Using dxlevel 81 is not a misconception. It has a valid use case. Same with the force ID settings. I haven't looked too much into those but people do strangely get a benefit out of them sometimes.

There are two incorrect things in the Delfy video that I spotted: r_lod 2 instead of r_rootlod 2 and the recommendation to use -threads -high in launch options (granted, I skimmed through it).

Anyways, the mastercomfig docs are open source! So you can improve them however you'd like.

How do you think a game engine or any benchmarking tool measures FPS? It uses frametime by measuring the time taken between the start and end of the frame, then takes the reciprocal of that (divides 1/frametime).

You may be thinking of dxlevel 98.

You don't have to do this anymore (and it doesn't have to do with VPKs anyway):

https://docs.mastercomfig.com/en/latest/setup/clean_up/

Also, as listed on the download page, you can access the mastercomfig zip package from here.

Having an FPS cap may be throttling your performance more than necessary. Try using an external FPS cap like Rivatuner.

It does use some DX8 shaders so it might negatively affect stuff on your machine.

frametime = 1/FPS
1/frametime = FPS

FPS is bad to measure as a relative thing, like "I just gained 30 frames!", instead of "I had 240FPS before and now I have 270FPS." That latter one is fine.

if you mean the frame pacing, that's harder to compare and measure between configs.

No, it controls shader quality. Reducing fillrate in this instance means reducing the required pixel fill cost.

This is irrelevant but given that I was mentioned and kind of misunderstood, I thought I'd respond. I didn't I said you would get higher FPS all the time with DX9 vs DX8. The more efficient renderer is rendering more, so it's just doing more, faster, but in more time than something that's doing less, slower. So it's understandable that it's getting less average FPS.

Anyways, DX9 can help a lot with drops, latency, resource usage, frame timing and performance over longer periods of time even if it isn't getting you higher average FPS. Finally, keep in mind that how you test dxlevel can affect your benchmarks. You should be setting dx9 and then launching without -dxlevel 90, or else you will experience some graphics options getting reset, causing lower FPS.

The things you were saying: "junk strings that aren't launch options"

Thanks for your interest in contributing to mastercomfig! There's a contributing document that contains packaging instructions here: https://github.com/mastercomfig/mastercomfig/blob/develop/.github/CONTRIBUTING.md#packaging

Beware that any sanitization of the launch options must be automated to be part of the generation process. So a filter of known bad launch options could work, as well as a heuristic for random garbage.

You can use your own personal knowledge of TF2 to develop mastercomfig. Where that knowledge comes from is not of my concern.

Yeah how Windows and Source interact for threading adds a lot of overhead and latency to the frame. That's sort of what I was working on that gave the biggest gains in TC2.

what's your hud

Limitation in Steam.

Already done.

I am pretty sure that it's just a problem with TF2, since split packets (>1260 bytes) are kinda not well implemented.

are you using a surface properties mod, like no footsteps?

New thread here: https://www.teamfortress.tv/58095/rgl-pugs-technical-alpha