Teeworlds - general

looks like @rand() used more recent C++ versions than Teeworlds generally uses and did not update the cmake files

@Dune ¯\_(ツ)_/¯

1

can some1 help https://prnt.sc/sfbljl

12:26

im not getting this error on windows

never had this type of error

You are probably missing a definition for a virtual function

Windows compile it, but linux not xD

@m!ki is this a modification? if yes, can you successfully compile vanilla Teeworlds?

it is modification

onto the second question?

success on vanilla

I wonder why you and I get such drastically different build times @Dune

12:36

@m!ki it most probably is a missing definition, MSVC can get quite lax by default

@Learath2 I do too, especially surprised that make is faster than ninja for you

12:37

maybe some core detection fuckery?

ninja beats make for me

12:38

slightly though

12:39

I tried bam -j 4 bam was using 8 threads I thought maybe it was some scheduling issue, but that didn't change much

you're on 4 cores ?

4 physical cores 2 hyperthreads per core

12:41

make -j4 and ninja handily outperforms bam on every machine I can get my hands on

especially if you don't specify any options, bam should use debug while make uses release mode

What version of bam are you using btw?

1

master

release mode compiles slower then debug to begin with

yes

12:42

which makes it even more surprising, it's very different from my experiments

@Dune https://gist.github.com/Learath2/d02814920cd3bb1e4c04735a96f2a31f

ComparisonVM

GitHub Gist: instantly share code, notes, and snippets.

13:29

it is possible that gcc vs clang could explain the discrepency maybe? This is on a fresh 2 core VM

13:29

ninja still beats bam in either case but clang allows make to beat bam aswell

Strange that its compiler dependent

13:30

I use gcc indeed

The default compiler is clang on macOS

I also use gcc

it might also be different compilation settings. I see -fno-exceptions -fstack-protector -fstack-protector-strong for bam and -fstack-protector-strong -fstack-clash-protection -fcf-protection

13:39

-fno-exceptions might have a faster build, I don't know about -fstack-clash-protection and -fcf-protection

Will NetEvent's Angle param be used at the near future?

for me bam is faster than cmake+ninja in release mode on linux in debug ninja is slightly faster on windows ninja is always faster for me

that reproduces my experiments on linux

@redix what kind of system? Do you factor in the config time? How many cores do you have?

Intel i5-8250U @ 1.80 GHz, 4 cores + SMT, config times excluded, arch linux vs win10

Also are we testing the same thing? I run bam config then just bam, compared to cmake .. then ninja

i'll add my data to the github discussion

My shitty archlinux vm with 2 cores still beats bam with ninja, though not as significantly as the 4core host

@Learath2 you should run bam config=release though (and that should give even worse results)

On the host that adds a flat 20s

14:37

Didnt test on the vm

I don't understand how you get this though. bam uses all cores?

Seems to, it does use the hyperthreads which I thought was an issue, but even if I limit bam to 4 cores it doesnt change anything

@redix slowdown on windows might be due to the download script?

yes, the download causes a small slowdown bam is also copying the data directory at build time. On windows this is really slow =\

15:49

when i keep the freetype+sdl files and use bam client; bam server it's still slightly slower than ninja

interesting

I repost my issue with rand's code here with more details oke i have a problem while compiling rand's mod using the last version of visual studio i asked to rand() and he replied that it was an issue coming from vanilla source code, and an update on this code might be the problem u can find the code here https://github.com/nheir/teeworlds The following line (line 814 in server.cpp) Msg.AddString(pCommandInfo->m_pName, IConsole::TEMPCMD_NAME_LENGTH); triggers the following exception

Unhandled exception thrown: read access violation.
pCommandInfo was 0xFFFFFFFFFFFFFFEF.

i don't rlly know what does it means, my programmers skills arent good enough P.S. : u need to write set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /std:c++latest") in CMakeLists.txt at the top of the file otherwise u would have an other error while compiling so if u have any concrete tips on how to solve that, im not good enough at programming to solve it by myself

nheir/teeworlds

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