice the size of NVIDIA's Tegra 3 and 33.6% larger than Apple's A5. With silicon, size isn't everything, but when we're talking about similar architectures on similar manufacturing processes, size does matter. Apple has been consistently outspending NVIDIA when it comes to silicon area, resulting in a raw horsepower advantage, which in turns results in better peak GPU performance.

Apple Builds a Quad-Channel (128-bit) Memory Controller

There's another side effect that you get by having a huge die: room for wide memory interfaces. Silicon layout is a balancing act. You want density to lower costs, but you don't want hotspots so you need heavy compute logic to be spread out. You want wide IO interfaces but you don't want them to be too wide because then you'll cause your die area to balloon as a result. There's only so much room on the perimeter of your SoC to get data out of the chip, hence the close relationship between die size and interface width.

Most mobile SoCs are equipped with either a single or dual-channel LP-DDR2 memory controller. Unlike in the desktop/notebook space where a single DDR2/DDR3 channel refers to a 64-bit wide interface, in the mobile SoC world a single channel is 32-bits wide. Both Qualcomm and NVIDIA use single-channel interfaces, with Snapdragon S4 finally making the jump to dual-channel this year. Apple, Samsung, and TI have used dual-channel LP-DDR2 interfaces instead.

With the A5X Apple did the unthinkable and outfitted the chip with four 32-bit wide LP-DDR2 memory controllers. The confirmation comes from two separate sources. First we have the annotated A5X floorplan courtesy of UBMTechInsights:

You can see the four DDR interfaces around the lower edge of the SoC. Secondly, we have the part numbers of the discrete DRAM devices on the opposite side of the motherboard. Chipworks and iFixit played the DRAM lottery and won samples with both Samsung and Elpida LP-DDR2 devices on-board, respectively. While both Samsung and Elpida do a bad job of updating public part number decoders, both strings match up very closely to 216-ball PoP 2x32-bit PoP DRAM devices. The part numbers don't match up exactly, but they are close enough that I believe we're simply looking at a discrete flavor of those PoP DRAM devices.


K3PE4E400M-XG is the Samsung part number for a 2x32-bit LPDDR2 device, K3PE4E400E-XG is the part used in the iPad. I've made bold the only difference.

A cross reference with JEDEC's LP-DDR2 spec tells us that there is an official spec for a single package, 216-ball dual-channel (2x32-bit) LP-DDR2 device, likely what's used here on the new iPad.


The ball out for a 216-ball, single-package, dual-channel (64-bit) LPDDR2 DRAM

This gives the A5X a 128-bit wide memory interface, double what the closest competition can muster and putting it on par with what we've come to expect from modern x86 CPUs and mainstream GPUs.

The Geekbench memory tests show no improvement in bandwidth, which simply tells us that the interface from the CPU cores to the memory controller hasn't seen a similar increase in width.

Memory Bandwidth Comparison—Geekbench 2
  Apple iPad (3rd gen) ASUS TF Prime Apple iPad 2 Motorola Xyboard 10.1
Overall Memory Score 821 1079 829 1122
Read Sequential 312.0 MB/s 249.0 MB/s 347.1 MB/s 364.1 MB/s
Write Sequential 988.6 MB/s 1.33 GB/s 989.6 MB/s 1.32 GB/s
Stdlib Allocate 1.95 Mallocs/sec 2.25 Mallocs/sec 1.95 Mallocs/sec 2.2 Mallocs/sec
Stdlib Write 2.90 GB/s 1.82 GB/s 2.90 GB/s 1.97 GB/s
Stdlib Copy 554.6 MB/s 1.82 GB/s 564.5 MB/s 1.91 GB/s
Overall Stream Score 331 288 335 318
Stream Copy 456.4 MB/s 386.1 MB/s 466.6 MB/s 504 MB/s
Stream Scale 380.2 MB/s 351.9 MB/s 371.1 MB/s 478.5 MB/s
Stream Add 608.8 MB/s 446.8 MB/s 654.0 MB/s 420.1 MB/s
Stream Triad 457.7 MB/s 463.7 MB/s 437.1 MB/s 402.8 MB/s

Although Apple designed its own memory controller in the A5X, you can see that all of these A9 based SoCs deliver roughly similar memory performance. The numbers we're showing here aren't very good at all. Even though Geekbench has never been good at demonstrating peak memory controller efficiency to begin with, the Stream numbers are very bad. ARM's L2 cache controller is very limiting in the A9, something that should be addressed by the time the A15 rolls around.

Firing up the memory interface is a very costly action from a power standpoint, so it makes sense that Apple would only want to do so when absolutely necessary. Furthermore, notice how the memory interface moved from being closer to the CPU in A4/A5 to being adjacent to the GPU in the A5X. It would appear that only the GPU has access to all four channels.

The A5X SoC A Word on Packaging & Looking Forward
POST A COMMENT

232 Comments

View All Comments

  • PeteH - Tuesday, April 03, 2012 - link

    I have to be honest, after reading through that link I didn't see anything that even implied working conditions had anything to do with the suicides of the factory workers. The only suicide for which there was any real information provided was that of the worker who killed himself after losing the iPhone prototype, and in that case the victim wasn't a factory worker, but someone in logistics.

    Did working conditions have anything to do with the factory worker suicides? Maybe, maybe not. There doesn't appear to be evidence either way.
    Reply
  • mr_ripley - Wednesday, April 04, 2012 - link

    I posted the Wikipedia link for all the links in the reference section.

    Here's a more direct report: http://sacom.hk/wp-content/uploads/2010/11/report-...

