https://bitsum.com/when-hyper-threading-hurts/ As far as I understand it, Process Lasso only prevents applications from using the virtual HT cores, but it does not disable Hyperthreading altogether. I guess this can yield in slightly better performance, but as Thomas already said, the result should be negligible in R3E.
Yes and that´s better as you want to be able to decide case by case, My big issue now after spending the week testing is the freezes in MP when other players are joining or leaving, especially in practice sessions, obviously. @Thomas Jansen any ideas on what causes that? I was thinking Windows Defender so I moved to AVG but it seems more or less the same.
There can be two parts to that, when a new car model/livery joins the server, your game will have to load those from your (hopefully SSD ). But it also has to download a tiny part of the model from our servers to decrypt it. So even if you have a blazing fast SSD, it might be the downloaded part that is causing the stutter, I imagine those servers might be affected by high player counts as well. So I don't think you'll be able to solve those stutters with any CPU tricks
For all having issues on older AMD Radeon, such as RX 6700XT (running with AMD 5600 on B550): 1) The 25.11.1 adrenalin driver causes stutter. I went back to 25.10.1 after clean uninstall 2) Do not use the Adrenalin meter overlay as it causes a little stutter 3) Use Vulkan with VSync on This gives me the best perfomance with no stutter at all.
Since moving to a 9070XT, I can't use DXVK anymore as it is (regardless of bundled version 2.4, or later 2.6.1, 2.7.1). Tested both 25.11.1 (it's from a later branch than the rdna2 driver, despite same name), and older 25.9.1 (5800X3D, W10). This is because DXVK + GPL shader caching is now partially broken with this card, so after starting R3E, the first session will always show a longer black screen due to initial shader compilation required to even show the track, and then exhibit occasional large stutters after start and during L1 on subsequent compilation (something that shouldn't even happen on dxvk >=2.4 on amd cards making use of VK_EXT_graphics_pipeline_library), same for new content initially stuttering in later sessions. But the main problem is that after exiting and entering the game, even loading the previously ran content will suffer large stutters again shortly into a session (despite a slightly quicker, but still abnormally long, first compile screen), as if most of previous shader cache doesn't persist. The only workaround I've found, is to set dxvk.enableGraphicsPipelineLibrary = False via a raceroom racing experience\Game\dxvk.conf file. This will return to pre-2.4 behavior on amd (as when we used dxvk before it came to R3E), so dxvk creates its own RRRE64.dxvk-cache file in the \Game folder rather than letting the AMD driver manage it. The downside, is that it'll also disable background shader compilation, so you'll initially get small stutters on L1 when loading content that hasn't been seen before. But after that, they'll remain properly cached for the next time you run R3E or similar sessions, and be free of hiccups. OTH, DX9 now runs decently, way better than the 6700XT ever did, without the previous brutal fame pacing issue when draw calls spike (p.e., long tight corners into wide scenery, T1 grid bunch ups, etc). It still remains more prone to occasional dips than dxvk, though. But it'll look as bad and subpar vs dxvk, without transparency MSAA on textures set to AT/A2C, so 2d horizon textures, fences, some track markings, and trees will be very aliased at the edges. Adding FXAA on top disguises it a bit, but image clarity will suffer. The perf drop is huge too, as in my fixed reference test, it falls from 234 to 163fps (dxvk, dx9) due to greater CPU overhead, starving the GPU. This specific problem wasn't present on a 6700XT where I ran R3E on 25.9.1 for the last months, or even 25.11.1 for rdna2 (non current rdna3/4 branch) for a couple of weeks. Note that I avoid overlays over dxvk. VK_EXT_graphics_pipeline_library (GPL) was working as it should (async compilation and properly driver managed shader cache) since last year. You can ignore this post if using rdna2 cards or other unaffected cards.
