MirceaKitsune

Oh, one more thing now that the main issue has been resolved and this thread can hopefully be a bit generic. When enabling r_useEntityScissors some ents will be cut off across visportals and not only. You can see it with a volume light here and the skybox leaving trails, sometimes the tops of guards helmets disappear into the void instead. Is this known and can anything be done to improve it? Are entity scissors possible to optimize to work for everyone in general? I took note of this one as for me it greatly improves FPS, by more than 10 in some places! It feels sad to let it go to waste. I think the other critique was it might decrease performance if too many entities are present... wonder if a triage can be made to avoid that like making it adaptive, to be able to default it as on with maximum efficiency. Also what was the driver issue with bindless textures and them blocking multidraw from working? IIRC multidraw was also a huge improvement as the GPU could work with copies of the same data. Is there a setting to re-enable bindless textures in case your driver does support them?

Manjaro seems to have gcc-12 so it's nice to know I compiled and tested with that. Wonder how much the compiler usually makes a difference in performance. Also curious how much running TDM on Wayland over X11 will help with any extra FPS (not expecting more). KDE / Plasma has been slow to become stable on WL but it seems in recent months they finally got there for the most part: If my drawing tablet is also detected by Gimp and power saving works and all, I plan to give it another try soon. IIRC the engine now uses SDL which should already be Wayland native so any potential improvements should be supported.

And now I'm concerned if I wasted everyone's time and should feel bad. Did another "svn update" to 10212, completely deleted the build directory, recompiled with the -DCMAKE_BUILD_TYPE="Release" flag: Now performance is back to normal levels, loading times are also lightning fast by comparison. I don't know if the new rev fixed something or "make clean" wasn't enough or that release flag makes such a difference... if it was local all along sorry for that But yes, everything is working like before for me now that this is solved. No performance improvement on AMD Linux... the issue lowenz reported may have been specific to Windows AMD after all, my low FPS may just be the best that is possible. Glad I can at least confirm everything seems to be well for everyone now.

I'd be surprised too, but just to be extra sure I tried it: Repacked tdm_base01.pk4 with the glprogs directory replaced by the SVN generated one. Same thing with that, whether I also keep the same glprogs folder in the core TDM directory or remove it. I compile the engine without any special parameters and used "make clean" first. My scripts in a nutshell: mkdir foo/bar/engine/build cd foo/bar/engine/build cmake foo/bar/engine make

Yes: Tried both with and without it, same result.

Retested all shadow setting mixtures as well as "r_shadows 0" with full engine restarts in between, same thing.

There is a small and simple feature I thought to suggest: In the Entity menu (n key by default), would it make sense to have a button to export the selected entity to a custom def? This would make it easier to create a common customized ent on your map, without needing to manually do so in a text editor. Practical example: You want to have a custom city watch used multiple times. First you add an atdm:ai_citywatch to the map. Then set your custom stuff like skin, def_head, def_vocal_set. There would be a button labeled Export or similar, which when clicked prompts you for a file name and custom entity name. This will create or override the entity atdm:my_citywatch in file def/my_file.def which starts with "inherit atdm:ai_citywatch" followed by your custom spawnargs.

Sorry about that, will do this next. I just tested every r_skip one by one, zero difference. Even setting r_fboResolution to 0.25 doesn't increase it by a single FPS which is very strange. Here are the average results with shadow quality set to Very High (r_softShadowsQuality 48) for maximum impact, map Written at the player start position and angle: 2.11 Beta 2, stencil & soft shadows: 105 FPS 2.11 Beta 2, stencil & hard shadows: 105 FPS 2.11 Beta 2, maps & soft shadows: 80 FPS 2.11 Beta 2, maps & hard shadows: 90 FPS 10211, stencil & soft shadows: 25 FPS 10211, stencil & hard shadows: 25 FPS 10211, maps & soft shadows: 25 FPS 10211, maps & hard shadows: 25 FPS Worth noting: If I look away and into the ground while staying at the same position, it will jump back to some 100 FPS. The max 144 FPS is still achieved in the main menu so there's no hard cap.

Since I just played Written in Stone and it has great detail I'll use it as a benchmark too, player start position and rotation. The performance degradation here is insane: 111 FPS on 2.11 Beta 2 engine, 25 FPS on SVN 10211 engine. Exact same Darkmod.conf or a fresh one same thing, simply using the different binary makes this difference. Tried all of the switches you mentioned and more, can't find anything that will bring the FPS back up. Only immediate thing I can think of is going through all the r_skip cvars in order again and see if any makes a difference.

