A8’s GPU: Imagination Technologies’ PowerVR GX6450

Last but not least on our tour of the A8 SoC is Apple’s GPU of choice, Imagination’s PowerVR GX6450.

When Apple first announced the A8 SoC as part of their iPhone keynote, they told us to expect a nearly 50% increase in graphics performance. Based on that information and on the fact that that Apple was moving to a denser 20nm process, we initially believed that Apple would be upgrading from A7’s 4-core PowerVR design to a 6-core design, especially in light of the higher resolution displays present on the iPhone 6 and iPhone 6 Plus.

Instead our analysis with Chipworks found that only four GPU cores were present on A8, which ruled out the idea of a 6-core design but did narrow down the options considerably. Based on that information and more importantly Apple’s Metal Programming Guide, we have been able to narrow down our options to a single GPU, the PowerVR GX6450.

The GX6450 is the immediate successor to the G6430 first used in the A7 and is based on Imagination’s PowerVR Series6XT architecture. Imagination first announced PowerVR Series6XT to the public at CES 2014, and now just a short eight months later we are seeing the first Series6XT hardware reach retail.

We have already covered the PowerVR Series6/Series6XT architecture in some detail earlier this year so we won’t go through all of it again, but we would encourage anyone who is interested to take a look at our architectural analysis for additional information. Otherwise we will be spending the bulk of our time looking at how GX6450 differs from G6430 and why Apple would choose this specific GPU.

From a technical perspective Series6XT is a direct evolution over the previous Series6, and GX6450 is a direct evolution over G6430 as well. Given a 4-core configuration there are only a limited number of scenarios where GX6450 outright has more hardware than G6430 (e.g. additional ALUs), and instead Series6XT is focused on adding features and improving performance over Series6 through various tweaks and optimizations to the architecture. Series6 at this point is actually over two years old – it was first introduced to the public at CES 2012 – so a lot has happened in the mobile GPU landscape over the past couple of years.

The closest thing to a marquee feature on Series6XT is support for Adaptive Scalable Texture Compression (ASTC), a next-generation texture compression technology that is slowly making its way into GPUs from a number of manufacturers. Designed by the consortium responsible for OpenGL ES, Khronos, ASTC is designed to offer better texture compression (with finer grained quality options) than existing texture compression formats while also being a universal format supported by all GPUs. In Apple’s case they have always been using PowerVR GPUs – and hence all products support PVRTC and more recently PVRTC2 – however ASTC being exposed allows them to take advantage of the quality improvements while also making game development and porting from other platforms easier.

Less visible to users but certainly important to Apple, Series6XT also includes new power management capabilities to reduce power consumption under idle and light workloads. Through finer grained power gating technology that Imagination dubs “PowerGearing G6XT”, GX6450 can now have its shading clusters (USCs) powered down individually, allowing only as many of them as are necessary to be fired up. As Apple continues to min-max their designs, being able to idle at a lower power state can be used to improve battery life and/or increase how often and how long the A8’s GPU uses higher power states, improving overall efficiency.

Apple iPhone GPU Performance Estimate: Over The Years

And, perhaps most importantly overall, Series6XT comprises a series of under-the-hood optimizations to improve overall performance. When it comes to the internals of PowerVR architectures we only have limited details from Imagination on how they operate, so in some areas we know quite a bit about what Imagination has been up to and in other areas their architectures are still something akin to a black box. At any rate Imagination’s goal for Series6XT was to improve performance by up to 50% – this seems to be where Apple’s 50% performance improvement claim comes from – though as we’ll see the performance gains on real world applications are not going to be quite as potent.

What we do know about Series6XT is that Imagination has made some changes to the structure of the USCs themselves. Series6XT still uses a 16-wide SIMD design, but in each pipeline they have added another set of medium/half-precision (FP16) ALUs specifically to improve FP16 performance. Now instead of 2x3 (6) FP16 ALUs, Series6XT bumps that up to 4x2 (8) FP16 ALUs. This is the only outright increase in shader hardware when you compare Series6 to Series6XT, and on paper it improves FP16 performance by 33% at equivalent clock speeds.

