2639 frames 18.458 seconds 142.97 fps ( 6.99 ms/f) 13.702 fps variability

CPU and overclock: i5 4460 @ 3.20GHz
Graphics Card: ASUS Radeon 270X
Driver version:15.20.1062.1004
dxlevel (default is 90): 81
Resolution: 1280x720
Full-screen or windowed: windowed
FPS configs enabled: modified comanglia toaster config (I'm just used to the ugliness, and it's easy to switch the few commands that matter for quad-cores)
Shadows enabled/disabled: disabled

notes: this is post-halloween '15, I really should've done one before the update just to check but I'm pretty sure I used to get noticeably lower than 142 fps on average

down from 250 to 220 with the halloween update.

dx8 vs dx9 post 2015 halloween update ?

I went from ~234 fps to 242 fps after the update in dx81

i5-4590,msi gtx 960,1600x900,fullscreen,shadows enabled
went from 120-125 to 145 on dx81 and cfg posted by spammyh (#184)

Yo lets get some people doing benchmarks gaining/losing fps from the update so we can figure out why.

I'll do one when I get home but my guess is that dx8 got a more extreme fps boost due to not seeing skins. I play on dx8 with a i7 4770k and my frames are higher and got more consistent with the update.
edit nvm looks like it definitely is dependent on the setup haha

/

dx9 runs as good as dx8 for me now after the update with a tweaked comangalias config (apparently high model and texture detail gives better fps now)

Source?

https://www.reddit.com/r/tf2/comments/3dmhlm/fps_fix_mat_picmip_1/

granted, this has been a thing for dx9 since gun mettle came out, I don't know why it'd specifically apply to this update

With a i5-4670k @ 4.7ghz and a gtx 960 on dx81, I am getting identical fps when changing between 2 and -1
edit: one thing to keep in mind is that the stv benchmark1 is from before skins so dx8 vs dx9 and -1 vs 2 could vary when skins are a factor.

CPU and overclock: AMD FX-8350 @4.00 GHz no turbo, no OC
Graphics Card: ASUS Radeon HD 7750 //Same chip than R7 250
RAM: 12 GB Kingston DDR3 RAM @ 1866 Hz
Driver version:15.20.1062.1004
dxlevel (default is 90): 81
Resolution: 1024x768 @85Hz fullscreen
FPS configs: Chris highframes

I guess more isn't always good, or maybe because of how short the test is the results vary.
The graphics card isn't the best part of my system by far

Nice. R7 250X has the same chip than HD 7770

Edit: spoilered results

I've still got a 2500k at 4.5ghz and a 7850, went from 225fps last week to 195 after the update.

Geforce GTX 970
Intel Xeon CPU E3 1231 v3 @ 3.40Ghz
16 GB RAM

CFG used: R4ndom + pazer's & Wareya's reccomendations.
aHUD for after scream fortress, my hud for before scream fortress.
Nohats mod, surface properties mod, nosmoke rocket mod.

Before Scream fortress:

fps variability totaltime numframes width height
214.5 17.7 12.3 2639 1920 1080
214 16.7 12.3 2639 1920 1080
215.6 17 12.2 2639 1920 1080

After scream fortress:

fps variability totaltime numframes width height
208.7 17.9 12.6 2639 1920 1080
205.3 17.5 12.9 2639 1920 1080
207.9 18.7 12.7 2639 1920 1080
179.8 20.7 14.7 2639 1920 1080 (My hud instead of ahud)

Was hoping for more frames from this update tbh

My cfg: http://pastebin.com/Q9BxWR3d
Steam status: Offline
Mat_antialias: 4

125,66 fps; 8 var
125,95 fps; 7 var
125,95 fps; 7 var

R4ndom cfg: http://pastebin.com/5THqg6pX
Steam status: Offline
Mat_antialias: 4

139,38 fps; 9 var
139,60 fps; 8 var
138,53 fps; 8 var

Rig:
i5 2500 NOT K - 3.3ghz
gtx 760 stock
8gb ram

All dxlevel 81.

