Announcement

Collapse
No announcement yet.

ODROID-XU's ARM big.LITTLE A7 + A15 Octa-Core

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by Popolon View Post
    Do not agree at all, Cortex A15 is already very efficient, at least more than any x86 or amd64 (in regards of computation power bu energy used), Cortex a7 is still more, but also far less powerfull.
    Do you have a reference for your assertion that it is more efficient than any x86?
    Anandtech ran some benchmarks not too long ago and haswell and the Apple A7 seemed to be the most efficient followed by the newer snapdragon. The A15 is pretty damn inefficient from everything i've read, but is a necessary step to higher ipc.
    OTOH, big.LITTLE, while seeming hacky, seems to work pretty well from following lwn's coverage.
    My preference, however, is the way qualcomm went about things.

    Comment


    • #17
      Originally posted by liam View Post
      Do you have a reference for your assertion that it is more efficient than any x86?
      Anandtech ran some benchmarks not too long ago and haswell and the Apple A7 seemed to be the most efficient followed by the newer snapdragon. The A15 is pretty damn inefficient from everything i've read, but is a necessary step to higher ipc.
      OTOH, big.LITTLE, while seeming hacky, seems to work pretty well from following lwn's coverage.
      My preference, however, is the way qualcomm went about things.
      You can't take AnandTech benches too seriously. He runs stuff like SunSpider so right off the bat you have to chuckle.

      Plus he takes practical Intel ad copy and makes an article out of it; e.g., all the pre-launch Haswell hype.

      Comment


      • #18
        Originally posted by johnc View Post
        You can't take AnandTech benches too seriously. He runs stuff like SunSpider so right off the bat you have to chuckle.

        Plus he takes practical Intel ad copy and makes an article out of it; e.g., all the pre-launch Haswell hype.
        I don't know of more professionally performed hardware tests. They seem like they are always at the forefront.
        They admit sunspider is pretty worthless since everyone optimizes for it so I don't know why they still use it.
        I've long heard the claim that AT shills for Intel but never seen any evidence.
        The thing is, anyone is welcome to refute his claim with their own results.

        Comment


        • #19
          Originally posted by intellivision View Post
          Do you have any proof for this claim?
          If not, then it's not factual and therefore FUD and I won't have a word of it.
          Dude, your reading comprehension needs updating. I specifically said "it's possible", and included a smiley. Hint: that means I don't know for sure for this chip.

          I have come across chips before where you cannot power-gate one part off without that part's driver.

          Comment


          • #20
            Here is a public source: Imagination's SGX kernel driver.

            https://android.googlesource.com/ker...sgx/sgxpower.c

            Using this info you could write your own driver to turn the thing off, if you really wanted to.

            Comment

            Working...
            X