The focus on FP16 is interesting, though for iOS it may be misplaced. These half-precision floating point operations are an excellent way to conserve bandwidth and power by not firing up more expensive FP32 ALUs, but the tradeoff is that the numbers they work with aren’t nearly as precise, hence their use has to be carefully planned. In practice what you will find is that while FP16 operations do see some use, they are by no means the predominant type of floating point GPU operation used, so the FP16 increase is a 33% increase only in the cases where performance is being constrained by the GPU’s FP16 performance.

FP32 performance meanwhile remains unchanged. Each USC pipeline contains two such ALUs, for up to four FP32 FLOPS per clock, or to use our typical metric, 128 MADs (Multiply-Adds) per clock.

The rest of Series6XT’s optimizations take place at the front and back ends, where geometry processing and pixel fill take place respectively. Imagination has not told us exactly what they have done here, but both these areas have been targeted to improve sustained polygon rates and pixel fillrate performance. These more generic optimizations stand to be more applicable to general performance, though by how much we cannot say.

One final optimization we want to point out for Series6XT is that Imagination has made some additional under-the-hood changes to improve GPU compute performance. We have not talked about GPU compute on iOS devices thus far, as until now Apple has not exposed any APIs suitable for it (e.g. OpenCL is not available on iOS). With iOS8 Apple is releasing their Metal API, which is robust enough to be used for both graphics and now compute. How developers put this capability to use remains to be seen, but GX6450 should perform even better than G6430.

Mobile SoC GPU Comparison
  PowerVR SGX 543MP2 PowerVR SGX 543MP3 PowerVR SGX 543MP4 PowerVR SGX 554MP4 PowerVR G6430 PowerVR GX6450
Used In iPad 2/iPhone 4S iPhone 5 iPad 3 iPad 4 iPad Air/iPhone 5s iPhone 6/iPhone 6Plus
# of SIMDs 8 12 16 32 4 4
MADs per SIMD 4 4 4 4 32 32
Total MADs 32 48 64 128 128 128
Pixels/Clock N/A N/A N/A N/A 8 8
Texels/Clock N/A N/A N/A N/A 8 8

The one wildcard when talking about performance here is going to be clock speeds. Apple doesn’t expose these and they aren’t easy to test for (yet), though in the long term Metal offers some interesting possibilities for nailing that down, or at least getting a better idea of relative clock speeds.

In any case, we’ll take a look at our GPU benchmarks in depth in a bit, but overall GPU performance compared to A7 and its G6430 is consistently better, but the exact performance gain will depend on the test at hand. Some tests will come very close to reaching 50% while others will be just 15-20%. The dependent factor generally seems to be whether the test is ALU-bound or not; because the USC has not changed significantly from G6430 to GX6450 outside of those additional FP16 ALUs, tests that hit the FP32 ALUs in particular show less of an improvement. Otherwise more balanced tests (or at least tests more defined by pixel fillrate performance) can show greater gains. In general we should be looking at a 30-35% performance improvement.

Why Four Cores?

One thing that admittedly surprised us in the revelation that A8 was using a 4-core PowerVR design was that we figured a 6-core design would be a shoe-in for A8, especially since Apple was on the receiving end of the density improvements from TSMC’s 20nm process. But upon further reflection an additional two cores is likely more than Apple needed nor wanted.

The biggest factor here is that coming from G6430 in the A7, performance has seen a solid improvement despite sticking to only four GPU cores. Due to the combination of performance improvements from the Series6XT architecture and any clock speed increases from Apple, A8 gets quite a bit more GPU performance to play with. The increased resolution of the iPhone 6 screen in turn requires more performance if Apple wants to keep native resolution performance from significantly regressing, which GX6450 is capable of delivering on. Never mind the fact that G6430 also drove the iPad Air and its much larger 2048x1536 pixel display.

