After losing Apple's iPad business, Intel has bled $7 billion while heavily subsidizing cheap x86 At

15678911»

Comments

  • Reply 201 of 217
    Quote:

    Originally Posted by prw View Post

     

    The major obstacles to Apple using ARM SOCs in laptops are USB 3.0 and Thunderbolt.  Apple didn't provide USB 3.0 until Intel started bundling support into its processor chips.  Even though USB 3.0 hubs are commonly available now, Apple would have to plunk the technology into its ARM SOCs to maintain a similar profile.  I hope they do so.  It takes hours to restore my iPad Air and iPhone 6+ now, since they only support USB 2.0.  Can you imagine a new Apple laptop without USB 3.0 ?

     

    Thunderbolt is a joint Apple and Intel technology, but only Intel is producing controller chips now.  How long before Intel moves that support into the processors ?  Then they could drop the external controller chips, and Apple would face a very expensive fork:  keep Intel processors or design their own controllers, or redesign their processors to include Thunderbolt.  

     

    If Apple overcomes these obstacles, I imagine a new version of Rosetta for Intel X86 instruction streams.  And they would still need to build-in virtual machine support.  

     

    I wonder - if Intel is licensing Atom designs, could Apple acquire a license and add an Atom core or 2 to one of their ARM Ann systems ?  That would be an interesting way to combine Windows virtual box and USB 3.0 support.  


    Actually Apple may not have to use either USB 3.0 or TB.  The reason?  USB 3.1 utilizing the reversible Type C connector.

     

    http://www.anandtech.com/show/8558/displayport-alternate-mode-for-usb-typec-announced

  • Reply 202 of 217
    canukstorm wrote: »
    Actually Apple may not have to use either USB 3.0 or TB.  The reason?  USB 3.1 utilizing the reversible Type C connector.

    http://www.anandtech.com/show/8558/displayport-alternate-mode-for-usb-typec-announced

    1) Remember, it's an mDP port, not a TB port.

    2) I could see Apple adding USB Type-C ports to a lightweight, small, and inexpensive ARM-based notebook but I would still expect them them to so the mDP port interface for displays. Having DP over USB Type-C is great, but it would just mean an adapter when used with displays that support DP, and Apple is already using the mDP ports for their external display.
  • Reply 203 of 217
    relicrelic Posts: 4,735member
    dasanman69 wrote: »
    There's a honest review by someone you'd think would be biased. Android tablets IMO are still lacking big time.

    I feel that way about most devices but I still manage to find a lot of uses for them, faults and all. I just own more than one, using their strengths to best of my ability. In a general consumer sense Apple has definitely won the tablet war and rightfully so, their interface is easy, fluid and decent apps to back the platform up. Coming from the direction of a programmer, tech enthusiast and hacker, iOS can be extremely infuriating at times do to Apple's totalitarianism. Regardless the iPad has definitely secured a place in my collection and is used for a very important thing in my life, music.

    There are very good Android tablets on the market that even the most staunch Apple user wouldn't mind using. The Nvidia Shield being one them, decent build quality, great features, in fact there's really nothing else you could possibly add to the thing that could make it any better. Nvidia K1 32Bit, 2GB Ram, 32GB, 1080P display, LTE, NFC, Mini SD, HDMI port, Miracast, active digitizer pen, fantastic but optional Bluetooth GamePad, solid 8 hours of battery and it just got updated to Android 5(no skin, pure), starting at 300. Than there are the Nvidia goodies, like; Grid, Stream (which actually works quite well), development tools, open boot loader making it easy to install Linux4Tegra or any other Linux for ARM distro, than using the built in HDMI it can be made into a really good desktop computer, a very useable and extremely quick one at that. Under Linux, with GeekBench, I get single-core 1190, multi-core 3960, plenty fast enough for a development desktop computer, especially one with some of the best video drivers for ARM that I have seen yet.
  • Reply 204 of 217
    lantznlantzn Posts: 240member
    Quote:

    Originally Posted by abazigal View Post



    It's articles like this which make me have to pinch myself every now and then to make sure I am not dreaming. What bizarro world is this where the competition has problems even giving their tablets away at cost, while Apple has no problems selling theirs at a healthy profit? (????)?

