Does your HUD have transparent viewmodels?

Well if a model has a lower LOD level than 2, it won't ever be used. r_rootlod sets the highest LOD level that can be used, so while r_lod 2 will always use LOD level 2, r_rootlod 2 will start at LOD level 2, and fade to even lower LODs if needed.

r_lod -1 uses automatic LOD, i.e, uses r_rootlod and fades through different LODs. Using anything other than -1 is not recommended (unless you're using 0 to force best quality), because if you try to use r_lod to force a lower LOD, you might not be using the ideal LOD (performance-wise) for that distance.

Sure, but it will take me some time to get it out though since I now use a Nvidia card and only AMD cards have this feature on Linux. I'm building a new PC (with Vega) in around December since this i7 870 is falling off now.

I did that for my own config. https://www.youtube.com/watch?v=faNeipYH4cw

I apologize for my awful gameplay, the overlay was displayed while I played, so I couldn't see anything :(

I've added that to the readme. Might move it to the wiki once I get more up.

Edit: moved to wiki.

May I use these screenshots? (and under what conditions/copyright license)

What are your launch options? Do you have any modifications installed?

dx9. it is the most efficient/mature source renderer.

dx8 sometimes gets more frames because it simply renders less, but most effects supported/enabled from dx9 can be disabled with cvars.

Thanks for the specs. The Intel Iris one is integrated, the driver will most likely prefer the R5 M230 when you're running TF2.

I meant the high quality PC config.

Thanks for that typo find. I've fixed it.

Well I've seen that's been bugged on various systems/devices, regardless of vendor, so I won't add a specific warning for AMD.

host_thread_mode 1 causes this. It doesn't work with local servers but helps in other situations. Use host_thread_mode 0 for when you're on a local server.

What are your specs? And I think it's more fair to use Comanglia's default, since it uses some of the multithreading CVars I use in mine (unless you've disabled those for this test).

Thank you! I'd love to see where I'm in terms of performance, especially on shitty laptops. If you could, please try to find CVars in the config that are suggested to be changed for bad CPUs/GPUs.

No problem!

That's strange. When did you last try my config and what do you mean by seize up?

High RAM usage is fine, though. It means the game is caching more assets for less stuttering and shorter load times.

It's just the memory settings in the cfg, as you said, though the only side effect should be high memory usage, not any performance issues.

As said in the bottom of my post with a note, I haven't gone through all of the mat CVars. I just haven't investigated bumpmap and specular and that's why they're not in the config. I don't know if they're good or bad yet. Sorry if my statement about "popular/most important CVars" was confusing. I'll change the wording to be less so.

// These settings will boost your performance on a 4 core CPU
// loop through the leaf asynchronously
cl_threaded_client_leaf_system 1

I found that the culprit for crashes was actually r_queued_decals 1 (at least in my case and one of my friend's), which should be turned off anyway with low decal counts.

Also, cl_threaded_client_leaf_system isn't actually used by default in the base Source 2013 engine, though I've left it in the config in the event that TF2's engine uses it (through extension of the base engine, or lack of an update for that system).

I haven't personally done any benchmarks, mostly because I use Linux and it won't be an accurate representation of performance since most configs don't have Linux specific CVars, and ToGL (which is what Valve uses to convert DirectX to OpenGL) messes up performance statistics.

I'd love if someone could make their own benchmarks at this point.

https://comfig.app/img/mastercomfig_banner.png

mastercomfig

a modern customization framework

• web app
  
• docs
  
• support me!
  
• discord server