I have the same issue ...just a black screen after loading. I just upgraded my card to a 9060 XT... also installed the adrenaline v 26.1.1 (by default). I thought the game was freezing and would have to restart my machine. I tried finding the dxvk.conf file, no luck,... not in the game file
@L1A1 If game stopped working after moving to an rdna4 (permanent black screen after loading a session), try launching R3E in dx9 mode and in graphics settings make sure to disable Motion blur (without it, you should be able to keep MSAA enabled). Should solve your black screen issue when loading to track in vanilla dx9. If instead you mean a temporary black screen due to dxvk (broken) GPL shader compilation which will last for some seconds until image comes back, yet fail to properly cache all shaders so it'll stutter badly ingame, that's entirely another matter. In answer to your remark, dxvk.conf is a just plain text file one has to create to manually config dxvk.
@L1A1 Btw, if you see a permanent cyan square under your mouse cursor only by the right and bottom areas of your screen, then set menu resolution the same as your desktop (i.e, problem will show on a 2560x1440 monitor if menu resolution is left at default 1920x1080, so you'll see that square when the mouse cursor goes beyond 1920x1080 coords with 0,0 as top left). That way, the square will only show when the cursor first appears before you move your mouse (as when pausing in-session, no matter where cursor is on screen), but not anymore permanently on menus as you move it over bottom+right areas. Will also show briefly if you leave your mouse over the splash screen when launching the game. This has been a bug with rdna3+4 cards + R3E, and sadly the latest CEF update wasn't enough to solve it. :shrug:
Hi, @AlleyViper and @Thomas Jansen in particular. Since a few weeks ago, maybe coinciding with tthe Q1 update, not 100% sure, I have had stuttering when loading the game in Vulkan mode. This can easily be seen in the steam monitor as there is always a big brown line constantly going up and down, indicating something is off as the line should be flat when the FPS is stable. Previously it was the opposite and I used this method as it was less taxing on GPU than DX9, even only on MSAAX8. I have gone back to the DX9 standard version and no issues so I was wondering if the Vulkan files got corrupted. I copied more recent ones from Github but the game looked all washed out and magenta so I went back to an older one, which seems to have done the trick but I need to test it a little bit longer. Is there an optimal version for Nvidia cards? I see the game standard one is from 2024.
I've reported recently in other thread why rdna4/3 cards will stutter and "miscache" in RE3 (i.e, in dxvk + dx9 + GPL but not rdna2) due to an LLPC compiler issue, but have no idea about current issues on NV as I don't own any modern one. On an old 2070S I still maintain, dxvk can go from ok+faster to stuttery at half fps in the next session, but I presume it's a vram management issue as it doesn't support SAM/ReBAR. Also, when testing alternative dxvk versions, I recommend you only copy tar.gz'd d3d9.dll to your vanilla Game\x64 (non-dxvk folder) and then launch the regular version. It's easier this way to not end up messing the original dxvk 2.4 version (as it for some reason includes a matching dxgi.dll). For your PM, I'll reply here: the proposed fix on that github link is to set a flag in dxvk to present your card as a (iirc) 6700XT to R3E. This flag should work on any modern dxvk release, so as for the game's own 2.4 dll. This can be currently done by creating a dxvk.conf file under game\dxvk.conf set with 'd3d9.hideNvidiaGpu = true' (note: must be placed in Game\ not on Game\x64 or Game\x64dxvk). You can also temporarily add 'dxvk.hud = fps' as a check if it's properly loaded, so an fps counter will show. If it works, have in mind that since dxvk 2.6.2, amd GPU's wont' have access anymore to nv render hacks, so you might lose transparency msaa if your card will also behave as such to dxvk after that change. Due to the "gpu swap", graphics settings ingame might need redone. GL edit: Sorry, can't remember how to do proper inline code in this forum, as '+' or `+` ``+`` ```+``` doesn't seem to work.
I renamed the 2 old DXVK files in game\x64dxvk so I can reuse them if needed. I want to try 2.7.1 to see if it uses less VRAM (long standing issued with DXVK) so I will add that line to my config file as you suggest and test it. Thanks.
Addiing the line to the config file and updating the dlls worked, no more stutter - for now. I can even turn shadows on again.
Cool, did it solve the other mentioned blue hue issue on most recent dxvk versions for your card too?
Yes, that was fixed with the command line, i believe. The stutter cam back later so I switched to Vulkan mode from originall and it went away again but there is clearly some underlying issue. Now I know that switching modes and deleting the vulkan dll files then reinstalling them is a workaround, though.