A5X chip in Apple's new iPad doubles RAM to 1GB - report

1235

Comments

  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by wizard69 View Post


    I do hope Safari is fixed, it crashes 4 or 5 time an hour for me.

    Note that is on my iPad one. Right now I have iOS 5.1 installed on my iPhone 4 and have a bit of use under the belt. It does look to be very stable, faster and with far less glitches. Hopefully this is a good sign.



    As to memory, it will help Safari a lot but Safari has many a bug that leads to all sorts of performance issues. Especially troublesome are the ones that force page reloads. If they have corrected these issues I could see my data usage going down.



    Wow! 4 or 5 times an hour? What sites are you visiting that result in so many crashes? If it isn't specific web sites, you definitely have defective hardware. Guess you'll just have to upgrade.
  • hirohiro Posts: 2,663member
    Quote:
    Originally Posted by Relic View Post


    Would be nice if there was a developer option to turn on true multitasking though. Sometimes I'm running a web application that gets all screwed up just because I switch over for a sec to change the music.



    Lost progress on web apps is more a function of hardcore memory conservation than multiprocessing. Changing from one tab to another can do the same thing in iOS Safari, and that is not even changing applications. The forced page reloads is a way to guarantee that an inactive TAB has no large memory required footprint. Moving from a native app to another with those kind of problems would indicate less than optimal state preservation by the app developer.



    I'm hoping that at some point the additional memory allows Apple to be just a little less time aggressive at reclaiming "inactive" RAM on iOS devices. I don't know if 1GB is enough to allow that or not though.
  • samabsamab Posts: 1,953member
    A lot of it is easy to figure out, but a few items can't be confirmed until we have a teardown.



    Gorilla Glass 2 is 20% thinner, so Apple took that space (and more) to put in a much bigger battery. A portion of the additional RAM will have to go to the GPU, so we don't know what end users will get in terms of free RAM.



    We don't know which GPU and at what clock speed the new ipad will have. Even if the new ipad just has a 543MP4 --- it doesn't mean that it will be 50% slower graphics than the ipad 2 (i.e. double the number of GPU cores but have to push 4x the pixel count = 50% slower graphics). Maybe Apple increases the clock speed of the GPU.
  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by Hiro View Post


    Lost progress on web apps is more a function of hardcore memory conservation than multiprocessing. Changing from one tab to another can do the same thing in iOS Safari, and that is not even changing applications. The forced page reloads is a way to guarantee that an inactive TAB has no large memory required footprint. Moving from a native app to another with those kind of problems would indicate less than optimal state preservation by the app developer.



    I'm hoping that at some point the additional memory allows Apple to be just a little less time aggressive at reclaiming "inactive" RAM on iOS devices. I don't know if 1GB is enough to allow that or not though.



    How could Apple be less aggressive? Isn't RAM reclaimed only when needed by something else? It seems that this is a factor of resource availability, not an aggressive strategy.
  • apple ][apple ][ Posts: 8,156member
    Quote:
    Originally Posted by Mikeb85 View Post


    Now this comes with a performance penalty, because Java apps don`t run natively, they run in a virtual machine (ie. emulator).



    So you're basically confirming what I claimed, that Java sucks. You're making excuses for poor performance on Android.
  • hirohiro Posts: 2,663member
    Quote:
    Originally Posted by wizard69 View Post


    Depending upon the specifIcs of the improvements to that GPU the actual on screen performance could very well be a wash. We will have to wait and see. Otherwise you are right RAM is very important here as is the GPU.



    Basic UI functionality is pretty non-taxing for a GPU. Until you introduce real-time transparency in elements, not just shadows, the GPU hardly notices. Pushing more pixels to a frame buffer is pretty easy too as that pixel to frame buffer relationship is all dedicated transistors. So doubling the basic processing units for 4x the pixels is a non linear relationship in favor of good GUI operation.



    On the flip side 4x the pixels is far more taxing for advanced graphics and blending when used in things like games or photo/video compositing/editing. So the 2x GPU capabilities probably just got performance to the place Apple was comfortable in going to the hi-res display for those apps.