2639 frames 15.610 seconds 169.06 fps ( 5.92 ms/f) 11.079 fps variability

CPU and overclock: i5 4460 @ 3.20GHz
Graphics Card: ASUS Radeon 280X Dual-X
Driver version: 15.201.1151.0
dxlevel (default is 90): 81
Resolution: 1920x1080
Full-screen or windowed: Fullscreen
FPS configs enabled: R4ndom's Config
Shadows enabled/disabled: disabled

2639 frames 19.270 seconds 136.95 fps ( 7.30 ms/f) 11.737

CPU and overclock: A8 5600k @4.0GHz
Graphics Card: ASUS Radeon 270X TOP OC
Driver version: 15.11
dxlevel (default is 90): 81
Resolution: 1280x1024
Full-screen or windowed: Fullscreen
FPS configs enabled: slightly modified comanglias
Shadows enabled/disabled: disabled

Notes: Installed the latest AMD beta driver that came out a few days ago and am getting around 16 more frames.

question: how difficult would it be to create a new general demo for benchmarking?

the current demo should still stick around for the sake of quick comparison over long stretches of time, but things like weapon skins aren't taken into account with it, when that's well-known to be a major FPS killer

It's easy to make demos.

If someone does make a new demo I would suggest:
-Posting it a new thread for the benchmark results (so as to not confuse any results here)
-Make the demo cover at least 30 seconds of gameplay
-Include at least 19 people in the server on different classes in an area close together.

trying to test process lasso, with a potential use-case:

//-dxlevel 98, twitch stream, youtube video + a bunch of chrome tabs in BG
2639 frames 21.835 seconds 120.86 fps ( 8.27 ms/f) 21.442 fps variability
2639 frames 20.788 seconds 126.95 fps ( 7.88 ms/f) 20.675 fps variability

//now w/ process lasso
2639 frames 20.620 seconds 127.98 fps ( 7.81 ms/f) 14.487 fps variability
2639 frames 20.828 seconds 126.71 fps ( 7.89 ms/f) 14.216 fps variability
2639 frames 20.668 seconds 127.68 fps ( 7.83 ms/f) 14.173 fps variability

thought I found some sort of proof that process lasso does something objectively useful, and restarted TF2 to check. then got:

//non-lasso
2639 frames 18.971 seconds 139.11 fps ( 7.19 ms/f) 12.719 fps variability
2639 frames 18.601 seconds 141.87 fps ( 7.05 ms/f) 12.209 fps variability
2639 frames 18.711 seconds 141.04 fps ( 7.09 ms/f) 12.199 fps variability

//on-lasso
2639 frames 18.894 seconds 139.68 fps ( 7.16 ms/f) 12.089 fps variability
2639 frames 18.838 seconds 140.09 fps ( 7.14 ms/f) 11.836 fps variability
2639 frames 18.885 seconds 139.74 fps ( 7.16 ms/f) 12.434 fps variability
2639 frames 18.878 seconds 139.79 fps ( 7.15 ms/f) 12.039 fps variability

...wherein I got consistently higher frames and lower fps variability all-around, despite only ever touching process lasso. other people in the update thread have noticed similarly. (also, any potential advantage process lasso gave immediately became very marginal, at best)

if you got lower frames this update, try restarting the game once and check benchmarks again

2639 frames 8.911 seconds 296.14 fps ( 3.38 ms/f) 19.225 fps variability @ 4.7 ghz
2639 frames 8.672 seconds 304.30 fps ( 3.29 ms/f) 21.929 fps variability @ 4.8 ghz

CPU and overclock: i5-4670k @ 4.7ghz
Graphics Card: EVGA GTX 960 SSC + slight oc
Driver version: 358.91
dxlevel: 81
Resolution: 1920x1080
Full-screen or windowed: Fullscreen
FPS configs enabled: slightly modified comanglias (hud_saytext_time 0 net_graph 0) + nohats mod + no effect mod + no explosion smoke script + bullet hole dust
Shadows enabled/disabled: disabled

added following dx levels and settings @ 4.7ghz :
2639 frames 9.611 seconds 274.58 fps ( 3.64 ms/f) 18.226 fps variability
dxlevel 81 without nohats/noeffect mods