    A world where a company actually make a product worth paying a premium for.  This happens all the time in many industries.  There are businesses where individuals are making quality items with attentions to detail that people are willing to pay for.  Cheap isn't the best in the longer run.  People are tired of things breaking or not working as advertised.

  • Reply 205 of 217
    gatorguygatorguy Posts: 23,469member
    alfiejr wrote: »

    ah, good of you to cherry-pick some Google PR quotes from 2014 about its Nexus line intentions - it's a testing platform, a prototype. at least, nowadays it is.

    but let's go back a few years to the beginning, shall we?
    OK. From 2010
    http://www.wsj.com/video/allthingsd-at-ces-andy-rubin-interview/BFC2C7A1-0F2C-4846-BC60-FC69F8F622F0.html

    Walt tries to make the same "Google is getting into vertical sales" argument. which of course they didn't and that even Rubin the promoter wouldn't agree with. At 3:40 in Rubin explains why the Nexus devices are needed, same explanation as Google uses now: The only proper way to develop the platform is by everyone in the engineering group focusing on the same phone hardware for the way Android features will operate, each and every day coming to work seeing he same piece of hardware. Focus. Even in the beginning Nexus phones were created for the engineers to test their code, find the bugs, make sure worked worked the way they anticipate. Google supplied software, OEM's supplied hardware.

    Then 9 minutes in Rubin begins talking about Google opening up an on-line store where licensees can offer their Android smartphones already set up with voice and data plans, and not just Nexus phones. HTC's Nexus One was simply intended to be the first. The carrier part of course didn't pan out as planned but Google still created and still maintains an on-line store where stock Android phones and tablets from various OEM's can be purchased. Other than offering carrier plans it's pretty much the way it was envisioned in 2010. Rubin just seems to have had unrealistically high hopes it would be a busier marketplace than it actually is.

    Some relatively small number of people really want that stock Android experience. Other than for the updates or those few developer tweekers I really can't imagine why myself. It shouldn't be any surprise at all that Nexus models are good choice for only a specific segment of the buying public.
  • Reply 206 of 217
    alfiejralfiejr Posts: 1,524member
    Quote:
    Originally Posted by Gatorguy View Post





    Google hasn't historically shown very much interest of getting into smartphone hardware as a business. There was the old Nexus One direct-to-customer sales program years back when there were very few Android devices for the public to experience. But Android was designed first and foremost as a widely-available alternative mobile OS rather than created to run on Google's own hardware. Even before the first Android device ever came to market it was the OEM's coming up with hardware platforms for the OS, not Google creating their own.

     

    just wrong. see above.

    Quote:
    Originally Posted by Gatorguy View Post





    OK. From 2010

    http://www.wsj.com/video/allthingsd-at-ces-andy-rubin-interview/BFC2C7A1-0F2C-4846-BC60-FC69F8F622F0.html



    Walt tries to make the same "Google is getting into vertical sales" argument. which of course they didn't and that even Rubin the promoter wouldn't agree with. At 3:40 in Rubin explains why the Nexus devices are needed, same explanation as Google uses now: The only proper way to develop the platform is by everyone in the engineering group focusing on the same phone hardware for the way Android features will operate, each and every day coming to work seeing he same piece of hardware. Focus. Even in the beginning Nexus phones were created for the engineers to test their code, find the bugs, make sure worked worked the way they anticipate. Google supplied software, OEM's supplied hardware.



