If KTX would start supporting some other strange mods more .ents would probably be needed to change item spawns on "normal" maps. Altough, .ent files are not really necessary since itemspawns CAN be build into qwprogs, but entfiles are so easy to work with. Not to mention that KTX qwprogs would be rather big after a couple of mods .

wow, good trick, but map may require few .ent files so this trick will fail. So adding variable which point where to get .ent files is a good idea and not a hack at all. But now i'm less motivated to add such variable untill we get real requirement of this.

ehm... "few .ent files" --> you mean at least 2 .ent files per map? I'm not sure it's possible to load 2 .ent files at same time :>. Or i dont understand....


Server know nothing about "gametype". So that's ideologically wrong to hack mvdsv this way. If you need something like this:
maps/4on4/dm3.ent
maps/2on2/dm3.ent
maps/ctf/dm3.ent
maps/ra/dm3.ent
maps/ca/dm3.ent
etc

You have to add a command like loadent or something. And load .ent files just like executing per-map configs .


I.e. just add command like "loadent" to mvdsv and make a '.ent'-loading logic inside mod, since it's mod's job.

That seems to be working just fine, thx for info. As long as .ent files are required only by ctf mode it's ok.

To distinguish .ent files for different modes you could use additional file extension, it could be like file.ent.ctf, file.ent.carena, file.ent.whateverelseyouimagine, or something like that .

F.e. CTF have additional media, so when u switch from CTF to non CTF server need reload map, so at the same time we can load corresponding .ent file. Additional media is a common thingie for mods which result in map reload on mods switch, so adding "loadent" command is unnecessary. Anyway add command or variable to server mean we need provide to server some additional info, f.e. how that will be look for ctf for both variants, loadent ctf/dm3.ent or sv_entdir ctf, as u can see we provide gamemode anyway.

How do I get the map cycle to kick in? I have check_maps enabled in my server.cfg. I've tried localinfo, doesn't work either.

Map rotation can be set up like this:

set k_ml_0 "dm2"
set k_ml_1 "dm3"
set k_ml_2 "e1m2"
set k_ml_3 "dm6"
set k_ml_4 "dm4"

Rotation might require

set k_matchless "1"

Maybe qqshka can confirm this?

Thanks. That worked. Didn't think to check the matchless config..

I "converted" QW.CS.FI's servers to run KTX.

The old KTPRO based duel/team ports do not exist anymore (27501-27506 and 29001-29003) but instead there are 8 ports of KTX running at 28001-28008 with 1on1/2on2/4on4 (3on3/10on10/ffa) gametypes.

Heh, KTX may be compiled in to qwprogs.qvm or OS native library (qwprogs.dll on WINDOWS and qwprogs.so on *NIX). We distribute qwprogs.qvm on site because it OS independant, but OS native library is much faster (less CPU usage), that may be critical since u up 8 ports. I dunno CPU usage on ur machine, but probably u will need use KTX in form of native library.

Also, seems u use old build (last release? ). Better to get latest KTX and MVDSV from CVS. KTX in CVS require MVDSV from CVS too since i've made few hours ago some changes in API.

Renzo - can help with compiling mvdsv and ktx for your OS . Have all BSD's and pack of Linush shells

With 1GB of ram and 100Mbps dedicated connection and Gentoo is 64bit. The cpu usage really seems a bit higher (even more than 7% per port) than before. So yes, compiled to .so would be nice. Now I'm off to sleep to get my less than 6 hour rest before I go to get bored in work.

Yeah, compilation help needed. configure/make fails miserably, probably missing something (plain gcc won't do?). Also the cpu% is way too high with 4on4 mode so...

Plz, Renzo, run few ktpro ports instead of few ktx.
Cos ktx unplayable (at least for me).

What files do I need to copy and what .cfgs I need to change?


B1aze what's the problem with KTX? I mean it's either KTX or KTPRO but not both. I don't want to waste time by wondering around two different sets of servermods/software if they begin to crash/bug/whatever.

If u successfully compile qwprogs.so and mvdsv(u really need mvdsv from CVS) and u use sample cfgs, then as I remmber u may need edit ktx.cfg find k_spw variable and switch value from 2 to 3 and use sv_maxrate instead of k_maxrate.