PowerVR Series6/6XT "Rogue"
GPU # of Clusters # of FP32 Ops per Cluster Total FP32 Ops Optimization
G6200 2 64 128 Area
G6230 2 64 128 Performance
GX6240 2 64 128 Area
GX6250 2 64 128 Performance
G6400 4 64 256 Area
G6430 4 64 256 Performance
GX6450 4 64 256 Performance
G6630 6 64 384 Performance
GX6650 6 64 384 Performance

These performance improvements in Series6XT have a cost as well, and that cost is suitably reflected in the estimated die sizes for each GPU. The G6430 was 22.1mm2 on the 28nm A7, while the GX6450 is 19.1mm2 on A8. Though GX6450 is smaller overall, it’s nowhere near the roughly 11.1mm2 a pure and perfect die shrink of G6430 would occupy. Limited area scaling aside, GX6450’s additional functionality and additional performance requires more transistors, and at the end of the day Apple doesn’t see a significantly smaller GPU because of this. In other words, the upgrade from G6430 to GX6450 has delivered much of the performance (and consumed much of the die space) we initially expected to be allocated to a 6-core GPU.

Overall the choice of GX6450 seems to be one of picking the GPU best for a phone, which is an area the G6430 proved effective with A7. As a step below Imagination’s 6-core PowerVR designs, GX6450 delivers a better balance between performance and power than a larger GPU would, which in turn is clearly a benefit to Apple. On the other hand this means A8 is not going to have the GPU performance to compete with the fastest SoCs specifically designed for tablets, though what this could mean for the obligatory iPad update remains to be seen.

A8’s CPU: What Comes After Cyclone? CPU Performance
Comments Locked


View All Comments

  • dingodingaling - Thursday, October 2, 2014 - link

    1. Is there circumstantial evidence to suggest that AnandTech are Apple Stooges?

    Yes, absolutely. The founder and owner of the site works for Apple and so does one of their long term “reporters”.

    2. Were AnandTech motivated to be Apple Stooges?

    Yes, absolutely.

    AnandTech are a commercial website, funded by advertising dollars. It is in their interests to drive up traffic and views of their site.

    They know that:

    a. iPhone users trawl every site that says anything in advance of an iPhone launch to pick up rumours, and after the launch to view “reviews” that help them avoid cognitive dissonance.
    b. Apple punish negative reviews by withdrawing access to review devices and invitations to events. http://www.cultofmac.com/255618/how-apples-blackli...
    c. Being punished like this will impact them severely from a commercial perspective. They are therefore commercially incentivised to have positive reviews.

    3. What means could they use to be Apple Stooges?

    Given we know they are motivated and incentivised to give Apple products good reviews, we can analyse how they do it. This can be in 5 basic ways:

    a. Use the Apple provided product, not one they brought off the shelf, meaning Apple can tune that one device for tests;
    b. Cherry pick tests where the Apple product will do well, and ignore those that they don’t;
    c. Carefully select the “comparison” devices, ignoring any that make Apple devices look bad;
    d. Make up tests where they think they need to cover up a hole, but ensure that no one else knows how the test works so they can’t repeat it, and
    e. Make “mistakes” occasionally and assume no one will notice.

    4. Is there evidence to suggest they are using these methods?
    a. Use the Apple provided product, not one brought from a store randomly.

    Yes we know they do this, and they admit it. Also looking at the Display tests there is evidence to suggest that the Apple provided product was tuned. They themselves had to admit their suspicions.

    b. Cherry picking tests.

    Yes, they do this.

    3DMark Ice Storm

    “On the synthetic benchmark 3DMark Ice Storm Unlimited, the iPhone 6 Plus scored 16,965. While that's well above the category average of 13,401, it fell below its Android competition. The S5 blew past with a score of 18,204, as did the HTC One M8 (20,640), the LG G3 (17,548), the OnePlus One (18,399) and the Note 3 (18,321)”


    CNet finds similar results



    An example from Arstechnica


    More results at:


    The Stooges select some of their tests (“Basemark X) but not others – why is that??

    c. Carefully select the “comparison” devices, ignoring any that make Apple devices look bad;

    Yes, they do this.