    Then 9 minutes in Rubin begins talking about Google opening up an on-line store where licensees can offer their Android smartphones already set up with voice and data plans, and not just Nexus phones. HTC's Nexus One was simply intended to be the first. The carrier part of course didn't pan out as planned but Google still created and still maintains an on-line store where stock Android phones and tablets from various OEM's can be purchased. Other than offering carrier plans it's pretty much the way it was envisioned in 2010. Rubin just seems to have had unrealistically high hopes it would be a busier marketplace than it actually is.



    Some relatively small number of people really want that stock Android experience. Other than for the updates or those few developer tweekers I really can't imagine why myself. It shouldn't be any surprise at all that Nexus models are good choice for only a specific segment of the buying public.

     

    ah come on. it is clear from what Rubin actually did in the years he was in charge of Android that he envisioned the Nexus line as much more than a prototype. and then he bought Motorola! and no, it wasn't just for the patents. it was to potentially become an OEM too.

     

    and it's clear his schemes truely "flopped" and Paige dumped him because of that. that was the history that DED referenced in his off-hand "flop" remark that you took issue with - he wrote many posts about it back during the Rubin era. and it was what stuck in my memory about Nexus too.

     

    i'm done with this argument. you will never budge from your prototype-only interpretation, no matter what the historic facts were.

  • Reply 207 of 217
    dasanman69dasanman69 Posts: 13,002member
    alfiejr wrote: »
    just wrong. see above.

    ah come on. it is clear from what Rubin actually did in the years he was in charge of Android that he envisioned the Nexus line as much more than a prototype. and then he bought Motorola! and no, it wasn't just for the patents. it was to potentially become an OEM too.

    and it's clear his schemes truely "flopped" and Paige dumped him because of that. that was the history that DED referenced in his off-hand "flop" remark that you took issue with - he wrote many posts about it back during the Rubin era. and it was what stuck in my memory about Nexus too.

    i'm done with this argument. you will never budge from your prototype-only interpretation, no matter what the historic facts were.

    Based on the evidence given. You, a opinion article from Time magazine, and GG, statements from Google, and Andy Rubin, who do you think proved their point? Or maybe he was just lying to fool everyone like it's said SJ did quite often.
  • Reply 208 of 217
    The source is deafening. No comments on Nexus beings "hardware reference" when they made the phone 32bit and the tablet 64bit?

    Talk about fragmentation for developers.
  • Reply 209 of 217
    Quote:

    Originally Posted by digitalclips View Post





    Well yes, prior to Apple adopting Intel, if there wasn't a an intel card involved it it was pure emulation and yes it sucked big time. Hence my suggestion that some sort of optional add on Intel processor dongle would be needed for VMs when (and I truly believe this is coming) Macs go native Apple and drop Intel. By the way, what ever happened to RISC?



    I assume you use Bootcamp for games, other than that I see no need for forgoing the side by side convenience of a VM in OS X.



    The R is ARM is for RISC, so it is alive and well today.  On top of that, even Intel's Core series chips uses RISC cores that are under a CISC decoder.

     

    We are both in agreement that we think an ARM mac is coming, our only disagreement is that you believe that Windows compatability should come via a dongle, and I think it should come via a hardware decoder like the Intel Core series CPUs.  The two main reasons why I don't like the dongle concept are 1) that like the old Intel card, you would need to duplicate most of the windows computer on the dongle.  Assuming that you can access the onboard GPU (which is a big assumption) you would still need onboard memory and a separate bus. and 2) Due to a limited market, these dongles will be relatively expensive, if they are made at all.  I believe this to be because of a combination of issue 1) and limited production runs not allowing for economies of scale.

     

    I also think that if Apple did switch to ARM without a solution for Windows compatibility, sales of the Mac would drop because of the need (real or imagined) for running Windows.

     

    I don't use bootcamp for games, but I do have some CPU intensive software for work that only runs on Windows.  I have tried running it in a VM, but the difference in speed justifies the reboot.  Like I said, I am not the average use case.  However if it was not for that software, I agree that the VM would be way more convenient.

  • Reply 210 of 217

    I had an Asus Atom based tablet for about a week before I returned it. It was horrible. Just horrible.

