Teeworlds - bridge

Teeworlds

IRC / bridge

One-way IRC channel bridge. If you want to be able to send messages to IRC, contact @Dune or @heinrich5991. https://www.teeworlds.com/?page=docs&wiki=rules/irc_rules

Between 2020-01-19 00:00:00Z and 2020-01-20 00:00:00Z

usually i go with sv_register 0 in debugging

i usually go with debug 1 in debugging :troll:

if you go debug in compilation you go dbg_dummies

00:53

dbg_dummies are the most usefull shit i have ever seen

+1

I'm writing golang stuff

01:10

that gets master server data

debugging is kinda hard

does anyone know how to build the masterserver with debug flags properly, as it doesn't seem to be printing any debugging at all

14:21

...

14:22

cmake -DCMAKE_BUILD_TYPE=Debug -DDEV=On .. doesn't seem to work?

14:25

._. I might be dumb

what do you mean, printing debugging?

I'd like to retrieve the server list from the master server utilizing go, but it fails or seemingly nothing happens and I'd like to debug that

the master server doesn't output anything, neither in debug nor in release mode

15:59

your best bet is probably using wireshark to monitor the port you're communicating on

well, it seems to output stuff when servers register to it

15:59

but nothing when a client requests the list, seemingly

16:00

is the client part also handled in masterserver.cpp?

yes

16:00

https://github.com/teeworlds/teeworlds/blob/44a0260372d170cdd51b78418708e506567f5513/src/mastersrv/mastersrv.cpp#L390-L407

16:00

this is the part where the masterserver sends the list

16:01

wait

16:01

it does have a debug output there (even in release mode)

16:01

sorry, I misremembered

dbg_msg, shouldn't it usually just print when built with debug flags?

n

16:02

o

oh

they always print

is it possible to work with multiple layers at once in map editor?

16:03

Image attachment

16:03

looks like

16:03

but how

check where pEditLayers is set

Image attachment

16:04

i guess shift it is

16:05

but idk how to do it exactly

16:05

in ddnet you can press shift and them click the layers you want to edit and select multiple ones

16:05

in vanilla i dont get anything happening while pressing shift

hm seems like its this area shift draggy thing and vanilla does not support multi layer editing but i have no clue what it is looping over then

i have put some work into the textrenderer. i basically added a caching mechanism and made a few general optimizations i could use some testers with different hardware

for me it gives + 5-40% fps increase on an nvidia gpu and + 5-20 % on an intel igpu. in extreme cases (serverbrowser + console) it can even go up to +80% just to mention: when you are already close to the gpu limit there won't be major improvements (e.g. 4K on an igpu)

17:29

for anyone who is interested... code and artifacts are here: https://github.com/axblk/teeworlds/tree/vbo

axblk/teeworlds

A retro multiplayer shooter. Contribute to axblk/teeworlds development by creating an account on GitHub.

nice thanks @redix

17:44

github artifacts are bae

@redix can you explain how this differs from the similar-sounding (to me, I don't know much about graphics) https://github.com/ddnet/ddnet/pull/861 ?

Pull Request using OpenGL 3.3 by Jupeyy · Pull Request #861 · dd...

A port to opengl 3.3. I will explain what i did here. To stay in the coreprofil i used the opengl 3.3 core profile you will find here: https://www.khronos.org/registry/OpenGL/specs/gl/glspec33.core...

17:45

@ChillerDragon add them for ddnet0.6?

0.6 in 2020 xd

17:45

yea but was thinking about it @heinrich5991

@heinrich5991 it only requires opengl 2.1 and you can easily enable the caching+batching for all texts in tw

17:46

i think in ddnet the textcontainer is only used for the chat

17:46

and the opengl 3.3 implementation switches shaders too often =\

what does that mean? "switching shaders too often"

17:48

aren't shaders usually static for the runtime?

it basically uses different shaders for everything, so you have a lot pretty expensive state changes

forgive me my curiosity: what is everything, what does it e.g. have different shaders for?

17:50

I could imagine: tilemaps

17:50

end of enumeration

https://github.com/ddnet/ddnet/tree/master/data/shader

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/ddnet

17:51

no clue why so many are used

17:52

usually that's just fine since you batch all draw calls per shader together... the tw rendering system does not

17:55

in general: it's no crucial since tw basically runs on a toaster... but it neither helps performance =\

17:56

but the performance is better for maprendering since it uses buffering on the gpu

opengl 2.1 should be alright hardware wise

comparison with ddnet: 750 fps vs 1300 fps in the serverbrowser on my nvidia card

which is which?

ddnet 750

18:05

seems like ddnet only uses the text container for a few ingame elements but not for the menus

hm ok

performance increases a lot in the editor when tile information is shown

i didn't read all the code but it seems like you have to use the textcontainer explicitly in ddnet... so you have to add single elements my approach just records text draw calls, batches them together and just updates the gpu buffer when something changed

18:11

btw: you can disable the buffer with gfx_text_cache 0

18:12

should still be faster than 0.7.4 since it has lower overhead

humm is it me or i notice no changes

18:18

uh well i have better fps on srv browser

18:18

compared to gamer client

well it depends on your hardware... and how much text it shown on screen with an intel hd 620 on wqhd i only get +5 %

18:18

in 1080 it's way more

0.7.4 and gamer i get 400-500fps servers page

18:19

with ur client 1k

ati radeon gpu

18:20

rx580

18:22

and around +100-200 fps ingame

with gfx_finish 0 i get 1900 fps in the server browser :O i thought this was 0 by default...

gfx_finish?

it waits until opengl finished the current frame

19:42

ddnet has 0 by default

19:54

didn't know about that lol

I don't understand what that means

it blocks the thread until all opengl operations you submitted are completely executed... the cpu waits for the gpu to finish all work. it's for example useful to determine how much time certain operations on the gpu took

Exported 92 message(s)