I'v change sample cfgs on site a bit today...

KTX @ *NIX compile how to:
1) get sources from cvs
2) ./configure
3) a: make vm - it will make a bytecode build. output is qwprogs.qvm and qwprogs.map. Second file isnt a must it only needed for debugging
b: make dl - it will make a native code build. output is qwprogs.so

to use vm-code use mvdsv with sv_progstype 2, for native-code sv_progstype 1

P.S. "make" or "make all" will build both native and byte codes

Can't get it to work.

Compiled the latest MVDSV 0.25.02-CVS (l), build 2528 from CVS and tried using both .qvm and .so

EDIT:

Compiled both MVDSV and KTX in other machine, copied them to the server and they seem to run fine (I actually installed ONLY qwprogs.so and renamed qwprogs.qvm to something else to be sure .so file was used) at much lower cpu%.

MVDSV obviously requires the +gamedir ktx to be the first argument in command line. Also something strange is going on with 64bit OS/compilations

oh, forget notice, u must set "sv_progtype 1" // progtype (0 = .dat, 1 = .so/.dll, 2 = .qvm)

Yeah, it is set like that in my mvdsv.cfg.


KTX is a great thing, easy to install and set up (thanks to the help provided on ktx.quakeworld.nu) with lots of features and cross-platform compatibility. However this is where the complaining begins:

- One 4on4 match can consume over 20% cpu-usage on A64 3000+/1GB ram with .qvm, which is unacceptable. Think about a machine half the power A64 3000+ provides and add another 4on4 port...

- KTX can be compiled to native .so but this is where the fun begins. There are no instructions how to do it. Also it requires more than just the GCC (lcc and q3asm obviously) and there are no instructions how to obtain them or set them up.

- No other than .qvm or sources are provided. This could mean admins wouldn't be able to use KTX if they have lousy server machine or insufficient skills to compile KTX themselves (I couldn't, luckily my friend knows shitloads of *nix so he was able to figure it all out... though it took time).


If the KTX crew is able to provide as good instructions for compiling qwprogs.so as is provided for setting up the KTX in .qvm mode things will work. If not, we'll be seing limited number of server ports due to too high cpu usage scattered around.

Thx goes to Empezar, he made site and sample cfgs (with help from mli and disconnect and also deurk), Empezar really great man, huge thx personaly from me.


Yea, btw how much CPU eats .so with all 8 players and match in progress?


Right, only one note, to compile .so u does't need lcc or q3asm, as disconnect mention above makefile may be used for compile both .qvm and .so, so then just use "make vm" for compiling qvm(here u need lcc and q3ams) and "make dl" for compiling .so(does't require lcc and q3asm), but instructions is not present, here u right.


Right, but somewhere in "deep of site" i provide link how to obtain sources via CVS command line


will be provided ASAP.

I'll check the cpu usage later today if people are playing 4on4, now I'm at the work (though runnin ezquake and testing the servers ) but we'll see.

Once the instructions are finished, this is going to be a great mod. Hopefully other server admins have the time to upgrade.

damn nobody of ezQ/mvdsv devs. have x86_64 box =:-(

I have AMD Athlon 64, but doubt that can help somehow.

Install some 64bit distribution (gentoo perhaps ) and it will help.

So I decided to ditch that CS.FI part too, new address (same ports as before) can be seen below my forum nick.


Server cpu usage on 4on4 game seems be about 4-8% nominal, spiking at less than 10% (I've seen 9,5% usage at the highest). So .so is considerably faster then .qvm but yet I'm hoping for higher level of optimization

New report on cpu usage: two 4on4 matches and two 1on1 matches going on:

cpu usage 13-18% on a64/3000+ (1800MHz) with 1GB ddr400.

I have the following problem on my x86_64 GNU/Linux:


ktx 1.32 and 1.33 the same...

anyway, if someone needs accout to this pc for ktx development purposes just drop me a pm

crow, check pm.

thanks for your help, it works in 32bits compat mode as you advised

so our ktx servers are online on dubstep.hu (p: 27500, 27501)

Ze Dutch (Holland mmkay?) server of XS4ALL qwteam (194.109.69.78:27500) is running KTX on test base as well.