Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
AMD Intel Hardware

AMD Beema and Mullins Low Power 2014 APUs Tested, Faster Than Bay Trail 66

MojoKid (1002251) writes "AMD has just announced their upcoming mainstream, low-power APUs (Accelerated Processing Units), codenames Beema and Mullins. These APUs are the successors to last year's Temash and Kabini APUs, which powered an array of small form factor and mobile platforms. Beema and Mullins are based on the same piece of silicon, but will target different market segments. Beema is the mainstream part that will find its way into affordable notebook, small form factor systems, and mobile devices. Mullins, however, is a much lower-power derivative, designed for tablets and convertible systems. They are full SoCs with on-die memory controllers, PCI Express, SATA, and USB connectivity, and a host of other IO blocks. AMD is announcing four Beema-based mainstream APUs today, with TDPs ranging from 10W – 15W. There are three Mullins-based products being announced, two quad-cores and a dual-core. The top of the line-up is the A10 Micro-6700T. It's a quad-core chip, with a max clock speed of 2.2GHz, 2MB of L2, and a TDP of only 4.5W. In the benchmarks, the A10-6700T quad core is actually able to surpass Intel's Bay Trail Atom platform pretty easily across a number of tests, especially gaming and graphics."
This discussion has been archived. No new comments can be posted.

AMD Beema and Mullins Low Power 2014 APUs Tested, Faster Than Bay Trail

Comments Filter:
  • by LWATCDR ( 28044 ) on Tuesday April 29, 2014 @09:30AM (#46867303) Homepage Journal

    This would be great for NAS if they make motherboards with a large number of SATA ports.
    Yes I know you can add a card but that drives up the costs and complexity.

    • by TheRaven64 ( 641858 ) on Tuesday April 29, 2014 @09:34AM (#46867347) Journal
      Seconded. I have an AMD E-350 in my NAS. It has 4 SATA ports (+1 eSATA), so that gives one optical drive and 3 drive bays, but leaves one of the drive bays empty and unusable. I went with AMD over Intel because Intel crippled (not sure if it still does) its Atom boards to 2 SATA slots, making them unusable for RAID-Z.
      • You could use all 4 SATA ports for HDDs, and install using a USB optical drive or stick when needed (which won't be very often..)

        • This is the route I'm probably going to go with my next PC. These days, even games bought on disc don't usually require the disc just to play the game (maybe the one and only upside of moving on to Internet-based DRM?), so it's mostly just used to install a game or rip a DVD/CD and then forgotten for a few months. A USB drive gets bonus points for letting me put the DVD drive wherever I want on my desk, rather than having to bend over to reach where the case is hidden away. They make them quite small too...

        • That's one RAID 5 with a hot spare, one RAID 10, or two RAID 1 arrays (or one spanning volume with mirroring, but you may as well use RAID 10). For a real storage server you're kind of limited.

          Six is a much better number. It would allow two fives, a ten and a one, a ten and a zero, a zero or one for the boot drive and a five with a hot spare, a single boot drive and a ten with a hot spare, two fives in a network boot or always USB booting, a six-disk ten with network/USB boot, and on and on.

      • Perhaps Intel branded boards are that way, but I have a Jetway mini-ITX NAS/server system running a D525 Atom with 6 SATA ports, which is perfect for one SSD (OS/boot), one DVD/CD, and then 4 RAIDed spinning disks.

    • There are a few motherboards in the segment already, which typically use LSI 2308 SAS controllers to provide 8 SAS/SATA (6Gb/s) channels. Some even support ECC RAM.

    • That would depend on the southbridge, not the CPU.

      The southbridges that have 6 or more 6Gbps SATA ports are:

      SB850/820M/920/950 (6)

      A50M/A60M/A70M (6)

      A75/A78 (6)

      A85X/A88X (8)

      A55E/A77E (6)

      • Really? What part of SoC with SATA controller did you miss?

        Quoting the story:

        These latest AMD APUs aren’t strictly CPU and GPU cores crammed onto a single piece of silicon. They are full SoCs with on-die memory controllers, PCI Express, SATA, and USB connectivity, and a host of other interface elements. Connect one of these APUs to some memory and storage, and some I/O ports and you’ve essentially got a complete low-power, X86 compatible platform, with modern Radeon graphics.

        Read more: http://ho []

      • by cdrudge ( 68377 )

        That would depend on the southbridge, not the CPU.

        These are full SoC. CPU, GPU, memory controllers, PCI Express, SATA, USB, etc are all on one chip. Attach IO ports, memory slots, and storage connectors and you got yourself a complete system.

    • by ncc74656 ( 45571 ) *

      This would be great for NAS if they make motherboards with a large number of SATA ports.

