[zup100]

I've done more tests on other maps, but first for the purpose of not cheating I set my processor on 2GHz to prevent dropping frequency to 800Mhz. It not helped but I wanted to be more pure and eliminate this issue if somebody will ask.

Also for this purpose I've made completely pure config if somebody want to ask.

sudo cpufreq-set -r -g userspace -u 2001.000MHz
sudo cpufreq-set -r -g userspace -d 2001.000MHz

Processor was always on 2GHz I've checked it by command in the terminal on other monitor showing me every 1 second about state of its frequency:

watch -n 1 lscpu

And the output was always:

CPU MHz:               2001.000

The strange thing as always in any 4.2 version was that htop (also on other monitor) was showing always 100 percent of usage on core 1 and 5 percent of usage on core 2. It was no difference between Quake3-UrT.i386 binary or Quake3-UrT.x86_64. There was the same (forgive me) shitty results.

Bohemia Version 4.2.012


The com_maxfps is max (125) so you see the drop, without any player on the map and not connected to any server. The similar results but of course worse fps with 8 players on this map resulted even to 60 fps drops in the same location seeing only one player!

Uptown Version 4.2.012 Make no stupid comment like "Buy a new GPU" I've made something differed so you can see on this resolution huuuuge drops on Uptown. And the resolution is? 400x300!!! Of course without any players and without connected to the internet server and the same condition like, processor is 2GHz and htop showing 100 percent ussage on core one and 5 percent of usage on core two!!!


87 fps!!! Why? And please make no stupid comment like "Buy a new CPU" Cause players with low end Core i5 or even Core i7 has the same problem. In my opinion (it can be totally wrong) its because even when you have 4 cores, in 4.2 version in usage is only one. And I will repeat again! Urt 4.1 is using always 2 cores on my Debian Squeeze all the time of course with ioUrbanTerror.i386 binary, cause 64 bit binary is useless! And I will repeat again that on Urt 4.1 I have never experienced on Uptown map without players any fps drops!!! The only condition when I have fps drops is on servers with 32 players. Then I have fps drops only to the 71 fps!!!

So lets compare:

1. In 4.2 version 8 players on the map and I can have drops to 60 or more (the test was done with seeing only one player) fps. I can have fps drops even with 400x300 resolution without any player, like 80, so imagine how much drops I could have with more players, smokes, etc...

2. In 4.1 version the only conditon when I have drops is where there is 32 players on the map and I can still play CTF with full smokes (cause drops are occasionally not constantly), grenades, etc. cause I have 71 fps occasionally, the constant is almost always 84 fps! I've made this to be in sync with my CRT monitor so I've done com_maxfps 82 (try it if you have CRT with 85 refresh rate) With Team Survival or Bomb Mode I've never experienced any fps drops!!!

So, when this 4.2 version will be finished? For me its a shame, cause I wanted to play 4.2 version after making cl_maxpackets 84 (so I was able to hit anyone then, not this shitty default settings) but there is a barier of fps drops and I do not know if even I will buy new processor it will be able to guarantee full 125 fps...

P.S.

Yes I have make process priority higher for Quake3-UrT.x86_64 application (even for .i386 binary) in the htop as somebody recommend, the results was the same...

P.S. x2

The only real guarantee in my opinion is to buy hi end processor for 1000 USD or more to resolve the problem of fps drops (of course I do not need a new faster processor for other purposes for my work). I would rather give you donation of 20 USD instead of buying hi end processor, but after spending many hours of testing and configing and seeking the solution without success, I do not see any real reason for doing that! And its the shame cause I wanted to do this!

This post has been edited by zup100: 07 June 2013 - 08:58 AM

[GlaD-don]

2000 mhz c2d is pretty slow man. q3 engine is single threaded, and therefore the only thing that matters is how fast the architecture is, not how many cores you have. the fact you sometimes saw both cores being used only means you used a custom audio engine that supports offloading the audio workload to a 2nd core.


I got FPS drops at stock (3.7 ghz) with my 2500k. It wasn't until I pushed past 4.2 ghz or so that they went away. I'm currently running at 4.5 ghz and the FPS drops are a thing of the past. I run with 32x AA, 16x AF, and 8x supersampling (horrible inefficient--literally draws each pixel 8 times), which is ridiculous overkill. My GPU never goes about 60% usage. This game is very, very CPU bound, which is why your resolution tests don't change anything.

you can build an entire PC for $500-600 that will get 125 fps on all maps. 1000 cpu--lawl.
also, i use a ~3ghz c2d every day, and it's slooow for just regular office task crap. had one at home for 3 years (clocked to 3.6 ghz) as well, and honestly wasn't expecting the improvement when I moved to sandy bridge. if you have the money, upgrade.

[zup100]

don, on 08 June 2013 - 05:23 AM, said:

2000 mhz c2d is pretty slow man. q3 engine is single threaded, and therefore the only thing that matters is how fast the architecture is, not how many cores you have. the fact you sometimes saw both cores being used only means you used a custom audio engine that supports offloading the audio workload to a 2nd core.

Not true!
I'm not using any custom audio engine, only standard original ioUrbanTerror.i386 in 4.1 version.

Not true!
On Windows in my custom timedemo I have 141fps on this C2D. On the same machine on Debian I have 182fps. On Debian I have always 100 percent of CPU usage, on Windows is differend ~70 of CPU usage. Install Debian man and then talk about performance in 4.1 version, but please do not shoot crap from your imagination.

