Announcement

Collapse
No announcement yet.

How Intel Laptop Performance & Efficiency Evolved From Nehalem To Broadwell

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

  • How Intel Laptop Performance & Efficiency Evolved From Nehalem To Broadwell

    Phoronix: How Intel Laptop Performance & Efficiency Evolved From Nehalem To Broadwell

    Last week I published a 7-way Linux laptop comparison with processors ranging from Sandy Bridge to Broadwell. Out of interest from readers in an even larger comparison, I've re-tested a Nehlaem-based "Clarksfield" laptop as well as a "Westmere" laptop to show how the raw performance and performance-per-Watt compare to the Sandy Bridge, Ivy Bridge, Haswell, and Broadwell devices.

    http://www.phoronix.com/vr.php?view=22589

  • #2
    One-page viewing is powerful.

    I find these older/budget laptop comparisons really useful for the performance-minded who can't afford the more expensive laptops.

    Comment


    • #3
      I found it all very difficult to parse. You have m and u cpus. Then you have i3, i5 and i7 all the while your benchmarks are multithreaded! So I can only compare:
      i3 m to i3 m
      i3 u to i3 u
      i5 m to i5 m
      i5 u to i5 u
      i7 m to i7 m
      i7 u to i7 u

      Comment


      • #4
        As is always the case, Michael generates lots of data but little analysis. If ALL you care about is running one of these particular programs, the data dump is helpful, but otherwise the lack of analysis makes the whole exercise pointless.
        A first obvious variable is single vs multi-threaded operation; a second is the extent to which AVX is useful (I'm guessing AVX and FMA are driving the obvious jumps in the FFTW performance). Knowing these sorts of things (exactly WHAT is a benchmark exercising) is essential to make a benchmark anything more than a dick-measuring exercise because it is ONLY that information that allows you to extrapolate from the benchmark to whatever code is of interest, whether it's 3rd party code you want to run or 1st party code you want to optimize.

        Comment


        • #5
          These results don't really surprise me all that much. The Core i series was relatively crappy until Sandy Bridge came along, and at which point AMD's revenue plummeted. When it comes to Intel, there's hardly ever a good reason to get a (dual core) i7 for a laptop, because as these tests show, the i5s hold up fairly well with a significant price drop. Considering the limited uses of laptops, the extra instructions and features of i7 tend to go to waste too. The only advantage i5 has over i3 is the turbo boost, but also at the cost of battery life.

          All that being said, if you want a high performance laptop, get an i5. If you want decent battery life and a good value, get an i3.

          Comment


          • #6
            Originally posted by tigerroast View Post
            One-page viewing is powerful.
            Is there a way to switch to one-page view? Multi-page view on a sites like these is autistic: you still have to scroll individual page and click transition to other pages; serve no purpose other than to annoy viewer whatsoever.

            Comment


            • #7
              Originally posted by magika View Post
              Is there a way to switch to one-page view? Multi-page view on a sites like these is autistic: you still have to scroll individual page and click transition to other pages; serve no purpose other than to annoy viewer whatsoever.
              Donate and become a Premium Member.

              Comment


              • #8
                Originally posted by schmidtbag View Post
                When it comes to Intel, there's hardly ever a good reason to get a (dual core) i7 for a laptop
                I don't agree: I do compile alot with my laptop and I find HT very useful.
                ## VGA ##
                AMD: X1950XTX, HD3870, HD5870
                Intel: GMA45, HD3000 (Core i5 2500K)

                Comment


                • #9
                  Thing is - the i5 and i3 chips have HT enabled as well.

                  And there has to be something wrong with the i5-520m - the i5 should run at 2.4-2.9 GHz, the i7-720qm at 1.6-2.8. Yet the i7 is twice as fast in the single-threaded FLAC and Stockfisch tests? Makes absolutely no sense.
                  Last edited by VikingGe; 12-21-2015, 05:50 AM.

                  Comment


                  • #10
                    Originally posted by darkbasic View Post

                    I don't agree: I do compile alot with my laptop and I find HT very useful.
                    Then you should get the quad-core i7, it is slower per thread, but is much faster for compiling.

                    Or, if you just want 2 cores with HT, the 2 cores with HT i5. The dual core i7, is basically a scam.
                    Last edited by carewolf; 12-21-2015, 06:27 AM.

                    Comment

                    Working...
                    X