    And a companion video: http://vimeo.com/17558439

    The video includes an interview of a survivor who is now paralyzed waist down.

    You can choose to patiently read and watch this report or just turn a blind eye like a lot of people do.
    Reply
  • PeteH - Thursday, April 05, 2012 - link

    I did read the report. It details unbelievably miserable working conditions in the factories, which I don't think anyone is disputing, and concludes that the way to change those conditions is to pressure the electronics companies making the bulk of the profits. None of the above comments dispute any of this. However it does not link working conditions to suicides among factory workers.

    And yet you continue to insist that there is a link, with no evidence to back it up. You make statements like, "over a hundred of them have committed suicide over the working conditions," "...scores of people killing themselves citing poor working conditions," and "there is no disputing the fact that these deaths are related to working conditions," but you provide only conjecture to back it up, no proof. When you do this people start dismissing everything you say out of hand, even the things that are accurate. And worse than that, you run the risk that other people arguing for better working conditions will be tarred with the same brush. Look at what happened to Mike Daisy.

    Again, I'm not saying working conditions didn't contribute to the suicides, I'm saying there is no evidence one way or the other. Until you have evidence (in the form of suicide notes, higher suicide rates among factory workers, etc.) please stop. You may actually be hurting the very movement you're trying to help.
    Reply
  • mr_ripley - Thursday, April 05, 2012 - link

    Well, I'm sorry if it is inconvenient for you that these individuals have not said it in so many words. Should we expect them to?... Hey, by the way, I know you're going to kill yourself but why don't your write down an explanation first so we can conclusivly say what the reasons are. And even though you are under a lot of stress right now and are clearly not thinking straight SPELL it our for me please...

    Evidence can come in different forms. Not all of it is directly incriminating, in which case the attention turns to the circumstances. So if these reports don't establish a reasonbly clear coorelation to you, then I am sorry but I disagree.

    You can nitpick on specific words in my comments and quible about words such as evidence. But what are you accomplishing here? Are you justifying your own guilt of purchasing a device manufactured here? Are you an Apple or Foxconn mouthpeice? Do they pay your for spreading lies like Foxconn factories are actually a good place to work (which has been said in the previous comments)? Really, it is people like you need to STOP.

    I'm not going to stop saying what I believe is right!! And unlike Mike Daisy I have not fabricated any evidence. At the most, you can complain that I have drawn incorrect conclusions and I am saying the same about you.
    Reply
  • PeteH - Thursday, April 05, 2012 - link

    It's inconvenient for me that you are lying. You're the one saying that there are, "...scores of people killing themselves citing poor working conditions," not me. Either show me a case where those people who killed themselves cited poor working conditions as the reason, or cease claiming it is fact. You do damage to the movement that's trying to improve things.

    People hear the news reports about Mike Daisy lying to Ira Glass and what they take away is not the specific lies (claiming to witness things had actually happened but that he had only read about), it's that he's a liar and the story wasn't true. They dismiss the whole issue of poor working conditions out of hand. That's what you risk when you lie to get people to listen.
    Reply
  • mr_ripley - Thursday, April 05, 2012 - link

    Go ahead, nitpick on specific phrases and completely lose the meaning. But the problem is easily corrected. One can argue that citing something does not have to be done on paper as you would in a professional article. To me the workers "cite" the existence of a problem through their actions as words have failed them.

    Still if you want me to rephrase I'll say "scores of people killing themselves in midst of poor working conditions.." Can you prove that this statement is inaccurate??

    And while you ask me for evidence have you ever bothered to see if you can find evidence that these deaths are not related to working conditions. Prove it to me and I'll take back everything I said.
    Reply
  • PeteH - Thursday, April 05, 2012 - link

    I think you missed the places above where I stated, "I'm not saying working conditions didn't contribute to the suicides, I'm saying there is no evidence one way or the other." That was my whole point. And I did explicitly state that there's no disputing the poor working conditions. So no, I have no problem with your revised statement.

    However, I don't think what I did was nitpicking at all. Nitpicking would be pointing out that a score is 20, so scores would imply at least 40, and I've only seen documentation of 17 suicides (I haven't seen numbers pre-2010). But that's not what I did.
    Reply
  • shompa - Friday, March 30, 2012 - link

    Manufacturing employees?

    Look at the world! There are about 20 countries in the world that are democratic and have great living standards. Its just 100 years ago since these countries had child workers and harsh condition.

    BTW. My county is on the "top countries" in the world. Still we have the largest suicide rate in our population in the world. Why are you not fighting against the Swedish government that drives thousands to kill them self each year? We live like slaves here with 80% taxes.

    BTW. Do you care if other companies use HonHai/FoxConn or is it just Apple? Are you writing the same thing about Dell/HP and all other companies that use FoxConn?

    What have you done?
    Have you donated money to a chinese worker? Or is Trolling the only thing you manage to do?

    Reply
  • grave00 - Friday, March 30, 2012 - link

    I was curious about this statement. Could you elaborate. What inconsistency is there?

    "On the iPhone Apple has been entirely too lax about maintaining consistency between suppliers. If it wants to be taken seriously in this space Apple needs to ensure a consistent experience across all of its component vendors."
    Reply
  • loboracing - Friday, March 30, 2012 - link

    I remember an ad that touted something new to "see and touch". The retina screen is the "see" part but what about the "touch"? Was that just a gimmick meaning you could touch the screen, or is there some sort of different feel to the screen? Reply

Log in

Don't have an account? Sign up now