      This. In my home server, I have an A4 on one of these [], which has six SATA ports. It's probably about as fast as the Core 2 Duo that was in it previously, it uses not much power at all (though probably still more than these new chips), and I think I didn't spend much more than $70 or so for the CPU and motherboard. I'm currently using one port for the boot drive and three for the data drives (JBOD with Greyhole). That

    • Asus P8H61 rev 2, mini-itx has 6 sata ports and USB 3. Have one sitting on my desk in a Cooler Master 110 case.
  • by rossdee ( 243626 ) on Tuesday April 29, 2014 @09:32AM (#46867319)

    Beema and Mullins
    successors to last year's Temash and Kabini

    I'm waiting for Darmok and Jalad

  • by edxwelch ( 600979 ) on Tuesday April 29, 2014 @09:49AM (#46867457)

    It's quite impressive that even though it's still 28nm, the GPU has 38% reduction in leakage current. That comes from improvements in the 28nm process. You would only get those sort of improvements by moving to a new node.

    • You can get improvements on the same lithography size by using different materials, different dopants, different dopant concentrations, different circuit geometry, different layer thicknesses, different voltages, improving the overall process, finer clock and power gating, faster dynamic voltage switching, etc.

      Shrinking circuitry is not the only way to reduce both static and dynamic power. It is merely the most commonly known factor since it has a 30+ years history of making transistors faster, cheaper, mor

  • Huh. (Score:4, Interesting)

    by serviscope_minor ( 664417 ) on Tuesday April 29, 2014 @09:53AM (#46867499) Journal

    Well, those benchmark results are pretty interesting.

    It seems to be pretty close to a low end Core i3 on a fair number of the tests which is very impressive, and the TDP is pretty good as well. It seems very competeive all of a sudden for low end laptops even considering the power draw which has been AMDs main weakness recently.

    Of course, it completely pwn3s the intel stuff at graphics as one might expect, but it is surprisingly respectable in the CPU department.

    Disappointingly it doesn't have HSA support. That would be cool, though I can see why they didn't bother for this iteration: not much out there can really make good use of HSA.

    • Except that when HSA apps come around, people will probably regret the decreased battery life that will come with the fact that you won't be able to run the expensive parts on hardware with lower control circuit density. I think HSA does have its uses in the low-power part of the computer market.
      • I think HSA does have its uses in the low-power part of the computer market.

        Oh yes sure, but it's going to take a while for people to write for it with the new desktop chips. Maybe there'll be enough software by the time the next gen mobile ones come out. I agree though that it's a great way to get lots of compute in a mobile power envelope.

    • Although it is indeed impressive and very competitive, the Bay Trail Atom Z3770 it bests is half a year old. One would expect one of the recently announced refresh Intel parts to perform as good as or better than the Z3770 on the CPU limited benchmarks: []

      When looking at GPU-dependent benchmarks, Intel is definitely hurting their necks while looking up to their competition and probably will be for some time.

      All in all: This is good news for the consumers.

  • Also with ARM (Score:5, Informative)

    by gnalre ( 323830 ) on Tuesday April 29, 2014 @10:23AM (#46867757)

    The summary missed the most interesting part of this chip in that it contains a ARM core running trusted zones Trusted Execution Environment.

    This makes them quite interesting for highly secure applications such as industrial embedded controllers

  • These APUs embed an ARM Cortex-A5 'PSP' for trusted computing.

    I'm wondering whether this arm core is utilized with respect to running Android apps, via AMD's collaboration with Bluestacks.

    A Cortex A5 is no speed demon compared to the quad-core beasts powering your smartphone. But it would only have to execute dalvik bytecode for an app, with an amd64 chip running the Windows host OS.

    (AMD haven't indicated they want to enter the Android market but this might give them an advantage over Bay Trail tablets, whi

    • Re:Bluestacks? (Score:5, Insightful)

      by amorsen ( 7485 ) <> on Tuesday April 29, 2014 @11:12AM (#46868293)

      In other words, it has a separate processor running proprietary firmware with full access to system memory. What could possibly go wrong?

      • by Khyber ( 864651 )

        The same thing that could wrong at most any other level of the architecture, or in software, or in firmware...

    • A Cortex A5 is no speed demon compared to the quad-core beasts powering your smartphone. But it would only have to execute dalvik bytecode for an app, with an amd64 chip running the Windows host OS.

      Why would it have to run the Dalvik bytecode? The entire point of Dalvik bytecode is that it's architecture agnostic. The ARM core would only have to run the binary code accessed via JNI, and given how much faster the AMD core is than an A5, it's likely to be faster to just emulate that. Unfortunately, people tend to only use native code on Android for things that are really performance critical...

      • Android builds for x86 also. Technically there is no need to emulate ARM or have embedded ARM hardware to run android. The android toolchain will compile native code to both x86 and ARM in the same APK. Android will decide which version of the native code to link at load time.
        • That's assuming that the app vendor actually does ship x86 binaries. Most only ship ARM, a few ship ARM and MIPS, very few also ship x86.
          • The SDK doesn't allow making "fat binaries"? One would sort of expect such capability these days.
          • The only application developers who need to bother with binary packaging are those who decide to use the Native Development Kit.