    And I generally like Android tablets.

  • Reply 211 of 217
    relicrelic Posts: 4,735member
    Quote:

    Originally Posted by graxspoo View Post

     

    I had an Asus Atom based tablet for about a week before I returned it. It was horrible. Just horrible.

    And I generally like Android tablets.


     

    I never been in a big fan of the Atom either, though I have to say the Windows 8 tablets run fairly well with them but their also using the quad-core variant. The worst tablet I ever used had an Atom CPU, the Samsung Tab 3 10.1, what a horrible piece of junk. After the system booted, there was only 340mb available for apps because TouchWiz is such bloated crap, Samsung should have been ashamed of themselves for this one. My brother inlaw bought one during an after Christmas sale last year. He thought there was something wrong with it so he asked me to take a look, nope, just a crappy tablet.

  • Reply 212 of 217
    crowleycrowley Posts: 10,453member
    popnfresh wrote: »
    You can't lose what you never had.
    I'll console myself with that next time I lose a race.
  • Reply 213 of 217
    relicrelic Posts: 4,735member
    Quote:
    Originally Posted by AJMonline View Post

     



    The R is ARM is for RISC, so it is alive and well today.  On top of that, even Intel's Core series chips uses RISC cores that are under a CISC decoder.

     

    We are both in agreement that we think an ARM mac is coming, our only disagreement is that you believe that Windows compatability should come via a dongle, and I think it should come via a hardware decoder like the Intel Core series CPUs.  The two main reasons why I don't like the dongle concept are 1) that like the old Intel card, you would need to duplicate most of the windows computer on the dongle.  Assuming that you can access the onboard GPU (which is a big assumption) you would still need onboard memory and a separate bus. and 2) Due to a limited market, these dongles will be relatively expensive, if they are made at all.  I believe this to be because of a combination of issue 1) and limited production runs not allowing for economies of scale.

     

    I also think that if Apple did switch to ARM without a solution for Windows compatibility, sales of the Mac would drop because of the need (real or imagined) for running Windows.

     

    I don't use bootcamp for games, but I do have some CPU intensive software for work that only runs on Windows.  I have tried running it in a VM, but the difference in speed justifies the reboot.  Like I said, I am not the average use case.  However if it was not for that software, I agree that the VM would be way more convenient.


     

    You can emulate x86 on the chip itself, Transmeta did it with their RISC processor and Nvidias new K1 uses similiar technology, binary translation (dynamic recompilation) where it's binary translation layer runs in software, at a lower level than the operating system. In fact the K1 Denver was originally intended to support both ARM and x86 code morphing technology from Transmeta, but was changed to the ARMv8-A 64-bit instruction set because Nvidia could not obtain a license to Intel's patents. You can still access this software layer and even flash the K1 with a custom emulation layer. This is one of the reasons why I like my Nvidia Jetson board so much, I get to experiment with things like x86 emulation on ARM. I currently use Tiny QEMU that only takes up 2.1mb, even has a DirectFB (framebuffer) interface, cool stuff. I've managed to get Windows XP up and running fairly easily, actually works quite well. So Apple could easily design their next SOC to include a x86 translation layer, that is if they were to get access to Intels pantents of course.

  • Reply 214 of 217
    relic wrote: »
    ajmonline wrote: »



    The R is ARM is for RISC, so it is alive and well today. On top of that, even Intel's Core series chips uses RISC cores that are under a CISC decoder.

    We are both in agreement that we think an ARM mac is coming, our only disagreement is that you believe that Windows compatability should come via a dongle, and I think it should come via a hardware decoder like the Intel Core series CPUs. The two main reasons why I don't like the dongle concept are 1) that like the old Intel card, you would need to duplicate most of the windows computer on the dongle. Assuming that you can access the onboard GPU (which is a big assumption) you would still need onboard memory and a separate bus. and 2) Due to a limited market, these dongles will be relatively expensive, if they are made at all. I believe this to be because of a combination of issue 1) and limited production runs not allowing for economies of scale.

