You can manually swap graphics_options.xml in Documents\My games\SimBin\RaceRoom Racing Experience\UserData\
@Thomas Jansen Thank you very much for this guide. Maybe you could add that with 9070 XT graphic cards either AA or motion blur must be deactivated otherwise it will result in a black screen after the loading screen.
I can too in the game, I was just reporting that the game doesn't adapt the auto-preset when I change the game resolution.
On my system (AMD 5600 + 6700XT), with vsync on in AMD driver, I still have slight frame drops every couple of seconds. Low frame rates in general. When using DXVK, frame rates are much higher, but there is no FPS cap. The setting in AMD driver does not work. Any advice?
You can use DXVK's own fps cap. Create a dxvk.conf file in "raceroom racing experience\Game" (not in \x64 or \x64dxvk subfolders), and on it: d3d9.maxFrameRate = 90 (90 is just an example, of course) I also set: d3d9.maxFrameLatency = 1 # dxvk.allowFse = True # dxvk.hud = fps You can comment out a line with # at the start, as in the above example for the dxvk's own fps counter (useful to check if dxvk is really running) and fse mode. dxvk.allowFse = True is only available from dxvk v2.5.3 (changelog) to restore pre-v2.5.2 behavior for better compat with VRR, so it's not needed on v2.4 (R3E's included dll). Also, for a cap to work on AMD settings, try setting it instead as a chill cap in the game's profile with same min/max values, not as FRTC in global settings. It'll then work when in session (not in main menu), I can confirm it on my 6700XT + dxvk. Also, with AMD it's preferable to use Chill with min=max to work as a fps cap in a game's profile rather than enabling global FRTC, because it introduces less latency in the chain than the later (the downside is slightly worse precision, and non working on menus).
Thanks a lot, mate! This helped! I just have some minor micro stutter. Did you leave all the other options in AMD driver off? Did you get reHUD running?
My driver options are untouched atm, as I run R3E on DXVK. Under DX9 I always get microstutter no matter what (cap, no cap), and some more noticeable stutters vs AI on first corners when they get entangled. Always had them since changing to a RDN2 card (also 6700X), it was previously fine on a RX480. OTH, on the 6700XT I don't notice microstutters under DXVK (nor large stutters on T1), and avg fps are also improved greatly over DX9. Radeon settings for R3E are left untouched (no chill, also), while the cap is set on dxvk.conf with a value I can easily keep constant and inside the VRR range of the monitor, vsync enabled ingame (but should only apply to avoid tearing on occasional framepace issues, given the dxvk fps cap is well bellow max Freesync limit). My cpu is a 5800X3D, so bear in mind that it might be doing a better job taming the 1% lows. Can't remember the last time I tried rehud, but if it happens to fail under dxvk, you might have to set Borderless instead of FS. In case of trouble while setting ingame display options under dxvk, change it before in DX9 mode, or edit directly Documents\My Games\SimBin\RaceRoom Racing Experience\UserData\graphics_options.xml without R3E running: <windowMode type="int32">1</windowMode> is FS <windowMode type="int32">2</windowMode> is Borderless Windowed (edit: ^ ignore suggestion above, see Maskerader's post bellow) Just in case, try also to start rehud after only after the game has started. GL!
DXVK default behavior is different than normal game, so even in Borderless window it still behaves like Fullscreen and ReHUD doesn't show up. You need to switch the game to a resolution that is smaller than your desktop, then the game will behave like a window. Some people just set up a custom resolution that is 1-2px smaller than native. I also think there is a setting for dxvk.conf to change this "forced fullscreen" behavior, but I don't remember it and never tried.