            All other developers who stick to the Android Development Kit's recommendations to NOT use the the NDK are just Java bytecode that gets recompiled to whatever the device needs either when the APK is initially installed (ART in Android 4.4+) or on-the-fly (JIT/Dalvik) as the application launches/runs.

    • by gnalre ( 323830 )

      The chip is specifically for security. It runs a embedded secure environment that the main processor can use to verify executables before they are run. It has nothing to do with android apart from the fact the same technology can be used to secure mobile devices(stop your phone being rooted etc)

      Being part of the main processor means it should be harder to break into unlike the intel TPM solution which requires a separate off-chip device

    • But it would only have to execute dalvik bytecode for an app

      Why would you want to use a low-performance ARM core to run translated Dalvik bytecode instead of using a high-performance AMD64 core to run differently translated Dalvik bytecode?

  • by Bryan Ischo ( 893 ) * on Tuesday April 29, 2014 @12:19PM (#46869031) Homepage

    The reason I say this is that the performance of Intel instruction set architectures has already reached a point of diminishing returns, and the rate of progress has slowed considerably. Intel's own parts are no longer doubling in performance every two years; they're improving marginally every two years instead (I'll throw up a completely made up number that is only very roughly accurate for the purposes of this post - let's say 20%).

    The amount of money necessary to drive substantial improvements in desktop processor technology no longer meets the amount of money available in the market to pay for such improvements:

    - The desktop processor market share is shrinking
    - For the vast majority of use cases, CPUs from several years ago were already "good enough"
    - We are nearing the end of the "easy" node shrinks, and possibly nearing the end of the "possible" node shrinks

    All of these combined means that there just isn't enough money in the market to drive significant performance increases anymore.

    The amount of money that AMD has to spend to get closer and closer to parity with Intel is less than the amount of money that Intel has to spend to stay ahead of Intel; thus, AMD will eventually catch up.

    The release of Beema and Mullins are evidence of this.

    How long will it take AMD to catch up? My guess is 2 - 3 years more years.

    At that point, Intel will no longer be able to easily have any competitive advantage over AMD because it would cost them far too much to move significantly ahead of AMD. Intel will be forced to gut its margins to stay competitive.

    That's my prediction.

    • If HSA succeeds, there might come a point in time in a few years when Intel and AMD will both be forced to use virtually identical manufacturing technology for their major chips (because nobody will have the financial resources or technological incentive to shrink further), only AMD will have by that time developed the HSA ecosystem to such extent that many applications will be significantly faster on AMD chips than on Intel chips because the wide-issue Intel units will have neither the spatial density nor
  • by default luser ( 529332 ) on Tuesday April 29, 2014 @12:45PM (#46869303) Journal

    For an "AMAZING" product like this is supposed to be, those are some awfully curated results right there. The product is only compared with a Core i3 (outdated Ivy Bridge) when it's actually going to win the test, and there's a plethora of multi-threaded tests that nobody on earth actually uses to put AMD's quad-core in the best light. Also, there are NO BATTERY LIFE TESTS to speak of, just "trust us" quoted TDP figures, and no pricing information.

    And while it is MUCH faster at games than Bay Trail, it's not fast enough to play ANY modern games, even on the lowest setting possible []. This leaves it firmly parked somewhere between tablet and ultraportable processing capabilities, so there's the question about product positioning.

    Call me back when AMD is willing to let reviewers just have at it. If your product does not suck, then it does not need to be coddled.

    • by Khyber ( 864651 )

      Most modern games are so shittily optimized that it's no surprise those GPUs aren't working well.

      • There is a huge gulf between the quality of PC games and tablet games due to the massive performance/storage differences, even in the realm of integrated graphics.

        That said, most modern PC games can be played on HD 4000 at 720p low/med settings. I know this because I play them on my HTPC with frame rates are at around the 30-50 fps mark. They play even better on AMD Trinity/Richland/Kaveri.

        The problem with Kabini et-al is they castrated the GPU to just 128 shaders and 8 texels per-clock, compared to 384/5

  • "especially gaming and graphics." I think the author is confused. Their current 100 or so watt A10's are exceptional at gaming plus the underlying processing. With 4.5 watts in this new one, you can watch an HD movie just fine or do other pure graphics operations but run Starcraft 2's AI scripts on the CPU portion? Oh HELL no!

    Let's take X Rebirth for example. You can run it on moderate graphics hardware but you need a beastly CPU, at least an original i3 minimum (so passmark rating of 1800+) just to
  • since I'm a gamer. Their cpu may not be the best but their graphics development clearly shows how well integrated it is in their cpu and far more superior than intel. For the average user, it might not be a big deal but it's more future proof because of that power.

The relative importance of files depends on their cost in terms of the human effort needed to regenerate them. -- T.A. Dolotta