Sorry for the delay. Not that I'm aware of, the port should normally be free. What I imagine must be happening is when TDM (launched by DR) is closed, Linux still sees the port as being busy for a while, despite even the process having been shut down. Never heard of such a thing before nor can easily imagine what must be happening, but at least it makes sense. This might not be a TDM specific issue in this case... unless the engine fails to tell the OS to close the port in the way Linux expects? Though perhaps an improvement could be made: What if darkmod doesn't crash when the port is busy, but instead just doesn't receive signals until it's freed? Would a more dynamic approach to busy connection ports make sense?

That's awesome, congrats! Though now we have to figure out why it horribly tanked for me instead Settings are the same only the engine differs, something in this change must have broken it.

I started putting together the basis of the project where I'll need my rotating / elevator stuff, still yet to get to that part. Hopefully without spoiling too much of my surprises, I'm toying with a dynamic daytime cycle of sorts. I'm easily able to obtain what I want using a light texture material... already have a working prototype which respects the lightgem and all, the skybox will use a fog light with a similar algorithm for the atmosphere. Only issue is I can't quite line up the color cycle to get the proper colors, seems I always get stuck when it comes to complex mathematics. red (sintable[(time * 1) / Parm4]) * Parm0 green (sintable[(time * 1) / Parm4]) * Parm1 blue (sintable[(time * 2) / Parm4]) * Parm2 ShaderParm4 is the length of a day. I'm using 5 for a test... it will be much larger in practice of course, a day will probably last somewhere between 5 and 10 real life minutes. The colors are off though... using a sintable / costable mixture works but I need the proper offsets: I want the sun to start / end at a full red, get yellowish just before / after the sunset / sunrise, and during the day just slightly increase its brightness toward noon... for the moon at night it should stay white just at very low intensity (0.1 should do). If anyone figures out a good formula please let me know, thanks!

Strange: With 10211 SVN engine I'm getting extremely low FPS all of a sudden, 20 FPS where I normally had 80 FPS. TDM 2.11 Beta 2 engine doesn't do it so it must be the new changes. It's not the debug engine, running thedarkmod.x64. Not shadow samples this time either, does it with both maps / stencil and soft shadows disabled. No improvement if I use a fresh Darkmod.conf or remove the generated glprogs directory.

@lowenz I'm not an engine expert but if that's alright I wanted to offer a few suggestions of my own to see if we can nail this down faster, after playing with the renderer cvars yesterday and noticing some interesting ones. First of all since I haven't seen this mentioned specifically: Have you ruled out the relatively new volumetric lights as a cause? Set "r_volumetricEnable 0". For me this barely makes a difference, just a few FPS. Otherwise an idea: Open the console, type r_skip, then press tab... this should print all skip variables. Can you try setting each to 1 and see if any magically brings the FPS back up? Remember to set them back to 0 as those disable core features and should only be used for debugging, also ignore r_skipBackEnd as that will stop drawing everything including the menu / console! For me each has a small contribution but none of them appears to be a silver bullet for the overall low performance, except r_skipFrontEnd because it literally makes everything white

It's hard to tell if I'm experiencing a completely different problem. What we have in common is we both use AMD cards, what differs is he's getting the issue on Windows while I'm on Linux / Mesa / amdgpu. The exact triggers and results seem to intersect in some areas. In case it helps mine is a ROG-STRIX-RX570-O4G-GAMING if that's a model which should be affected by common driver stuff: Today I read about the new 2.11 Beta 2 and decided to update. First thing after starting a map, I noticed performance was down to 20 FPS where it would normally be 60 FPS. After a minute of digging I found that's because r_softShadowsQuality reset itself to 24 I think. I set it back to Low / 6 and things went back to normal, but this was a good opportunity to notice how extreme the performance impact of having even a modest setting can be. I could have sworn it wasn't as bad an year ago, albeit the maps I was playing also made a big difference. This is mainly on shadow maps though: If I use stencil shadows the impact is lesser with more samples. Even if I turn soft shadows off however, I feel I get abnormally low FPS on both Stencil and Maps. This is great news: Compiling the engine from SVN should still work on my end, I'll wait for the patch and see what it changes for me. I really feel something must be up with the AMD driver, at least some of the performance degradation must have a common root between what both me and lowenz are seeing... if not I'm wrong and it's a learning experience, but I sense the engine changed something all AMD cards of a certain type aren't happy with.

