Jump to content
Bohemia Interactive Forums

Brisse

Member
 View Profile  See their activity

  • Content Count

    1195

  • Joined

  • Last visited

  • Medals

  • Medals

Posts posted by Brisse

  1. Posted

    14 hours ago, djotacon said:

    Autodetect settings: 28.5 fps ( The autodect change the object distancie view to 800m.).

     

    Good. Back when I had an FX-8350 it used to set it to 1200 I think, which was more or less unplayable. I always lowered it to 800m manually which makes the game ugly but playable. Sadly, the low object distance is necessary on those CPU's.

     

    Now, with the Ryzen 1700X, autodetect sets it to 1900m and yet the game runs so much better than with the FX on 800m.

    • Like 1

  4. Posted

    Low vs standard

    https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=968ce9dc-0bd1-11e7-9182-0edaf8f81e27

     

    Standard vs. high

    https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=9dc87b16-0bd2-11e7-9182-0edaf8f81e27

     

    High vs. very high

    https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=b431826c-0bd2-11e7-9182-0edaf8f81e27

     

    Very high vs. ultra

    https://cdn.knightlab.com/libs/juxtapose/latest/embed/index.html?uid=c70620be-0bd2-11e7-9182-0edaf8f81e27

    • Like 2

  7. Posted

    6 minutes ago, djotacon said:

     

    That's I dont understand why in the AMD the benefits are nearly zero.

     

    One of my theories is based in the PHYSX + NVIDIA GAMEWORKS based in the sudden fps drops when there's explosions but is only a theory.

     

     

    It has nothing to do with AMD vs. Intel or Nvidia physx/gameworks.

     

    The reason the FX doesn't work well in Arma is it's poor single threaded performance. The FX is best in embarrasingly parallel tasks. In Arma, you are better off with a cheap dual core with strong single threaded performance.

  9. Posted

    28 minutes ago, nedflanders said:

     

    Did you play with HPET and Energy Profile yet?

     

    I did, but I saw no significant difference in Arma 3.

     

    It's interesting to see what happens when you switch between balanced and performance profiles though.

    In balanced, it seems to stack as many threads as possible on as few physical cores as possible, while using logical cores before physical cores so that it can park as many cores as possible. I bet this is great for energy efficiency while not giving up too much performance.

    In performance, there's no core parking and no downclocking. Threads are first and foremost put on it's own physical core before putting them into the extra logical cores only if the physical ones are already saturated. Probably less energy efficient, but faster.

  13. Posted

    @abudabi

     

    Looks like my Windows 10 is properly aware of physical cores, SMT and cache, unlike what people are saying on reddit.

     

    This is what I get from coreinfo:

     

    Quote

    C:\>coreinfo -c

    Coreinfo v3.31 - Dump information on system CPU and memory topology
    Copyright (C) 2008-2014 Mark Russinovich
    Sysinternals - www.sysinternals.com

    Logical to Physical Processor Map:
    **--------------  Physical Processor 0 (Hyperthreaded)
    --**------------  Physical Processor 1 (Hyperthreaded)
    ----**----------  Physical Processor 2 (Hyperthreaded)
    ------**--------  Physical Processor 3 (Hyperthreaded)
    --------**------  Physical Processor 4 (Hyperthreaded)
    ----------**----  Physical Processor 5 (Hyperthreaded)
    ------------**--  Physical Processor 6 (Hyperthreaded)
    --------------**  Physical Processor 7 (Hyperthreaded)

    C:\>coreinfo -l

    Coreinfo v3.31 - Dump information on system CPU and memory topology
    Copyright (C) 2008-2014 Mark Russinovich
    Sysinternals - www.sysinternals.com