    I also think that if Apple did switch to ARM without a solution for Windows compatibility, sales of the Mac would drop because of the need (real or imagined) for running Windows.

    I don't use bootcamp for games, but I do have some CPU intensive software for work that only runs on Windows. I have tried running it in a VM, but the difference in speed justifies the reboot. Like I said, I am not the average use case. However if it was not for that software, I agree that the VM would be way more convenient.

    You can emulate x86 on the chip itself, Transmeta did it with their RISC processor and Nvidias new K1 uses similiar technology, binary translation (dynamic recompilation) where it's binary translation layer runs in software, at a lower level than the operating system. In fact the K1 Denver was originally intended to support both ARM and x86 code morphing technology from Transmeta, but was changed to the ARMv8-A 64-bit instruction set because Nvidia could not obtain a license to Intel's patents. You can still access this software layer and even flash the K1 with a custom emulation layer. This is one of the reasons why I like my Nvidia Jetson board so much, I get to experiment with things like x86 emulation on ARM. I currently use Tiny QEMU that only takes up 2.1mb, even has a DirectFB (framebuffer) interface, cool stuff. I've managed to get Windows XP up and running fairly easily, actually works quite well. So Apple could easily design their next SOC to include a x86 translation layer, that is if they were to get access to Intels pantents of course.

    Damn!

    Now, you've done it!

    I surfed Transmeta, the K1 ...

    Here's something from an ARS writeup:
    Transmeta's Crusoe team did just that. They started over again, but this time instead of asking "how fast can we possibly make this," they asked "how efficient can we possibly make this, and still have it run x86 apps acceptably." Thus, Crusoe's designers were working towards two primary design goals that dictated the decisions and tradeoffs that they made. Transmeta wanted the Crusoe to have:
    • full x86 compatibility
    • the lowest possible power consumption
    • a level of x86 application performance that provides for a reasonably good user experience

    http://archive.arstechnica.com/cpu/1q00/crusoe/m-crusoe-1.html


    And something about the K1
    When it designed Project Denver, NVIDIA chose to step away from the out-of-order execution engine that typifies all modern high-end ARM and x86 processors. In an OoOE design, the CPU itself is responsible for deciding which code should be executed at any given cycle. OoOE chips tend to be much faster than their in-order counterparts, but the additional silicon burns power and takes up die space.

    What NVIDIA has developed is an in-order architecture that relies on a dynamic optimization program (running on one of the two CPUs) to calculate and optimize the most efficient way to execute code. This data is then stored inside a special 128MB buffer of main memory.

    The advantage of decoding and storing the most optimized execution method is that the chip doesn't have to decode the data a second time once needed -- it can simply grab that information from memory. Furthermore, this kind of approach may pay dividends on tablets, where users tend to use a small subset of applications. Once Denver sees you run Facebook or Candy Crush a few times, it's got the code optimized and waiting -- there's no need to keep decoding it for execution over and over.

    http://hothardware.com/News/Nvidias-64bit-Tegra-K1-The-Ghost-of-Transmeta-Rides-Again/


    And, of course, IBM was experimenting with something similar (x86 translation) on its PPC chips ...

    And, of course, there were [are ?] issues about licenses to use the Intel x86 instruction set.


    So, we have the K1 a 64-bit RISC processor with 2 CPUs and a large cache -- with the ability to run ARM RISC instructions quite well ... and presumably run x86 CISC instructions well enough ... And what makes this possible is an extra CPU core doing a dos-y-dos and allemande-left with the other core and the large cache ...


    An extra CPU core and a large cache ... An extra CPU core and a large cache ... where have I heard that before???


    1000


    http://www.anandtech.com/show/8562/chipworks-a8


    Edit: Oops wrong image -- will update!