    And Apple is in a good place because I don't see iPad changing resolutions again for a long time, so increases in component performance will directly turn into user percieved increased device performance. Contrast that against desktops and laptops where the screen sizes/resolution keep increasing to eat up all available GPU performance.
  • hirohiro Posts: 2,663member
    Quote:
    Originally Posted by dfiler View Post


    How could Apple be less aggressive? Isn't RAM reclaimed only when needed by something else? It seems that this is a factor of resource availability, not an aggressive strategy.



    Not in iOS, there are different design philosophies for the embedded device design compared to the full laptop/desktop design. The first directly impacting issue is there is no backing store in the iOS virtual memory (VM) subsystem. So RAM that is left in the inactive queue indefinitely in OS X (as long as the user does not need it) is not left there in iOS. The forced refresh of web pages when moving from tab to tab is a direct reflection of this as when you move from tab A to tab B, tab A's view is/was not maintained in RAM for instant loading upon return to tab A.



    Lots of other examples of why the app switching state saving is the way it is are for the same reasons in the basic VM subsystem. At some point with more RAM available the tradeoff becomes possible to not immediately free the inactive memory. Where Apple will decide that line is exactly I don't know, but I do know for static pages on my original iPad Safari on 5.1 is acting more nicely this way where several days ago I was always seeing full page refreshes no matter what.
  • rogifanrogifan Posts: 10,669member
    Quote:
    Originally Posted by chris v View Post


    God, I hope so. That's the most frustrating thing about the iPad. Quite a few times, I'll enter some text, say in a window at a site like this, switch to another tab to grab a link, & when I switch back, the page refreshes & my text is gone. If I want to write anything at all involved on a message board, I do it in notes, then copy and paste the thing. It's a pain.



    Also, I hope iOS 5.1 does something about Safari's bad crashiness. It's nice that it auto-recovers the links that were open when it crashed, but I have to re-launch it 4 or 5 times a day, it seems like.



    Glad to know I'm not the only one experiencing this. I don't know what was worse, the constant Safari crashes or page refreshes. And this would happen to me whether I had 2 tabs going or 5. It didn't seem to matter. That's why I'm so excited to get 3rd gen iPad next week.
  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by Hiro View Post


    Not in iOS, there are different design philosophies for the embedded device design compared to the full laptop/desktop design. The first directly impacting issue is there is no backing store in the iOS virtual memory (VM) subsystem. So RAM that is left in the inactive queue indefinitely in OS X (as long as the user does not need it) is not left there in iOS. The forced refresh of web pages when moving from tab to tab is a direct reflection of this as when you move from tab A to tab B, tab A's view is/was not maintained in RAM for instant loading upon return to tab A.



    Lots of other examples of why the app switching state saving is the way it is are for the same reasons in the basic VM subsystem. At some point with more RAM available the tradeoff becomes possible to not immediately free the inactive memory. Where Apple will decide that line is exactly I don't know, but I do know for static pages on my original iPad Safari on 5.1 is acting more nicely this way where several days ago I was always seeing full page refreshes no matter what.



    That's a fine description.



    To be specific, we were discussing how safari tabs are lost when memory is reclaimed for other purposes. Observation as a user has lead me to believe that this is not done proactively. If that is the case, it isn't possible for apple to be "less aggressive" about when cached tabs are lost.



    Your experience of not seeing any tabs being cached is surprising. My experience is the opposite. Safari's cache is preserved unless that memory is needed by some other app or tab. I've grown somewhat proficient at predicting when tabs will need to be reloaded. This allows switching between a half filled out web form and another tab... if you're careful about not loading too heavy of a page it is possible to view other pages and not lose that data entry.
  • Quote:
    Originally Posted by wizard69 View Post


    This is where Apple really sucks, they could easily just tell us how much RAM and how fast the cores are. Instead they play games.



    Oh and before anybody says anything it does make a difference, iPad has been RAM starved for far to long.



    Without real multitasking or complex apps, why would you need lots of RAM?
  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by I am a Zither Zather Zuzz View Post


    Without real multitasking or complex apps, why would you need lots of RAM?



    Fast task switching for one. The ability to cache graphic laden web pages. Even trivial video editing can benefit from lots of ram. Gaming. Etc.