    For the display, the comparisons for max brightness is missing phones for example the Note III which Tom’s hardware measure at 555 Nits. Why the missing phones?? Surely the Stooges have tested these.

    One of the missing phones is the Note IV which displaymate tested and rated as the best Smartphone display – and their testing includes both iBends.

    “The Best Performing Smartphone or Phablet Display that we have ever tested.”



    Suspicious that its missing!

    For contrast where are phones like the Galaxy S5? This comparison shows that the S5 has amazing contrast.http://www.phonearena.com/reviews/Apple-iPhone-6-v...

    d. Make up tests where they think they need to cover up a hole, but ensure that no one else knows how the test works so they can’t repeat it

    Yes, they do this.

    For battery Life their test is simply wacko – the only site that has the iBends winning battery life tests and probably due to the rigged nature of the “web browsing” test.


    Tom’s hardware has also has completely different results to the Stooges for battery life and web-browsing http://www.tomsguide.com/us/iphone-6-battery-life,...

    Phonearena also have ‘web browsing’ tests that show the iBends aren’t great


    Arstechnica also found the Ibend battery life to be average.


    e. Make “mistakes” occasionally and assume no one will notice.

    Yes, they do this.


    The iBend6+ has better measurements than on Tom’s Hardware.

    Geekbench 3

    The iBend6+ has worse scores on Tom’s hardware than the Galaxy S5 and Note III.

    GFXBench 3.0 Manhatten Offscreen

    Stooges are rating the IBend6 and iBend6+ higher than GFXBench themselves have tested it at.



    The measurement for the iBend6+ is different on Tom’s hardware (its better!), but also different for all the other phones.

    More “mistakes” for the colour temperature and gray scale accuracy are completely different in that test as well.

  • vFunct - Thursday, October 2, 2014 - link

    Give it up.

    Your Android phones area garbage. They're fake iPhones. Apple invented the modern smartphone, and all the innovations like Retina displays. Android keeps trying to copy Apple, but they're never as good or as fast.

    Why get a fake iPhone when you can get a real iPhone instead?

    Fake iPhones are terrible. Get a real iPhone instead.
  • KuyaMarkEduard - Sunday, October 5, 2014 - link

    Hello Dingodingaling. Are you certain that this practice of Apple is still existing even today, as we speak?

    ""Apple PR's dirty little secret:

    Summary: Apple PR maintains a blacklist of journalists that it refuses to talk to. This includes any media outlet that posts anything even remotely negative or heaven help you, a rumor.

    Apple's public relations department is notoriously tight-lipped and only responds to a limited subset of the mainstream media, and usually only the outlets that write positive things about its products.

    If you dare to write an unflattering piece about Apple or -- heaven forbid -- post a rumor you're almost guaranteed to lose your access to Apple. I know this firsthand because I'm the poster child of Apple's PR blacklist. (I was part of a precedent-setting legal case with Apple in 2005, which I won on appeal in 2007 -- thanks to the Electronic Frontier Foundation.)

    Say what you will about my work, but I call 'em like I see 'em.

    I write good things about Apple, I write bad things about Apple and I also publish rumors when I believe that they're credible or plausible. I write about things that I find interesting and about topics that will benefit my readers. Sometimes Apple likes what I write other times it doesn't. Apple and I have classic love/hate relationship.

    But one thing's for sure, I'm not an Apple cheerleader. If like reading puff pieces about Apple there are a number of websites and blogs that will gladly oblige. Or heck, just dial up apple.com/pr.

    Case in point: On February 7 when Arun Thampi posted on his blog that Path was sneakily uploading iPhone user's address books to its server -- without permission -- I called and emailed Apple. Apple didn't reply. Then and I blogged about it.

    On February 8 when Dustin Curtis blogged that Apple makes a standard practice of approving apps that upload the entire contents of your iOS address book to developer’s servers I again called and emailed Apple. Apple didn't reply. Then I blogged about it.

