Announcement

Collapse
No announcement yet.

Core i7 8700K vs. Ryzen 7 1800X For NVIDIA/Radeon Linux Gaming

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

  • #41
    Here's what I find particularly interesting:
    In games where the GPU is bottlenecked, Ryzen seems to be a nudge ahead of Intel, whereas in Windows it's usually a tiny bit behind.
    In games where the CPU is bottlenecked, Ryzen seems oddly far behind, which isn't so much the case in Windows.

    I'm curious to see how different the results would be when testing different schedulers, because for some of these tests (like DXMD) it doesn't really make sense why Ryzen performed so much worse with the 1080Ti.

    Comment


    • #42
      If anyone wants to compare their CPU against Ryzen, i have a geekbench result here: https://browser.geekbench.com/v4/cpu/4471803

      it's really good measure for basic information. For Gaming, Singlethread result matters, if you cpu pulls out more points then you will have higher framerates.

      Comment


      • #43
        Originally posted by xpander View Post
        If anyone wants to compare their CPU against Ryzen, i have a geekbench result here: https://browser.geekbench.com/v4/cpu/4471803

        it's really good measure for basic information. For Gaming, Singlethread result matters, if you cpu pulls out more points then you will have higher framerates.
        Yup, here's mine on an OC-ed Ryzen 7 1700 https://browser.geekbench.com/v4/cpu/4508642

        NOTE: they don't show RAM speeds, what was yours? I'm at 3000 as 3200 makes the system freeze up at times for some reason. Nice to see the 1700x isn't that far from the 1700 (OCed)

        Comment


        • #44
          Originally posted by Almindor View Post

          Yup, here's mine on an OC-ed Ryzen 7 1700 https://browser.geekbench.com/v4/cpu/4508642

          NOTE: they don't show RAM speeds, what was yours? I'm at 3000 as 3200 makes the system freeze up at times for some reason. Nice to see the 1700x isn't that far from the 1700 (OCed)
          since i use 4x8GB sticks i cant get more than 2800mhz currently, but i have it on CL14, which makes small difference, with 2 sticks i can get 3200 easily but i'm lazy to go swap them again to run geekbench

          Comment


          • #45
            Originally posted by xpander View Post
            its because opengl games aren't threaded and singlecore perf counts, so higher frequency will win + intel has slight advantage on IPC when it comes to clock per clock.
            My ryzen R7 1700X @3.9ghz performs same as Haswell i7 4790K @4,4ghz on singlethreaded scenarios. When it comes to multithreading things change quite a bit
            That doesn't make much sense, count int small IPC advantage i7 have (say 5 to 10%) and count in clock advantage (even with assumption djdoo made about auto-overclocking from UEFI) of 1GHz (4000MHz vs 5000MHz), that's still 25% plus 10%, about 35% gaining over 60% of advantage??? That doesn't make much sense to me, even if we assume that 1800x didn't turbo at all (was at 3,6GHz), and 8700 did turbo to 5GHz, it's ~40% plus 10%, meaning 50% advantage at best.

            Speaking strictly by architecture, in worst case scenario, you could make up those 64% simply by using ~35-40% clock/IPC advantage, and giving rest to the iner-comunication difference. But even that doesn't explain it, sicne test were done where only one CCX is used and there's no meaningful difference in performance.

            So yeah, while clocks/IPC/CCX can explain part of it, something else is going on here for such a big difference, and it is probably in the applications (games).

            Comment


            • #46
              Ryzen should only be used with 3200 14T or better if you want to use it for gaming, like for instance gskill sells amd optimized sets https://www.gskill.com/en/product/f4-3200c14d-16gfx ram with these properties is the best memory for amd. But even in the best case, intel is still tens of percents faster for gaming, where single core speed matters most.
              I agree with other posters that michael needs to say the exact memory type used/and how its configured, with game benchmarks, for them to be relevant.
              Last edited by pheldens; 17 October 2017, 12:21 PM.

              Comment


              • #47
                Originally posted by leipero View Post
                So yeah, while clocks/IPC/CCX can explain part of it, something else is going on here for such a big difference, and it is probably in the applications (games).
                Shitty DirectX -> OpenGL wrapper ports that have high single threaded bottlenecks.

                Comment


                • #48
                  Originally posted by bosjc View Post

                  Shitty DirectX -> OpenGL wrapper ports that have high single threaded bottlenecks.
                  Most likely, good point.

                  Comment


                  • #49
                    for relevant comparison compare 8 core vs 8 core or 6 core vs 6 core

                    Comment

                    Working...
                    X