Not true!
In first relase of 4.2 version there was fps drops even on turnpike even without players on the map!!! On current version of 4.2 version there is no fps drops in the same condition, of course with players I have fps drops even on turnpike. So we have some work done, but its only partial -shame...
Again! Please do not shoot crap from your imagination, cause I speak not only for myself but also for players with low end Core i5, i7 who has fps drops. Please inquire deeply and look at the facts, then speak if you want help people!

Its a shame that people like you always talk about partially solving the problem and not giving any real economical solution. Its like saying that if you want to drink water you must buy only water in plastic bottles!

Its a shame that there is no real attention from developers (or its very slow attention) and also from players. And we are tolerating problem, not giving solution or its very slow or expensive solution. Almost always saying that "I don't have problem, so you shouldn't have either. And if you have a problem then buy a new machine, do not expect from developers any work to be done..."

There is some improvements, but its partial. There is also a possibility for more work to be done, cause trend has been established and we have seen clearly some improvements. In graphics, animations, and partially in fps! Engine sucks and developers should be passionate enough to inquire to address the problem! With passion and not only explanation without meaning, 4.2 version could be enjoyable and even better from 4.1 version.

Lack of passion is not only developers problem but also players who tolerate the problem. The need is for speed! For critique and for constructive positive meaningful encouragement. If I and my friends have fps drops I'll post it, look at it, inquire to it, be angry about it, speak about it, as it is! Do it! Help developers see the problem, help others with the problem, help yourself!

[zup100]

First of all I want to begin by apologizing everyone for my harsh language. In fact I was angry, but this is not the reason for trying making you (especially developers) less than me.


Now, with current version with new cfg, more and more I enjoy the 4.2 version. Fps still drops, but I hope that in new coming updates problem will be fixed. This post will maybe help you figure out the problem.

Today I was curious what will happen if I switch off "Core Multi-Processing" in Bios and with the same conditions, but this time only with one core do some tests. The same condition as before with 2GHz frequency but now only one core. After reboot htop showing only one core.

Before restart I've done (as a precaution) timedemo test in my custom demo in 4.1 version the results was (without switching off Multi-Processing") as usual 180fps.


After switching "Multi-Processing" off in the bios and restart the results was this :

1. Urt4.1 -112 fps

2. Urt4.2 -the same fps on Uptown (garden, etc., like in previous posts in pictures, so there was no reason to post the same images), and other maps like before -no changes whatsoever!


Summary is this:

It looks like there is some bug in 4.2 binary in Debian Linux OS affecting priority or system usage. This bug is not experienced in 4.1 version in Debian 64bit, cause compared two core performance to one core this is too much difference to say that 4.1 32bit binary is not using two cores. Maybe technically it not uses but the system does cause 4.1 binary is always using 100 percent of both cores and with one core 100 percent of one core.
So this are the numbers comparing 32 bit 4.1 binary and 4.2 64bit binary (it doesn't matter too much if its 32 bit or 64 bit binary on my 64bit Debian system):

1. 180 fps in timedemo (custom demo), 100 percent of both cores in usage and 112 fps in timedemo, 100 percent of cpu usage after switching off "Multi-Processing" in Bios.

2. 89 fps in garden on Uptown, 50 percent of both core usages and 89 fps in garden on Uptown after "Multi-Processing" switched off in Bios resulting in 100 percent of CPU usage of one core.

So the downside is that performance is worse. The good is that now I can listen music, speaking via ts3, have active browser on other monitor simultaneously, while playing Urt4.2! Without of course affecting the game performance!

[Banned b1naryth1ef]

I suggest trying a personally built copy of the official ioUrT client (https://github.com/B...r-UrbanTerror-4). I believe some of the optimization and stuttering fixes are not in the current build, although I may be wrong. Honestly don is right, it's simply impossible to try and support every CPU model and speed on the market, not to mention impossible to test. There WILL be differences between versions, and you may get better or worse FPS.

I'll try to look into the issue you described above, but I have a feeling it may be a weird edge-case behaviour.

[24/7.x3r]

I am guessing the binaries were already made 15days ago?
https://github.com/B...master/binaries
Use: Quake3-UrT.i386

If not (I assume you have a good level of experience with linux zup100)
1. Download all the files
2. Install development tools for your linux distro
3. cd to the directory you downloaded to (where the make file is)
4. Do 'linux32 make' in terminal
5. Move Quake3-UrT.i386 in the build directory to your UrT folder (always backup your old binary)
5. Play

I would like to add, when I use to use Linux, UrT 4.1 did not play as well as it did on Windows.

This post has been edited by x3r: 28 June 2013 - 04:50 PM

[GlaD-don]

>hi help me plz
>lol fuk u for helping me u r wrong (and i will spew nonsense like "low end i7")
>ok i not get reply, better give backhanded apology

[zup100]

don, on 29 June 2013 - 03:51 AM, said:

>hi help me plz
>lol fuk u for helping me u r wrong (and i will spew nonsense like "low end i7")
>ok i not get reply, better give backhanded apology

I do not expected any reply after my apology. I though that I will leave the subject and wait for new update after league ends. And yes, you're right I wasn't always stable emotionally.


Anyways, thanks guys, x3r and b1naryth1ef for suggesting new compiled binary! Its like surprise, the compiling process its like Christmas present. I'm beginning to unpack it! Thanks!