    Logical Processor to Cache Map:
    **--------------  Data Cache          0, Level 1,   32 KB, Assoc   8, LineSize  64
    **--------------  Instruction Cache   0, Level 1,   64 KB, Assoc   4, LineSize  64
    **--------------  Unified Cache       0, Level 2,  512 KB, Assoc   8, LineSize  64
    ********--------  Unified Cache       1, Level 3,    8 MB, Assoc  16, LineSize  64
    --**------------  Data Cache          1, Level 1,   32 KB, Assoc   8, LineSize  64
    --**------------  Instruction Cache   1, Level 1,   64 KB, Assoc   4, LineSize  64
    --**------------  Unified Cache       2, Level 2,  512 KB, Assoc   8, LineSize  64
    ----**----------  Data Cache          2, Level 1,   32 KB, Assoc   8, LineSize  64
    ----**----------  Instruction Cache   2, Level 1,   64 KB, Assoc   4, LineSize  64
    ----**----------  Unified Cache       3, Level 2,  512 KB, Assoc   8, LineSize  64
    ------**--------  Data Cache          3, Level 1,   32 KB, Assoc   8, LineSize  64
    ------**--------  Instruction Cache   3, Level 1,   64 KB, Assoc   4, LineSize  64
    ------**--------  Unified Cache       4, Level 2,  512 KB, Assoc   8, LineSize  64
    --------**------  Data Cache          4, Level 1,   32 KB, Assoc   8, LineSize  64
    --------**------  Instruction Cache   4, Level 1,   64 KB, Assoc   4, LineSize  64
    --------**------  Unified Cache       5, Level 2,  512 KB, Assoc   8, LineSize  64
    --------********  Unified Cache       6, Level 3,    8 MB, Assoc  16, LineSize  64
    ----------**----  Data Cache          5, Level 1,   32 KB, Assoc   8, LineSize  64
    ----------**----  Instruction Cache   5, Level 1,   64 KB, Assoc   4, LineSize  64
    ----------**----  Unified Cache       7, Level 2,  512 KB, Assoc   8, LineSize  64
    ------------**--  Data Cache          6, Level 1,   32 KB, Assoc   8, LineSize  64
    ------------**--  Instruction Cache   6, Level 1,   64 KB, Assoc   4, LineSize  64
    ------------**--  Unified Cache       8, Level 2,  512 KB, Assoc   8, LineSize  64
    --------------**  Data Cache          7, Level 1,   32 KB, Assoc   8, LineSize  64
    --------------**  Instruction Cache   7, Level 1,   64 KB, Assoc   4, LineSize  64
    --------------**  Unified Cache       9, Level 2,  512 KB, Assoc   8, LineSize  64

     

    I won't be able to overclock my RAM. I'm stuck on 2133mt because I'm running four dual rank modules. They are only rated for 2400mhz CL14 in their XMP profile anyway so I'm not loosing that much by running four sticks.

  18. Posted

    Don't have options for HPET or SMT in my UEFI. Asus have hidden these settings on this mobo for some reason :S

     

    Tried disabling HPET in Windows though. It's actually disabled by default in Win 10, but I had enabled mine at some point because I needed it for a specific thing I was doing. After that I just left it enabled. I disabled it again with the command "bcdedit /deletevalue useplatformclock " but I saw no difference in performance while running YAAB. Nothing significant anyway. I was like a percent above previous bench but that might have just been random.

     

    I might try some more of dwardens suggestions later when I have time.

  23. Posted

    Ryzen 7 1700X

    PRIME X370-PRO

    4x16GB DDR4-2133mhz 15-15-15-36-51-2T (frequency is limited due to using 4 dual rank modules)

    R9 Fury

     

    YAAB with it's standard settings at 1440p yielded 30fps average. Similar settings on my previous FX-8350 gave around 24fps but I'm not 100% certain the settings and game version is the same so that's not an apples to apples comparison.

     

    I find that YAAB seems to run like crap, but subjectively I have gotten a huge boost during normal gameplay. I ran the auto detect settings and the game set itself to about double the view distances I used to run previously, and it still runs much smoother on the Ryzen than the FX.

     

    I know some games have run better with SMT off so I was interested in testing that, but I actually couldn't find an option for turning SMT off in my UEFI :/

     

    Also read about some issues where Windows parks cores when it shouldn't, and having HPET active has negative performance impact. I didn't bother working around any of these as I hope they will be fixed by software updates. I know proffesional reviewers worked around these issues by disabling HPET and setting Windows power management to high performance.

     

    Edit: Tried to tick "Enable Hyper-Threading" in the launcher. Lost 8% in YAAB, but not sure if it's due to the setting or simply due to the random nature of YAAB.

  25. Posted

    33 minutes ago, Tankbuster said:

    So, great for Windows 10 and the few applications that use a lot of cores then? How useful for us Arma users? Not very, I suspect. And that's a shame because Intels dominance of this market is really unhealthy,

     

    I've got news for you. Ryzen wasn't exclusively designed for Arma. It looks like it's going to be an awesome performer for those of us who actually use our PC's for more than just gaming, while still being perfectly viable for gaming, but of course it won't beat Kaby Lake in gaming.

     

    I expect to build a Ryzen system next week. I might try some Arma benchmarks then, but I don't expect to see any different results from those already posted. The results posted above are perfectly in line with Ryzens single core performance, which is what matters in Arma 3.

×