I did a quick test with that: Noticed "r_softShadowsQuality 4" looks roughly as alright as 6 but gains a notable bit of extra FPS. Problem is that breaks the menu setting causing Soft Shadows to appear as Off. Actually could we have more options for the menu entry? Like a Lowest setting for 4 samples, 8 for Low, 16 for Medium, 32 for High, 64 for Highest. Or something among those lines, but at least have the lowest start from just 4. Or why not just use a slider? The sharpness effect has one, feels like it would make more sense here. This is the sort of thing users may want to easily tune based on their hardware and / or perceived visual loss.

Thanks. Played with those settings and the ones on the wiki page. Strangely enough nothing else improved performance noticeably except for "r_useEntityScissors 1": This gives me almost +10 FPS in places which was quite a beloved surprise! Any reason not to be using this by default? What about r_shadowMapSinglePass and r_shadowMapCullFront? I've those on for ages with shadow maps in use, they work flawlessly for me with visible improvements. Note that "r_shadowMapSinglePass 2" still causes an instant crash... let me know if you'd like me to try debugging that, though last time I asked about it I believe I was told it's no longer being maintained due to driver limitations which is unfortunate. I understand "r_useAnonreclaimer 1" should also be faster so I'll autoexec it too. No difference at least with my map's portal setup, but also not seeing any issues with so at least it's seemingly safe to try.

Just curious, given I don't get quite the performance I feel I could on Linux / Mesa / amdgpu: Are there any performance optimizations or significant features that are off by default, introduced since 2.10 and that make sense to test for 2.11? Currently I only have the following in my autoexec.cfg which to my surprise weren't turned on by default last I checked: r_useMultiDraw 1 r_useMultiDrawIndirect 1 r_shadowMapSinglePass 1 r_shadowMapCullFront 1

An unfortunate side effect to be aware of: The spectrum light won't respect shadows that should be cast by normal surfaces. I can easily imagine that's because the light completely ignores anything that isn't on the custom spectrum including shadow casters. Would be nice to have a solution, especially as this should mean symbols can be illuminated through solid walls. A bearable inconvenience at the end of the day however, might shorten the UV light radius if it seems too bad.

And done. Very gorgeous FM, loved it overall: Story was great and the environments some of the best! Only found about two thirds of the loot and just two secrets, maybe next time I play it I'll be more thorough.

Ooof. I can't give up soft shadows, too pretty... +80 FPS just by using hard shadows though? Curious if you're still on Mesa 22.2.4 like me, think you said that Mesa 22.3 is supposed to improve performance a bit so fingers crossed. I don't get anything as bad as that on Linux / amdgpu though. Still feels like performance is lower than it should be... as in I drop below 100 FPS in areas that aren't full of complex geometry and lights enough to feel justified.

Still planning to finish this soon, working on my own FM and didn't find the time to return to it. Got into the area I was having trouble with, now time to find the objective in there. In the meantime a joke that had to be made, for those who reached this zone and know what I'm talking about...

Speaking of reflection bugs and flickering: In the FM I'm working on (not public yet) I placed the puddle decal on some floors. I noticed that from certain positions and angles the puddle turns white, other decals seem to be working fine so it's likely specific to reflections. Should be possible to replicate using the button to create a decal on the selected face, moving the resulting patch 1 grid unit above the brush surface, then texturing it with "textures/darkmod/decals/dirt/puddle". Between the screenshots below, all I did was look a little to the right to trigger the new appearance, which can be seen becoming whiter and losing its reflection. There's also a difference of 20 FPS between the two, likely meaning the puddle shut reflections off... just now realized how great the performance impact of having puddles can be

Interesting. I'm on amdgpu on Linux / Manjaro but don't believe I ever saw such flickering. Closest I can say I've seen, excluding the quicksave flashing which I understand was fixed, is when turning the view around too quickly and a lot of portals loaded the whole place turned white or was invisible for a few frames. No black boxes like in that video though.

That is correct, func_static typically shouldn't be used for the purpose. I'm looking at func_mover but can't quite tell what I'd need to script myself or will work automatically: If I have it targeting multiple positions specified by another func_* (target0, target1, target2, etc) can each button call it to go to that specific target like the Wiki indicates, accounting for both position and rotation? multistate doesn't seem to support different rotations so might not be what I'm looking for in this case.