    1000


    http://www.macrumors.com/2014/11/12/a8x-graphics-8-cores/
  • Reply 215 of 217
    I got exactly this far:

    "This subsidized, clearance sale product dumping makes it quite incredible that Apple can sell any iPads at all, let alone remain the world's leading tablet vendor"

    This statement seems to be made without irony, simply stated as though it's a true headscratcher. What about these shitty things puts Apple in any danger?

    We aren't talking about Galaxy or Nexus-quality pads here. We're talking about bargain basement junk. The kind of crap that banks give away when you open a new account. I'm not insulting the Atom (I use Atom-powered boards to run tty-only Linux machines for a retrocomputing lab) but I am insulting the large majority of products that Atoms go in. Chintzy garbage.

    That's the problem with the Android market overall. There's a lot of Android machines out there! But most of them are crap. El Goog and Samsung will sell a good number of their products, which are equal in quality to an iPad, it's just for every Nexus 10, there's 67 "Wulong 487JLD91 10" LCD Touchtab Tablet Computers" sold in India for $95 each.

    Let's not forget that the Next Billion won't be running Lollipop or iOS 8, and they'll prob be not-running them on hardware that any of us would scoff at.
  • Reply 216 of 217
    relicrelic Posts: 4,735member

    "The advantage of decoding and storing the most optimized execution method is that the chip doesn't have to decode the data a second time once needed -- it can simply grab that information from memory. Furthermore, this kind of approach may pay dividends on tablets, where users tend to use a small subset of applications. Once Denver sees you run Facebook or Candy Crush a few times, it's got the code optimized and waiting -- there's no need to keep decoding it for execution over and over."

     

    This explains why I experienced lag on the first day of using my Nexus 9. To further experiment, I opened and closed 10 of my most used applications 5 times each and then rebooted, everyone one of them opened instantaneously. I then installed a new app, Photoshop Touch, opened it up for the first time, took 11 seconds, that's a long time, closed it and opened it 5 more times, than rebooted. It opened up in 2 seconds, huge improvement, I chose Photoshop Touch because it's a fairly large app with lots of functionality. So even after rebooting the optimized instruction sets stay in cache, that's cool. 

     

    Nvidia is going to introduce a new application that will allow a person to manipulate this cache, it will be part off their new Denver development board application collection, geesh say that 10 times in a row. You will be allowed to set which instruction sets take priority or even store certain code permanently. This is going to be very beneficial to my mini VM Linux server. I created an ARM Linux Distro that will launch a Tiny QEMU VM for each of the tasks that I need completed. Instead of just using one Linux install with all my software packed in, my VM server will launch a separate Linux distro that's specifically modified to handle each task, containing just those libraries and applications to make it happen, I need a Media encoder, a separate VM will start, a Node Cluster, another VM will start. This way everything is easily managed and resources controlled. I already have a version of this running on my K1 32Bit development board and it works great, the Denver will hopefully just speeds things up a little, I'm also hoping for a 4GB RAM version too.

     

    I'm really having a lot of fun with these ARM development boards, there isn't much you can't do with one. Even if you looking for an inexpensive desktop computer, at 190 bucks, you can't go wrong. My little Jetson is defiantly fast enough to run whatever you can throw at it, including Doom 3 with everything on high, 1080P, 40 - 50 FPS. I know it's an older game but still looks darn impressive if you asked me.

     

    I'm also highly anticipating MIPS 64Bit and will defiantly grab a development board once it becomes available. In the mean time I have requested one of Imagination's new 32Bit MIPS boards called the MIPS® Creator CI20. You can't buy one just yet but you can get one for free by submitting a request to them, include your project synopsis, I've already received a notification that my proposal would be excepted, so just patiently waiting for my board, doooottt, ddoooott, dddooooo, is it here yet, damn, I hate waiting for stuff.

  • Reply 217 of 217
    Nice to be able to afford to lose Billion$ of $. The bad press portends a good omen, Intel is up 3.5% on the bad news.
Sign In or Register to comment.