That might be an nv vs amd thing, p.e, if they prefer dxgi or not. On my AMD system, even with dxvk.allowFse = True on >=2.5.3 I still get instant alt+tab switching as with modern FS dxgi games, and last time I tried not even a month ago, sealhud (and the official demo hud) rendered fine without any workarounds. Didn't check ReHud, though. But yesterday, when I tried R3E fresh on another system with a 2070S to check for an unrelated silent crash issue, under vanilla dxvk v2.4 I faced slow alt-tabs (delay, black screen in between), as if it was using legacy dx11 FSE presentation. And that could be making it harder for overlays to work. Edit: @Maskerader , just in case, checked it further on the 2070S with FS set ingame, 581.57 on W10 Spoiler: Fast alt+tab testing vs. DXVK version 2070S, DXVK 2.4 (R3E's), nvcp present mode Auto: - dxvk.allowFse = True / or none set (default) -> slow alt+tab - dxvk.allowFse = False -> slow alt+tab This setting likely isn't recognized before 2.5.3, so it always forces Fse. A few notes: - Forcing "Prefer layered DXGI swapchain" for the vulkan presentation method in NVCP made no difference. - Repeating tests with "Borderless Windowed" set ingame instead of Fullscreen also made no difference, always resulting in a slow alt+tab, so it seems ignored for DXVK's preference. - Lastly, re-checked vanilla 2.4 on the 6700XT. Can confirm it behaves different to the NV card. It'll fast alt+tab in/out, so it seems to differ in presentation mode. 2070S, DXVK 2.6.1 (d3d9.dll placed on \x64, same as main 6700XT system), nvcp present mode Auto: - dxvk.allowFse = True -> fast alt+tab || that was unexpected! - dxvk.allowFse = False / or none set (default since 2.5.3) -> fast alt+tab Even forcing present mode to "Prefer Native" allowed fast alt+tab. Tl;dr, for users having such issues with ReHud or overlays, it might be worth trying a dxvk version from 2.5.3 to 2.6.1 due to the aforementioned change in 2.5.2, as Fse became disabled by default to ensure better compat with additional gdi windows. I'd avoid 2.5.2 specifically, as I can't even launch R3E with it on a 6700XT (the 2070S still can), and also because "dxvk.allowFse" can only be manually toggled on later versions.
The setting I'm talking about had something about "fullscreen" in its name, it wasn't allowFSE (unless allowFSE is one of severall possible names for the same setting).
Donno, can't find a setting with that name documented, or in past modifications. For the reason "Prefer layered DXGI swapchain" wasn't working on the tests above, it seems it needs extra flags on nv inspector to apply for dxvk, but that's moot on >2.5.2 anyway if the intent is just to avoid FSE. Ops, I'll stop the OT. Missed that this ain't the other dedicated dxvk thread, but the latest settings guide.
Thomas, regarding CPU core use, if we were to use process lasso, how many cores/threads are the optimum to put the game on? Is it better if they are physical cores, performance or efficiency ones and should hyperthreading be on or off? Also, to avoid micro freezes in MP when players join a session are there any game or window processes that could be moved to other cores/given priority etc and what resource hogging processes could be moved to other cores?
@DaReaLDeviL made a very nice post about this stuff: https://forum.kw-studios.com/index....oblem-with-the-openxr-menu.20804/#post-258695 I can add that the game uses two major threads, one for physics and one for rendering. VR uses an extra thread for some VR specific rendering stuff. Then there are a couple of things here and there that are done on separate threads, but they use very little resources. So giving the game room for 5 threads should be optimal to prevent any halting, fewer should be possible, but I'm not sure how the game would distribute the threads in that case, so it might not work that well.
Thanks I did read that one hence my questions for the expert, is it better to separate the main 64 exe from the weblauncher.exe on different cores or does it not matter? Also, hyperthreading on or off or doesn´t matter?
I'd say giving the main exe some cores completely for itself will give you the biggest improvement, because stutters happen when the render thread is migrating between cores or waiting for something else to finish. All the other threads are pretty low utilisation. Hyperthreading off is probably better in theory for RR, as we don't use many threads, but I wouldn't cripple your system just for RR
It is possible to set hyperthreading for each physical core, so I am thinking of 3 performance cores only for RRE EXE with HPT off and then moving everything else to the other 3 performance cores and all the efiiciencty cores too with HT on. Is that a plan? Defender needs to be on a separate core/s too but I am not sure if it matters whether P or E. Do you know? I´ve already seen improvements using the AMD tips from the OP but there are still some lows that are causing freezes, especially when players join a lobby.
Sorry, I should have written on a program to program basis so have it off for RRE but on for all other apps.
Process Lasso does indeed give the option to have H/T on or off on case to case basis e.g.set for each program. https://www.reddit.com/r/macgaming/comments/gmpqj5/guide_double_your_fps_with_process_lasso/