Jump to content

Ready

Members
  • Posts

    649
  • Joined

  • Last visited

Everything posted by Ready

  1. Thanks! But that looks complicated. Or not? Does this need to be done after every update? Seems to be a lot of work for an improvement that is not clear to me (yet). So far I did not notice any difference. But I am not really accustomed yet to the Apache, so I won't really notice much yet.
  2. Installed the optimized Core, Liveries and Cockpits for A-10C and AH-64D. I checked but only had the optimized cockpit for the A-10C before. Thanks @edmuss for mentioning it. Thanks @Taz1004 @lefuneste01 for the great optimizations.
  3. The bars charts I create are usually about 20 seconds long averages. Can be all kinds of missions. The most recent ones are recorded on our own MP server, where I am in a cockpit of a cold A-10C. I run discord and an OBS recording on the side, which consume also FPS, just to get an idea of what my starting position will be. This for me is a good enough indicator because I just focus on the delta between two settings. This way I can compare two settings every 5 minutes or so. This morning I did about 10 changes in an hour. It takes quite some time closing stuff, rebooting, changing BIOS settings, booting(BIOS posting sometimes can take up to a minute or more with the BIOS changes), starting Windows, VR applications, OBS, DCS (sometimes I have to restart DCS a couple of times to get my eye-tracking to work), run the test, look at the results and adjust settings and deciding the next change to test (I keep a script/log with changes/progress). Once I am happy with the optimizations of all the settings I changed, I will run some stress tests to see the numbers over a longer stretch like 10 minutes or so.
  4. Made a few modifications. BIOS Turbo Game Mode back enabled. PBO Curve -30 Infinity Fabric 2166 UCLK DIV1 Mode UCLK=MEMCLK Validations (to see if still in place) MSI Afterburner (check if auto starts with Windows) Clock: Core +150, Memory +1000, Fan Power Limit 95% HAGS ON instead of HAGS OFF (HAGS OFF reduces about 10FPS) Process Lasso - Main/Probalance OFF Windows Page File - Set manual to fastest SSD. Other drives OFF Windows USB Selective Suspend OFF DCS excluded from Defender (both the file and the process) Disk Space - at least 25% free (to prevent reduced write performance because of decreased size of Dynamic SLC Cache) null
  5. I use Quad View Foveated Rendering in order to be able to use The OpenXR Toolkit, run the OBS Mirror for recording and see the OpenKneeboard inside my recording. It is similar to Varjo Foveated and can also be configured for focus area and peripheral etc. It still works and only has a tiny little bug that sometimes disable eyetracking when starting up. I notice this in the Hangar view and keep restarting until it is gone, which usually takes 1-2 times if it appears. Thanks I will also give it a go
  6. Yes exactly. I use some of the default missions that comes with the airframes when you buy it, easy for reference. I sometimes also shared track files so others could run it as well and then it is easy to compare the experience. For my tuning I just keep the headset on the desk without moving it. I center the view and then it always is the same. The DCS in-game dashboard also shows FPS and FTs and I am also running the OpenXR Toolkit and run an overlay to display these settings. What I like about the overlay is that it is fixed to your POV. The DCS in-game dashboard stays fixed in one place, which is less easy to use when looking around.
  7. Well I just did the opposite I directly went for 5500 and 1.235v and slowly decreased to see where it would become stable. Timings on my Infinity Fabric quickly screw up stability, so there I went down to 2100. There is no default benchmark as far as I know. I have been running track files in the past and then used the same file to try various settings. Did that for about 150 tests or something like that. Wish there was a 3DMark file with DCS or something in it. I now just do the same thing over and over. I recreate the baseline for my usual flying; loading up our Wing's MP server, turn on my apps and go from there. I sometimes just put a plane in a hold over large detailed terrain like Kola. Once I find a good configuration, I only test the increments and compare those.
  8. Further CPU overclocking I added the following changes BIOS - CPU Infinity Fabric 2133Mhz Edit 3-12-2024 Dialed back the Inifinity Fabric to 2100 due to instability in game. PBO Curve -40 Scalar 10x Game Mode ON Process Explorer automatically ON, DCS excluded from first two cores MSI Afterburner - GPU Memory Clock +1.000 Core Clock +150 Fan Power Limit 95 User defined software automatic fan control ON (default curve)
  9. Process Lasso Looked into Process Lasso again. Setting CPU affinity for certain cores exclusive to DCS and the rest of the cores to other applications mainly has impact on the 1% lows. Better to keep Process Lasso turned off. null
  10. Results of my CapFrameX tests comparing various BIOS / overclock settings on my DCS performance. Configuration Same case (open case, no case fans). Same SSD's with same OS, same DCS. New CPU, new CPU Cooler, new System board and new Memory. DCS in VR Varjo Aero ppd 39. DCS System settings: see screenshot. Overclock BIOS (version 0606): EXPO ON, PBO +200, Curve Optimizer -40. Rebar ON. + Ryzen Master settings (see screenshot) Overclock GPU (MSI Afterburner: Core Clock +202, Memory Clock +883, Power limit 85%) improves the 1% percentile FPS by 10. Running on Multiplayer, 132nd Wing server, A-10C, cold and dark parked at Bodo on the Kola Peninsula. OBS recording (using OBS mirror), which consumes about 6-10 fps. (I mostly record my flight so I include this in my benchmarks) Conclusions 9800X3D with AM5 and DDR5 is a huge step up from my 5800X3D. I can not compare one-to-one through CapFrameX, because my previous tests were taken on our previous MP map (PG, no trees) and I heavily increased the visuals in the DCS settings (including secondary shadows and terrain object shadows default). I know the new map consumes more performance (way more trees and textures) and by looking at previous numbers I estimate the bottom line improvements to be roughly 25+ FPS and 25+ in the 1% lows. X3D Turbo Mode must be OFF: it has a negative impact on DCS in VR. Core Performance Boost must be OFF: it has a negative impact on DCS in VR. Additional timings from the Ryzen Master and MSI Afterburner work very well and possibly can get some more tweaking. Further experimentations: FCLK, currently 2100, maybe one or two steps higher. Scalar, now on 5, possible increase to 7 or 10. RAM speed, now at 6000, maybe 6200. nullPushing the GPU higher core clock (250?) and memory clock (1.000) Test system for stability, fly around large missions in Apache.
  11. My memory issue turns out to be very slow testing of the RAM. It takes about a minute for the BIOS to post, after which the memory runs at 6000. The following BIOS options still need to be tested for impact on DCS in VR; 1. Comparing the manual overclock via Ryzen Master (see below video) vs enabling PBO (+200 and -40 curve) 2. Enabling or disabling X3D Turbo Mode (which disables SMT) Overclocked my 9800X3D to 5500Mhz using AMD Ryzen Master, following
  12. CPU Overclocking PBO optimizer ENABLED +200 -30 Next: Following the steps in this video Link to the details: https://skatterbencher.com/2024/11/06/skatterbencher-82-ryzen-7-9800x3d-overclocked-to-5750-mhz/
  13. GPU overclocking Check power limit of 90, +145 core, 1.1v available, memory oc +500 and custom fan curve. (thanks @Reed) All You Need To Know | MSI AFTERBURNER Overclocking & Undervolting Full Walkthrough | MSI Strange curve at my Volt/Freq editor nullnull null
  14. Thanks for the SDI tip. Did not know about it.
×
×
  • Create New...