    With that said, i think apple has made excellent decisions regarding the trade offs associated with spec-ing mobile devices While perhaps not perfect, the end result has been superior to any other approach the entire rest of the industry has tried.
  • neo42neo42 Posts: 287member
    Quote:
    Originally Posted by rednival View Post


    First off, Android is most definitely Linux based. Google just decided to make it's own Java implementation to run most of Android applications. I think they are probably regretting that. First, it does have performance issues. Especially when you cram it on some of the cheap devices that run Android. Also, since Android went mainstream (maybe even because of Android), Oracle acquired Sun and has aggressively perused anyone with so called "unlicensed" Java implementations. Android Java was actually based on Apache's Java implementation which was already in dispute, so it was a poor, poor decision for Google for that reason alone.



    That said, you can write Android software in C++ or even Objective C. In fact, it is encouraged for games.



    That is truth. That is fact.







    I guess I am an idiot for liking my iPad and having an Android phone I actually like. There are things my phone does that my wife could never do with her iPhone, and my phone is not rooted. The extras I have matter to me and keep me from buying an iPhone. They are things that matter on my phone, but do not matter on my tablet.



    Believe it or not, there are legitimate reasons to choose Android over iOS. I do not believe Apple is an evil empire that destroys your freedom. I am perfectly fine with Apple controlling the experience on their devices. Some times though, it boils down to function over form.



    I hate to say this because I know people will attack me without knowing what they are talking about. This is the reality though. I have over a year with both iOS and Android:



    Apple has a major edge over Android on style and ease of use. No doubt.



    Android has a major edge over Apple on capability and customization. No doubt.



    Android also has malware, background process that can cripple your device, and ad ware. It isn't for everyone. But if you want freedom to customize, your choices are Android or a jailbroken iPhone.



    THAT'S the reality. Decide what matters to you and make your choice.



    Personally, I don't find it to be as easy as most people. I supposed that makes me a moron for not being brainwashed that actually thinking rationally. I need to upgrade my phone soon and I am still leaning towards Android since I now have a Macbook and an iPad. Of course, I need to learn more about jailbreaking. A jailbroken iPhone might be the best of both worlds.



    Did you really think I was being serious? I guess you missed my reply to the other guy who thought I was being serious too.
  • mikeb85mikeb85 Posts: 506member
    Quote:
    Originally Posted by Pendergast View Post


    It does matter, because otherwise the Tegra 3 is just a paper beast.



    Also, the Tegra 3 is only just catching up to the *year-old* iPad 2 and its A5?



    nVidia`s ARM chips suck. Tegra 2 was the worst performing dual-core chipset, Tegra 3 will be the worst quad core. Qualcomm`s S4 (which is coming out in the new HTC smartphones) is much quicker than the Tegra 3 despite only being dual-core.



    Apparently, Huawei`s quad-core K3 chipset is significantly quicker than the Tegra 3, and maybe even the S4 (Huawei are touting it as the world`s quickest smartphone chipset), and then the new Exynos chips will likely be unveiled soon, not to mention the ARM A15, etc...
  • jnjnjnjnjnjn Posts: 588member
    Quote:
    Originally Posted by umrk_lab View Post


    Impressive, indeed especially if we go back to history :



    In 1975, the 80 MHz Cray-1 was announced. Excitement was so high that a bidding war for the first machine broke out between Lawrence Livermore National Laboratory and Los Alamos National Laboratory, the latter eventually winning and receiving serial number 001 in 1976 for a six-month trial. The National Center for Atmospheric Research (NCAR) was Cray Research's first official customer in 1977, paying US$8.86 million ($7.9 million plus $1 million for the disks) for serial number 3. <...>



    The 80 MFLOPS Cray-1 was succeeded in 1982 by the 800 MFLOPS Cray X-MP, the first Cray multi-processing computer. In 1985, the very advanced Cray-2, capable of 1.9 GFLOPS peak performance, succeeded the two first models <...>



    In 1979 I was programming on the TRS-80. It's Z80 had an frequency of 1.77 MHz (I had to look that up, I thought it was about 2.5MHz but if I remember correctly the frequency was updated later on) and no floating point instructions, not even integer multiplication.

    It seems almost impossible to create programs on such an limited device but I remember that I was really impressed by it because program listings (in assembly) of several meters could execute in the blink of an eye.

    It's really incredible what you could do with 29GFLOPS.



