Announcement

Collapse
No announcement yet.

26-Way Intel/AMD CPU System Comparison With Ubuntu 16.10 + Linux 4.10 Kernel

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

  • 26-Way Intel/AMD CPU System Comparison With Ubuntu 16.10 + Linux 4.10 Kernel

    Phoronix: 26-Way Intel/AMD CPU System Comparison With Ubuntu 16.10 + Linux 4.10 Kernel

    In preparation for Intel Kaby Lake socketed CPU benchmark results soon on Phoronix, the past number of days I have been re-tested many of the systems in our benchmark server room for comparing to the performance of the new Kaby Lake hardware. For those wanting to see how existing Intel and AMD systems compare when using Ubuntu 16.10 x86_64 and the latest Linux 4.10 Git kernel, here are those benchmarks ahead of our Kaby Lake Linux CPU reviews.

    http://www.phoronix.com/vr.php?view=23994

  • #2
    In all honesty, those old AMD CPUs are not as bad as they are projected to be. We also keep forgetting that Intel has had a process node / technology (finfet) advantage for quite a long time now. Most users may not even notice a difference between the two under normal day to day tasks. Games are a different story, but with AMD's push of Vulkan and their gaming consoles, they're also slowly warming up to the idea of multi-core rendering. But I can't wait for Ryzen though. Especially because it will mean more competition in the high-end computing space. More competition will lead to more awesome technologies and better prices, and that's gonna be good for us consumers. All in all, 2017 seems to be very exciting! Thanks for the round-up!

    Comment


    • #3
      The system with the i5-3470 clearly has some kind of problem here, since this CPU can't be so much slower than the i5-2500K: Both have four cores without Hyperthreading. The Sandy Bridge i5-2500K is running at 3.3 GHz with 1/2/3/4 turbo steps, while the Ivy Bridge i5-3470 is running at 3.2 GHz with 2/3/4/4 turbo steps. Maybe you want to look into this issue further...

      Comment


      • #4
        Originally posted by arunbupathy View Post
        In all honesty, those old AMD CPUs are not as bad as they are projected to be. We also keep forgetting that Intel has had a process node / technology (finfet) advantage for quite a long time now. Most users may not even notice a difference between the two under normal day to day tasks. Games are a different story, but with AMD's push of Vulkan and their gaming consoles, they're also slowly warming up to the idea of multi-core rendering. But I can't wait for Ryzen though. Especially because it will mean more competition in the high-end computing space. More competition will lead to more awesome technologies and better prices, and that's gonna be good for us consumers. All in all, 2017 seems to be very exciting! Thanks for the round-up!
        No kidding, I really expected the FX 83xx processors to be worse. They really held their own though overall. I really would like to see some OpenGL vs Vulkan on Linux tests varying the CPUs and keeping the video card the same.

        Comment


        • #5
          AMD's FX series tends to perform much worse in Windows, and is where it really got it's reputation hurt. Many Windows applications aren't optimized for the module system AMD used. Remember - it's based on their Opteron chips, which are focused on servers. The architecture was never bad but it wasn't designed with consumers in mind.

          Comment


          • #6
            Originally posted by schmidtbag View Post
            AMD's FX series tends to perform much worse in Windows, and is where it really got it's reputation hurt. Many Windows applications aren't optimized for the module system AMD used. Remember - it's based on their Opteron chips, which are focused on servers. The architecture was never bad but it wasn't designed with consumers in mind.
            Yeah, I have to agree with you here. In fact I personally believe all AMD has to do to take the performance crown overall is to add a third integer unit per pipeline. They really needed 22nm and then 14nm a long time ago. CMT type architectures are superior for sure when considering performance scaling per thread. What often happens is developers create a main thrad and then create a helper thread to split off functions from the main loop. Often times it can help lower latency or improve responsiveness, but doesn't realy translate into a higher performance. It really is a better architecture, but AMD scewed it up by calling the actual core a module and the integer pipelines a core. So it's because of that people don't consider -HOW- most apps are multithreaded, just how many threads it was written to spawn.
            Last edited by duby229; 01-08-2017, 12:46 PM.

            Comment


            • #7
              I find the Darktable results bizarre. It makes no sense to me why the i5-6500 and i5-6600k have half the performance of the slower i5-4670. It's almost like Darktable is picking the wrong code path on those CPUs.

              Comment


              • #8
                Originally posted by arunbupathy View Post
                In all honesty, those old AMD CPUs are not as bad as they are projected to be.
                I bought an Athlon X4 845 (Carrizo, 3.5GHz). Compared to my haswell (locked to 3.5GHz and 4 cores), its scimark scores are ~15-20% lower, but it takes more than twice as long
                (36vs 16minutes) to build a current wine git snapshot in a ramdisk. haswell has 32gb ddr3l-1600, carrizo has 32gb ddr3-2133. So the amd core itself might not be that bad, but combined with caches and memory controller it has no chance against intel.
                Last edited by mlau; 01-08-2017, 01:13 PM.

                Comment


                • #9
                  would be really nice if you could do some gaming benchmarks betweeen i5, i7 and AMD FX cpus by using some higher end nvidia GPU

                  Comment


                  • #10
                    Originally posted by mlau View Post

                    I bought an Athlon X4 845 (Carrizo, 3.5GHz). Compared to my haswell (locked to 3.5GHz and 4 cores), its scimark scores are ~15-20% lower, but it takes more than twice as long
                    (36vs 16minutes) to build a current wine git snapshot in a ramdisk. haswell has 32gb ddr3l-1600, carrizo has 32gb ddr3-2133. So the amd core itself might not be that bad, but intel's memory controller and caches are far far far superior.
                    Yes, Also a very good point. It's not so much the memory controller as it is the caches. Although AMD's newer Zen Architecture uses a totally new cache derived from Samsungs fabrication process. I really hope it finally resolves the difference between Intel and AMD's cache system performance once and for all.

                    Comment

                    Working...
                    X