Announcement

Collapse
No announcement yet.

Early Benchmarks Of The Intel Core i7 7700K On Linux

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

  • Early Benchmarks Of The Intel Core i7 7700K On Linux

    Phoronix: Early Benchmarks Of The Intel Core i7 7700K On Linux

    For those curious how Intel's flagship Core i7 7700K "Kabylake" processor is performing under Linux, my sample arrived yesterday and I've begun putting the CPU through its paces...

    http://www.phoronix.com/scan.php?pag...rly-Linux-Data

  • #2
    A bunch of those results between the 7600K and 7700K really show how the wider execution width in Skylake & Kabylake is put to the best use in applications that can take advantage of hyperthreading.

    Comment


    • #3
      Whenever you see two CPUs with the same amount of cores and the one with HT performs better this is due to stalls in the CPU (branch mispredicitons and cache misses), which is the sole purpose of HT in the first place. Execution width has little to do with it.

      Comment


      • #4
        Originally posted by efikkan View Post
        Whenever you see two CPUs with the same amount of cores and the one with HT performs better this is due to stalls in the CPU (branch mispredicitons and cache misses), which is the sole purpose of HT in the first place. Execution width has little to do with it.
        That's flat out wrong. Execution width is what provides all the resources on the CPU core in the first place that are unused and therefore available for hyperthreading.

        For all the slobbering that goes on around here about POWERPC, you think that people would actually understand that IBM's newer POWER chips are *extremely* wide (noticeably wider than Skylake) and they employ up to 8-way SMT on those cores or else they actually don't perform very well considering how huge & complex they are.

        Comment


        • #5
          ^ I think you are both correct. For a chip with a long narrow pipeline and small cache (ahem, many intel desktop chips, although less so in recent generations), HT serves to ameliorate branch mis-predictions and cache misses. But in a big wide fat chip like IBM POWER, using SMT4 and SMT8 designs brings big performance gains via parallelism.

          Comment


          • #6
            Need 6700k comparisons to make anything of this.

            Comment


            • #7
              Originally posted by rabcor View Post
              Need 6700k comparisons to make anything of this.
              I don't have a 6700K but as said in the article, many more CPUs being added to this comparison for its featured review.
              Michael Larabel
              http://www.michaellarabel.com/

              Comment


              • #8
                Originally posted by chuckula View Post
                That's flat out wrong. Execution width is what provides all the resources on the CPU core in the first place that are unused and therefore available for hyperthreading.
                You are mistaken. The issue width required for a core without HT and for one with HT is exactly the same. Are you mixing it up with instruction decode?

                Comment


                • #9
                  We have found that all the iterations of HT since Sandy Bridge don't scale well. Physical cores scale linearly on a calculated slope. Logical cores are a bell curve where the slope can increase unpredictably. We rarely count them in CPU buys.

                  Comment


                  • #10
                    Originally posted by edwaleni View Post
                    We have found that all the iterations of HT since Sandy Bridge don't scale well. Physical cores scale linearly on a calculated slope. Logical cores are a bell curve where the slope can increase unpredictably. We rarely count them in CPU buys.
                    Well, HT can only help maximize the utilization of a core, meaning a core with HT may perform closer to the theoretical maximum for a single core, never above it.

                    -----

                    I would love to see a comparison with i7-6800K and i7-6900K, especially considering how pricy i7-7700K is over i5-7600K.

                    Comment

                    Working...
                    X