The Apple iPad Review (2012)
by Vivek Gowri & Anand Lal Shimpi on March 28, 2012 3:14 PM ESTThe 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.
Facebook
Twitter
RSS
234 Comments
View All Comments
name99 - Friday, March 30, 2012 - link
Just to clarify, this is NOT some Apple proprietary thing. The Apple ports are following the USB charging spec. This is an optional part of the spec, but any other manufacturer is also welcome to follow it --- if they care about the user experience.darkcrayon - Thursday, March 29, 2012 - link
All recent Macs (last 2-3 years) can supply additional power via their USB ports which is enough to charge an iPad that's turned on (though probably not if it's working very hard doing something). Most non-Mac computer USB ports can only deliver the standard amount of USB power, which is why you're seeing this.Your Lenovo *should* still recharge the iPad if the iPad is locked and sleeping, though it will do so very slowly.
dagamer34 - Friday, March 30, 2012 - link
I did the calculations and it would take about 21 hours to recharge an iPad 3 on a normal non-fast charging USB port from dead to 100%. Keep in mind, we're talking about a battery that's larger in capacity than the 11" MacBook Air.snoozemode - Thursday, March 29, 2012 - link
http://www.qualcomm.com/media/documents/files/snap...Aenean144 - Thursday, March 29, 2012 - link
Anandtech: "iPhoto is a very tangible example of where Apple could have benefitted from having four CPU cores on A5X"Is iPhoto really a kind of app that can actually take advantage of 2 cores? If there are batch image processing type functionality, certainly, though I don't know if iPhoto for iOS has this type of functionality. The slowness could just be from a 1.0 product and further tuning and refinement will fix it.
I'm typically highly skeptical of the generic "if the app is multithreaded, it can make use of all of the cores" line of thought. Basically all of the threads, save one, are typically just waiting on user input.
Anand Lal Shimpi - Thursday, March 29, 2012 - link
It very well could be that iOS iPhoto isn't well written, but in using the editing tools I can typically use 60 - 95% of the A5X's two hardware threads. Two more cores, at the bare minimum, would improve UI responsiveness as it gives the scheduler another, lightly scheduled core to target.Alternatively, a 50% increase in operating frequency and an improvement in IPC could result in the same net benefit.
Take care,
Anand
shompa - Friday, March 30, 2012 - link
*hint* Use top on a iOS/Android device and you will see 30-60 processes at all time. The single threaded, single program thinking is Windows specific and have been solved on Unix since late 1960. Todays Windows phones are all single threaded because windows kernel is not good at Multit hreding.With many processes running, it will always be beneficial to have additional cores. Apple have also solved it in OSX by adding Grand central dispatch in their development tools making multithreaded programs easy.
Iphoto for Ipad: Editing 3 million pixel will demand huge amount of CPU/GPU time + memory. Apple have so far been able to program elegant solutions around the limits of ARM CPUs by using NOVA SIMD extensions and GPU acceleration. An educated guess is that Iphoto is not fully optimized and will be at later time.
(the integrated approach gives Apple a huge advantage over Android since Apple can accelerate stuff with SIMDs. Google does not control the hardware and can therefore not optimize its code. That is one of the reasons why single core A4 was almost as fast as dual core Tegras. I was surpassed when Google managed to implement their own acceleration in Andriod 4.X. Instead of SIMD, Google uses GL, since all devices have graphics cards. This is the best feuture in Android 4.x.)
name99 - Thursday, March 29, 2012 - link
[/quote]Apple’s design lifespan directly correlates to the maturity of the product line as well as the competitiveness of the market the product is in.
[/quote]
I think this is completely the wrong way to look at it. Look across the entire Apple product line.
I'd say a better analysis of chassis is that when a product first comes out, Apple can't be sure how it will be used and perceived, so there is some experimentation with different designs. But as time goes by, the design becomes more and more perfected (yes yes, if you hate Apple we know your feelings about the use of this word) and so there's no need to change until something substantial drives a large change.
Look, for example, at the evolution of iMac from the Luxo Jr version to the white all-in-on-flatscreen, to the current aluminum-edged flatscreen which is largely unchanged for what, five or six years now. Likewise for the MacBook Pro.
Look at the MacBook Air. The first two revs showed the same experimentation, trying different curves and angles, but Apple (and I'd say customers) seems to feel that the current wedge shape is optimal --- a definite improvement on the previous MBA models, and without anything that obviously needs to be improved. (Perhaps the sharp edges could be rounded a little, and if someone could work out the mechanicals, perhaps the screen could tilt further back.)
And people accept and are comfortable with this --- in spite of "people buy Apple as a fashion statement idiocy". No-one will be at all upset if the Ivy League iMacs and MBAs and Mac Minis look like their predecessors (apart from minor changes like USB3 ports) --- in fact people expect it.
So for iPhone and iPad. Might Apple keep using the same iPhone4 chassis for the next two years, with only minor changes? Why not? There's no obvious improvement it needs.
(Except, maybe, a magnet on the side like iPad has, so you could slip a book-like case on it that covered the screen, and switched it on by opening the book.)
Likewise for iPad.
New must have features in phones/tablets (NFC? near-field charging? waterproof? built-in projector like Samsung Beam?) might change things. But absent those, really, the issue is not "Apple uses two year design cycles", it is "Apple perfects the design, then sticks with it".
mr_ripley - Thursday, March 29, 2012 - link
"In situations where a game is available in both the iOS app store as well as NVIDIA's Tegra Zone, NVIDIA generally delivers a comparable gaming experience to what you get on the iPad... The iPad's GPU performance advantage just isn't evident in those cases..."Would you expect it to be if all the games you compare have not been optimized for the new ipad yet? They run at great frame rates but suffer in visuals or are only available at ipad 2 resolutions. The tegra zone games are clearly optimized for Tegra while their iOS counterparts are not optimized for the A5x, so of course the GPU advantage is not evident.
This comparison does not seem fair unless there is a valid reason to believe that the tegra zone games cannot be further enhanced/optimized to take advantage of the new ipad hardware.
I suspect that the tegra zone games optimized for A5x will offer a tangibly superior performance and experience. And the fact that the real world performance suffers today does not mean we will not see it shortly.
Steelbom - Thursday, March 29, 2012 - link
Exactly this.