8.4GB

it's monday

those control entity shadows, not map shadows. without shadows, the map would be completely bright (mat_fullbright 1)

wait where's the 6th button on the mouse, i can only find mouse 4/5

also, it works fine for me, but i'm on linux. so i guess it is a logitech software issue. maybe your buttons are set to be programmable?

Not sure about this and unfortunately have no way of testing this myself, but copy this file into a custom folder (or edit the one in mastercomfig if you're using my config): https://raw.githubusercontent.com/mastercoms/tf2cfg/master/mastercomfig/dxsupport_override.cfg, then under

"1"{
"name" "81"

change

 "ConVar.cl_particle_retire_cost"	"10"

to

"ConVar.cl_particle_retire_cost"	"0"

I don't know any specifics, just heard reports of it happening from some people.

It uses the Source Engine to get mouse movements.

Well, the material system prefers certain video modes of w, h and refresh. Not sure if deviating from those video modes will decrease performance, but it might.

it's sunday

You can also use mat_viewportscale % (% = 0.5 for 50% res, etc) to reduce the game resolution without reducing the HUD resolution.

What preset are you using? Players not showing up might be solved using host_thread_mode 0

On some systems, raw input may cause input lag.

Just drop mastercomfig in your tf/custom folder and it will work.

It's no secret that the glow outline has performance issues, which I believe will be fixed in the next major update. However, that FPS drop is a lot bigger than what I would expect.

mammoth imminent

http://www.youtube.com/playlist?list=PLHy7G7ndrUmq933tLi_nTlFlvmJKZpmAi

EDIT: they were never non-private

also, someone got the thumbnails http://www.valvetime.net/threads/thumbnails-of-new-private-team-fortress-2-videos.257895/

Those aren't errors, those are me forcing values for cheat console variables using dxsupport overrides.

Try using mat_dxlevel then mat_savechanges.

2639 frames 19.082 seconds 138.30 fps ( 7.23 ms/f) 9.900 fps variability
2639 frames 19.023 seconds 138.73 fps ( 7.21 ms/f) 8.932 fps variability
2639 frames 19.048 seconds 138.54 fps ( 7.22 ms/f) 8.677 fps variability

Glad you got it working better. However, I wouldn't recommend net_compresspackets 0, since I believe splitpackets are more CPU intensive than packet compression. Also, I would recommend doing config changes in a new config file exec'd after comfig so that you can easily update the comfig file, and so that you can easily share your own "preset" for others to use.

All of those should help. The less things the engine has to render, the better your performance will be. The biggest performance hindrances are particles and player models/animations so focus on getting those kinds of mods.

4.4.9 enables host_thread_mode by default again.

This command improves FPS by a lot, and I had it enabled by default before but then disabled it due to people asking for help with their local servers, which host_thread_mode 1 breaks.

I've decided that it is best to keep it on because the majority of people probably wouldn't remember to turn host_thread_mode on, but will remember to turn it off if they're in a local server that isn't working.