nohats mod doesn't work on sv_pure 2 servers anymore after the last update that's why i guess .

I played a scrim just yesterday and it was working fine...

played a lobby doesn't work for me, Harbs said he is emailing tf2 team about it .

well shit...

I unstalled it even thought I don't need FPS cfg. I ust like the way TF2 looks with lower settings, feels good and stable.

yeah , I played alot with the hats mod , and the HATS were off .

Laptop Dell Inspiron N5040
Specs:
Intel Pentium 2.13ghz
Integrated Intel HD Graphics
3GB of RAM

Been using your old one, so I'm looking forward to using this when I get home.

Also, I hope I can take a small slice of credit in reminding you about your config several days ago.

it crashes when I use it?

Their's is 2 commands in this CFG that are not stable for people that don't use quad cores.

cl_threaded_bone_setup 1
r_queue_decals 1 (some have said this is unstable while cl_threaded_bone_setup 1 isn't for them)

I have an fx-8320 (8 cores for reference) and r_queue_decals is really unstable for me. Probably just an amd thing, or maybe it's just my particular chip.

I've tested this on a system with a AMD Phenom II x4 945 and one with the 940. It's still entirely possible it's not stable for everyone, but not necessarily all AMD cpus.

i use a quad core...

Did you miss this?

Updated my CFG small fps gain 2-4%

// ----------------------------------------------------------------------------
// HUD
// ----------------------------------------------------------------------------
hud_saytext_time 10 //Default is 12 and measured in seconds, this is the amount of time chat stays on your screen
voice_enable 1 //Default is 1, This is the voice chat in the game 1 = on, 0 = off
hud_deathnotice_time 5 // Default is 6, This the the death notices that usually appear at the top right of the screen
hud_achievement_glowtime 0 // Default is 2.5, not 100% what this is about a 1.5% fps boost.
hud_achievement_count 0 //Default is 8, Max number of achievements that can be shown on the HUD
hud_achievement_tracker 0 // Default is 1, Show or hide the achievement tracker

I need to do more testing apparently, cause I tested this just before tonight's update and I had a full 8 extra FPS, and immediatly after I have about 14 less fps than before!?!

-edit-
So this is my framerate on my last cfg for this update

2639 frames 12.561 seconds 210.10 fps ( 4.76 ms/f) 9.532 fps variability

What it was...

2639 frames 11.770 seconds 224.22 fps ( 4.46 ms/f) 11.665 fps variability

So my update to my CFG still stands as it's still a small FPS boost, but you will get less FPS cause this update did something odd :/

Yeah because full moon mode on.Don't worry if off, again u take back +10 fps.

That actually effects FPS by that much...

6.25% decrease in FPS :/

That's... interesting. Why would it decrease FPS that much just by a value change? What does it do visually other than allow Halloween hats? If that's the cause, does using the NoHats mod counteract it?

While idk -everything- that the Halloween/Full Moon mode does here's some of them.

Halloween Hats

New Textures for Health Kits

I suspect this is most likely inefficient code for checking if Halloween/Full Moon Mode is turned on.

yes because diferent code im use nohats mode and confirm if halloween mode ON have -6 -10fps

I don't know what it was in this config but my mouse feels a lot snappier than when I used m0re's or Chris' configs. I never really had a problem with my FPS when using those, but the mouse always felt...floaty? Anyways, it's perfect with this config so thank you.

pretty sure I am the only one of those cfgs that added the "m_rawinput 1"

Updated the CFG

r_occlusion 0 // disables ambient occlusion. 2% FPS boost

// ----------------------------------------------------------------------------
// Test Shit
// ----------------------------------------------------------------------------
cl_localnetworkbackdoor 0 //Network optimizations for Singleplayer, disabling has about a 1-2% fps boost
cl_cloud_settings 0 // I believe this disables cloud syncing of CFGs but not 100% sure
//cl_notifications_show_ingame 0 // Pretty sure this disables Trade/Duel Notifications
sys_minidumpspewlines 1 // Basically the number of lines saved to a log file from console. No FPS boost but 15% lower FPS variance
cl_loadondemand_default 0

I can't believe I missed r_occlusion in the past, was literally a no brainer.