    Later. Rinse. Repeat.

    Then I got an idea. Since Apple PR never responds to my voicemails or emails, maybe they'd respond to the guys that do have access. So I contacted several prominent Apple pundits (who shall remain nameless) that are known for their access to Apple (some of whom get replies from Apple "every time") and I asked them to enquire about Apple's stance on enforcing its policy on address book uploads.

    And you know what? None of them would do it.

    (Update: ironically there's a couple of exclusive stories out today about Mac OS 10.8/Mountain Lion which certain members of the Mac Illuminati had access to a week early.)

    Why? They'd probably say that Apple wouldn't comment. But someone's got to ask if they expect Apple to reply. I mean come on! Apple's not going to press release its shady developers that steal your contacts.

    The fact of the matter is that most journos with access to Apple are afraid of losing it. They're afraid of asking the tough questions. They're afraid of getting blacklisted. Like me.

    So then I contacted the Wall Street Journal.

    There's a prominent columnist at WSJ that has lots of access to Apple. Arguably the most access to Apple. Apple loves the Journal. Apple sends controlled leaks to the Journal. Apple gives unreleased product to the Journal. Surely, Apple would have to respond to the Journal. Right?

    Well guess what? Apple spokesman Tom Neumayr replied to AllThings today about the issue of developers stealing your contacts without permission. (More on that later)

    Gee? I wonder why?

    I'll tell you: AllThingsD is a wholly owned subsidiary of Dow Jones & Company Inc., which is a member of The Wall Street Journal’s Digital Network (which includes WSJ.com, MarketWatch, Barron's, and SmartMoney).

    It's simple really. Apple needs the Journal. The Journal doesn't need Apple. And the Journal's not afraid of getting on Apple's PR blacklist -- because it would never happen.

    Other wags with access, but without the clout of the Journal are probably afraid of getting blacklisted if they probe around too much -- or ask the tough questions.

    My point is that if Apple PR actually read blogs and responded to queries from bloggers things like Address-gate might not explode into giant issues that end up in the Wall Street Journal. Apple could have nipped it in the bud a week sooner by simply replying to my email or voicemail with something to the effect of "yes, we're aware of the issue and we're looking into it."

    Instead, Apple makes a conscious point of ignoring certain journalists hoping that unsavory issues like Address-gate blow over and that no one will notice. Well guess what, I'm persistent. And if Apple doesn't reply, I'll contact the people that I know at the Journal -- or my Congressman.

    And before you cry "sour grapes!" consider this. I've been blacklisted by Apple for over 10 years. I never get invited, I never get replies. I'm long over it. This doesn't have anything to do with me. It's about you and your privacy. I called and emailed Apple PR because I care about my (and your) private contacts and I wanted to know why Apple isn't enforcing its own privacy policies.

    If you don't care about developers stealing your contacts, that's fine. But I do.""

    Because if so, with the iBend issue, how can now then we be assured that each and every-time Apple will say a thing about the issue, will always be true, and not fabricated?

    This is indeed, an eye-opener to all the Serious tech-readers out-there…, …Of-course, this does-not include the Fanboys.
  • EricGee - Friday, October 3, 2014 - link

    If both new iPhone editions are virtually the same phone, why doesn't the 6 also have landscape view. Both these devices should have an iPad GUI. Landscape view is the most ergonomic position. Having to switch grips from portrait to landscape increases the chance for accidental drops. Landscape also decreases the likelihood for cumulative trauma which include carpal tunnel and digititis. In short, it's just easier to navigate your phone seamlessly by keeping it in your preferred orientation. Lastly, it's sad that I'm waiting for a jailbreak. My plan is not to jailbreak. But if Apple developers can't be aware to what the mass looks for in functionality, jailbreaking may be the only answer. If you would reference Infinidock, this app allows for multiple icons to be docked. Not just four icons. Making these small additions to the user interface would vastly improve the users experience. If developers aren't willing to make such adjustments, then I suggest allowing open code for Cydia jailbreak developers that will allow for users to freely mod our devices.