2639 frames 8.763 seconds 301.16 fps ( 3.32 ms/f) 20.742 fps variability
hud_deathnotice_time 0

2639 frames 8.559 seconds 308.33 fps ( 3.24 ms/f) 22.129 fps variability
cl_drawhud 0

2639 frames 10.082 seconds 261.75 fps ( 3.82 ms/f) 18.448 fps variability
dxlevel 90

2639 frames 9.664 seconds 273.08 fps ( 3.66 ms/f) 16.989 fps variability
dxlevel 90 + cl_drawhud 0

Highscore?!

I believe you just won.

Checking in after updates.

CPU and overclock: i7 2700k @ 4.3Ghz
Graphics Card: GTX 660ti
Driver version: 358.91
dxlevel: 90
Resolution: 1920x1080
Full-screen or windowed: Windowed Borderless
Config: Personal config emphasizing fps

Extras: Have 2 monitors in total hooked up to the same graphics card
Running Windows 7.

Previous Results:

2639 frames 13.608 seconds 193.92 fps ( 5.16 ms/f) 14.320 fps variability
2639 frames 13.599 seconds 194.06 fps ( 5.15 ms/f) 13.640 fps variability
2639 frames 13.610 seconds 193.90 fps ( 5.16 ms/f) 14.000 fps variability

New Results:

2639 frames 13.112 seconds 204.38 fps ( 4.89 ms/f) 14.957 fps variability
2639 frames 13.013 seconds 203.80 fps ( 4.93 ms/f) 14.336 fps variability
2639 frames 13.158 seconds 204.56 fps ( 4.99 ms/f) 14.106 fps variability

.

CPU and overclock: i7 4790k @4.4GHz
Graphics Card: SLI GTX980 (~1129MHz stock clocks)
Driver version: 358.50
dxlevel (default is 90): 90
Resolution: 1920x1080
Full-screen or windowed: Windowed
FPS configs enabled: None, settings are

http://puu.sh/llbyR/ead824296b.png

Shadows enabled/disabled: Enabled

2639 frames 26.348 seconds 100.16 fps ( 9.98 ms/f) 6.794 fps variability
2639 frames 27.120 seconds 97.31 fps (10.28 ms/f) 5.904 fps variability
2639 frames 26.434 seconds 99.83 fps (10.02 ms/f) 6.144 fps variability
2639 frames 26.220 seconds 100.65 fps ( 9.94 ms/f) 5.955 fps variability
2639 frames 26.131 seconds 100.99 fps ( 9.90 ms/f) 6.265 fps variability

And just for fun, the same settings at 3840x2160

2639 frames 26.676 seconds 98.93 fps (10.11 ms/f) 7.040 fps variability
2639 frames 27.386 seconds 96.36 fps (10.38 ms/f) 6.331 fps variability
2639 frames 26.773 seconds 98.57 fps (10.15 ms/f) 6.373 fps variability
2639 frames 26.827 seconds 98.37 fps (10.17 ms/f) 6.150 fps variability
2639 frames 26.750 seconds 98.65 fps (10.14 ms/f) 6.167 fps variability

lol tftv stream coming at you in 4k downscaled

Edit: Odd that my GPU never went over about 35% usage and neither of my GPUs went over 30%. Adding my 1366 MHz overclock didn't change anything either since they never had to get themselves over 900MHz or so.

your fps seems kind of off David (I have an i7-4790k @ stock and even w/ dx9 I get higher fps w/ a crappier vid card.) Are you able to see the CPU core loading when it's running?

Are you running the game at close to max settings though? I do that because I only use TF2 to stream for TFTV so I'm not bothered about getting 120fps, so long as my game is at 60 or higher it's fine for the stream. I said that my CPU doesn't go above 30% usage or so but I'll do a test now to see if that's all on one core or something.

Nope, spread across all 8 physical and logical cores, none going higher than about 30-40% usage.