Before (framerates are overall lower do to the last update with full moon mode on)

2639 frames 12.561 seconds 210.10 fps ( 4.76 ms/f) 9.532 fps variability

After

2639 frames 12.243 seconds 215.55 fps ( 4.64 ms/f) 9.587 fps variability

https://dl.dropboxusercontent.com/u/92187841/gfx.cfg

-edit-
Until the TF team releases new commands or makes some noncheat required it's going to be extremely difficult to find new settings spent roughly 14hrs this past week testing nearly every allowed setting in the game...

r_occlusion 0 // disables ambient occlusion. 2% FPS boost

// ----------------------------------------------------------------------------
// Test Shit
// ----------------------------------------------------------------------------
cl_localnetworkbackdoor 0 //Network optimizations for Singleplayer, disabling has about a 1-2% fps boost
cl_cloud_settings 0 // I believe this disables cloud syncing of CFGs but not 100% sure
//cl_notifications_show_ingame 0 // Pretty sure this disables Trade/Duel Notifications
sys_minidumpspewlines 1 // Basically the number of lines saved to a log file from console. No FPS boost but 15% lower FPS variance
cl_loadondemand_default 0

2639 frames 12.561 seconds 210.10 fps ( 4.76 ms/f) 9.532 fps variability

2639 frames 12.243 seconds 215.55 fps ( 4.64 ms/f) 9.587 fps variability

Thanks bae

Looks like some interesting stuff, I'll add those to my config. Past that, I'll probably print all cl_ and mat_ commands to a text file from console and look for potential improvements.

I just wish convar documentation was better, because this can only be painful to do. Hopefully they improve it in Source 2.

I've tested basically every allowed command from here
https://developer.valvesoftware.com/wiki/List_of_TF2_console_commands_and_variables

btw their's A LOT of really unstable commands to either enable or disable.

That list is outdated by two years, new convars are sure to have turned up or been missed. In fact, I found some experimental ones that I'm about to benchmark with my config;

ai_expression_optimization 1       // Disables NPC expressions when you can't see them, not sure if it works on regular players in TF2, but that's why it's experimental.
//r_pixelfog 1                     // Supposedly increases FPS via fog, but it doesn't seem like it. Feel free to play with it to test. I'll run benchmarks later.
fast_fogvolume 1                   // Undocumented but I didn't experience any FPS loss with it enabled, will run benchmarks later.
filesystem_native 0                // I think there's an FPS gain with this, anyway.
host_thread_mode 1                 // runs host in threaded mode if multicore

ai_expression_optimization 1       // Disables NPC expressions when you can't see them, not sure if it works on regular players in TF2, but that's why it's experimental.
//r_pixelfog 1                     // Supposedly increases FPS via fog, but it doesn't seem like it. Feel free to play with it to test. I'll run benchmarks later.
fast_fogvolume 1                   // Undocumented but I didn't experience any FPS loss with it enabled, will run benchmarks later.
filesystem_native 0                // I think there's an FPS gain with this, anyway.
host_thread_mode 1                 // runs host in threaded mode if multicore

ai_expression is definitely just for the AI and alot of other commands disable expressions
r_pixelfog I tested it's definitely not helpful turned off in Dx9 at least with a FPS cfg might be helpful in a High Quality CFG, and does nothing in Dx8

filesystem_native 0
host_thread_mode 1

These seem promising, but I have a feeling host_thread_mode is going to only affect local listen servers but I could be wrong.

Forgot to mention I also added

net_queue_trace 1
pretty sure this will help with optimizing the net code for TF2 but I could be wrong.

I gave them a try on my crappy PC and got these results with the benchmark;

(all experimental benchmarks and r_pixelfog on)
2860 frames 73.944 seconds 38.68 fps (25.85 ms/f) 2.656 fps variability

(filesystem_native set to 1, host_thread_mode set to 0)
2639 frames 68.168 seconds 38.71 fps (25.83 ms/f) 2.682 fps variability

By the looks of it, the FPS variance definitely went down from yours either way, but I don't think they did anything that would affect the demo. Perhaps it'd be more effective in real play, I don't know, but they don't seem to harm anything. I am, however, leaving r_pixelfog 0 commented since it seems to lower FPS in some situations.

Added some new convars;

mod_load_anims_async 1             // Loads model animations async
mod_load_mesh_async 1              // Load model mesh async
mod_load_vcollide_async 1          // Load model vcollide async

Note that with the above benchmarks and the one below, all of my standard graphics settings are set to their lowerest in windowed noborder at 1366x768. Gibs/ragdolls are on.

2639 frames 66.843 seconds 39.48 fps (25.33 ms/f) 2.654 fps variability

Looks like a very small FPS boost if anything, not a big deal but if it's not detrimental and doesn't cause crashes I'll go with it.

Didn't notice this, thanks. I'll add it.

Did you run 2 different demos cause the number of frames should be the exact same. Specifically the 2639 Frames is the one that was provided in the benchmark thread.