    J.
  • jnjnjnjnjnjn Posts: 588member
    Quote:
    Originally Posted by samab View Post


    A lot of it is easy to figure out, but a few items can't be confirmed until we have a teardown.



    Gorilla Glass 2 is 20% thinner, so Apple took that space (and more) to put in a much bigger battery. A portion of the additional RAM will have to go to the GPU, so we don't know what end users will get in terms of free RAM.



    We don't know which GPU and at what clock speed the new ipad will have. Even if the new ipad just has a 543MP4 --- it doesn't mean that it will be 50% slower graphics than the ipad 2 (i.e. double the number of GPU cores but have to push 4x the pixel count = 50% slower graphics). Maybe Apple increases the clock speed of the GPU.



    No they didn't (increase the clock speed of the GPU) and that's because Apple already showed a performance sheet during the keynote. The new quad core GPU is twice as fast as the dual core GPU of the iPad2 so its cores have the same frequecy.



    J.
  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by jnjnjn View Post


    No they didn't (increase the clock speed of the GPU) and that's because Apple already showed a performance sheet during the keynote. The new quad core GPU is twice as fast as the dual core GPU of the iPad2 so its cores have the same frequecy.



    J.



    We should probably still wait for real confirmation of the clock rate.



    Doubling of cores rarely translates into doubling of actual performance. If anything, an exact doubling of performance suggests a slight bump in clock speed.
  • jnjnjnjnjnjn Posts: 588member
    Quote:
    Originally Posted by dfiler View Post


    We should probably still wait for real confirmation of the clock rate.



    Doubling of cores rarely translates into doubling of actual performance. If anything, an exact doubling of performance suggests a slight bump in clock speed.



    In this case it does. The PowerVR 543 is extremely scalable and it's performance is an almost exact multiplication of the number of cores (up to 8).

    It's also unwise to overclock a GPU designed for a specific frequency, especially because that means that it's power usage will increase disproportionaly. History learns that Apple underclocks it's iPad, iPhone and iPod processors to have almost the same performance but a lot less power usage.

    But I am curious about the feature size of the A5X processor. Is it 28nm or 32nm? Did they redesign the CPU and GPU for this feature size or is it just an translation to the new feature size?



    J.
  • dfilerdfiler Posts: 3,420member
    Quote:
    Originally Posted by jnjnjn View Post


    In this case it does. The PowerVR 543 is extremely scalable and it's performance is an almost exact multiplication of the number of cores (up to 8).

    It's also unwise to overclock a GPU designed for a specific frequency, especially because that means that it's power usage will increase disproportionaly. History learns that Apple underclocks it's iPad, iPhone and iPod processors to have almost the same performance but a lot less power usage.

    But I am curious about the feature size of the A5X processor. Is it 28nm or 32nm? Did they redesign the CPU and GPU for this feature size or is it just an translation to the new feature size?



    J.



    Is evidence to illustrate that assertion about an exact doubling of performance with a doubling of cores. Even when hardware has very efficient scheduling, there is still overhead and software considerations.
  • hirohiro Posts: 2,663member
    Quote:
    Originally Posted by dfiler View Post


    Is evidence to illustrate that assertion about an exact doubling of performance with a doubling of cores. Even when hardware has very efficient scheduling, there is still overhead and software considerations.



    Scheduling overhead is a myth. Multi-core schedulers aren't any more complex than non-trivial single core schedulers.



    The place where multi-core does not get the true n-times linear increase in performance has to do with accessing shared OS resources. So the apps do get linear increase in performance for their non-sharing code, but the probability an app will need to wait for access to its own shared data or the OS to be ready to do something when the app makes an API call goes up. As OSes get better designed with more granular division of capabilities and lock-free accesses this also reduces the frequency and length of waits.
  • wizard69wizard69 Posts: 11,726member
    That is on my IPad 1.



    Part of that appears to be due to many bug fixes. I also get the feeling that the GPU driver was improved as scrolling seems to be far smoother. 5.1 looks like a fantastic upgrade.



    As a Side note it looks like a bug creep into mail.



    Quote:
    Originally Posted by Mazda 3s View Post


    Hopefully this will cut down on page refreshing in Safari.



Sign In or Register to comment.