    Believe it or not, the ability to freely mod our phones to the users preference is what separates the iPhone and Android users. Hence the jailbreak... Don't patronize your loyal consumers by making such availability this coming S model.
  • James Wimberley - Friday, October 3, 2014 - link

    Anand team: Please consider adding ruggedness tests for mobile devices. Desktops spend their lives in sheltered environments, with a few heatwaves and power cuts and surges. Laptops are occasionally dropped. But tablets and smartphones are constantly being dropped, put in linty pockets, exposed to rain and coffee, bent, scratched, and connected to low-quality chargers. If you a ea SEAL, you need a different order of ruggedness, like the Toughbook. But normal consumer use is still quite deamnding.
  • e34v8 - Friday, October 3, 2014 - link

    I have three main complains here.
    The is the screen resolution. We all know 326 ppi is far, far from enough. I have always been able to see aliasing on iPhone screens (never on my Nexus 5) and aside form that, the image is just not that clear and sharp as on a good 1080p screen. Maybe the 6+ will offer great viewing experience.
    The second complain is size. Yes, I think that Apple is making the right move with bigger screen size. Better late than never. The 4.7" are not the ideal size for me, but this is a major improvement. But wtf were Apple thinking, when they made a 4.7" phone, with the dimensions of a 5 or a 5.2"phone? Why? Apple customers value compactness. Just compare it with an old 4.7" phone like the Optimus G - 131.9mm vs 138.1mm .
    And last, but not least - 1gb of ram. Are Apple so greedy? This is typical planned obsolescence. I still can't believe it. 64 bit SoC and a gig of ram...

    The lack of OIS or wireless charging is also not good.

    This phone should cost a lot less.
  • blackcrayon - Friday, October 3, 2014 - link

    "We all know 326 ppi is far, far from enough" - Stopped reading there. It's enough for the vast majority of people, and is certainly a better idea than making a 500 ppi phone that lags.
  • e34v8 - Saturday, October 4, 2014 - link

    Well, that's your problem. You should read someone's comment, before you reply.

    I agree with you. I also hate phones that lag, stutter or miss frames in the UI. That is the reason I do not like Samsung and their Touch Wiz. But that's not the case with my Nexus 5. It's totally stock, but believe me, it's blazing fast and smooth. And it still manages to have a gorgeous 455 ppi screen.

    Anyway, there is no such thing as a perfect phone, but Apple could have offered much more, having in mind that it's a flagship expensive device.
  • michael2k - Monday, October 6, 2014 - link

    1) Your eyesight is superior, most aren't nearly as good as yours.
    2) The Optimus G is 8.45mm thick vs the 6.9mm of the iPhone 6
    3) You talk about typical planned obsolescence, yet in the same breath rave about the Nexus 5. The Galaxy Nexus from 2011 won't see either the 2013 KitKat nor the 2014 Android L, whilst the 2011 iPhone 4S saw both the 2013 iOS 7 and the 2014 iOS 8. I would expect a 2014 iPhone 6 to see at least iOS 11, and still be usable! What do you think your 2013 Nexus 5 will see?
    4) OIS isn't that important, especially when it already has one of the best cameras in the industry
    5) This phone already sold out. Costing less is probably the last thing it needs. Ordered on Sept 20th and expect to see my phone in October 20th.
  • e34v8 - Thursday, October 9, 2014 - link

    The fact that my eyesight is or isn't superior does not change the facts. Ignoring or denying something, without any arguments is not productive. You say that OIS is not important and that's it. Why? I can tell you, I'm never going to buy a phone without OIS again. It's much easier for me to catch perfectly focused shots and shoot steady videos.

    Anyway, everyone has their own view on things. I think Apple are capable of making a better phone - with thin bezels, full HD screen, more than 1 gig of ram, bigger battery... but they chose not to. I'm not some fanboy that will blindly ignore faults. The iPhone 4 was my last iPhone. I'll continue to wait for something from Apple that's worth the money.

Log in

Don't have an account? Sign up now