OpenArena Message Boards

Hi guys!
What about inserting a pre-defined DEMO file with the next OpenArena release?
Do you remember the old times when "timedemo" command was used to run benchmarks with Q3A? Technically, it would be possible with OA, too... (DO NOT LINK) h t t p s : / / openarena . wikia . com/wiki/Demos#Timedemo:_use_demo_as_benchmark
A demo (named "demo088" or something similar?) that all users would have by default, would allow them to compare their systems.

What do you think about it?

Good idea.
But it should be a real demo with defrag for example - or maybe neons demo map

It should be a demo with lots of crap going on. it's a benchmark after all

a player doing a looooooooooong jump is not an idea of a benchmark

hmmmmmm - you are right XD

Defrag? Uhm... It should include baseoa stuff only.

I think it should use a good looking map, include various player models, a good number of players (or bots).

If possible, it would be nice to include stuff as rocketjumps, plasmaclimbs, mid-air rocketing/railing...

some skilled players here who like to do such a demo?
do we have to rename to "OA DemoQPlayer" or similar for this?

It would be nice if some of us of the forum may organize a certain day/time on a certain server to record a such demo, with "us" appearing there.

We could record more demos at that time, from different players pov, and then choose the best one. One of us could even convert them to avi and use them to create an OA 0.8.8 promotional video to post on youtube or such.

But anyway, for benchmark purposes, even a single player fighting against bots could be enough.

Guys, any suggestion about the map to use for that?

And anyone out there that wants to record the demo?
I could try, but I'm not extremely skilled and thus the demo would not show very good things... (and I'm not used to play with g_synchronousclients enabled).

A graphic-heavy map, of course.

oa_shouse, czest2ctf and czest1dm come to mind.

Hi! I've recorded a demo on a clean 0.8.8 RC installation, to make a test.

- My hardware is about a 2003-2004 machine, so it suffers a bit
- I used g_syncronousclients. I'm not very comfortable with it, and when I tried a couple of RJs, they were unsatisfying (I'm having some little troubles with my wireless mouse, too)
- I played locally, fragging a good number of stupid "I can win" bots.
- I played oa_shouse.
- You can see some "WARNING: unhandled AL error: Invalid value" errors appearing even in the demo.

I think it would be better if someone better than me (in skills and hardware) could do a better demo.

Anyway, I attach it to this message if you want to try it for a DO NOT LINK[/b]) h t t p s : / / openarena . wikia . com/wiki/Demos#Timedemo:_use_demo_as_benchmark]timedemo (http://([b).

Here's my result: 

3398 frames 8.0 seconds 426.1 fps

(http://quakefour.net/oa/demo.jpg)

Sounds very good. Mine did it @ little more than 50 fps... @640x480

Anyway the problem with these benchmarks is that there are  a billion of possible video settings combinations (resolution+texture mapping quality+color depth+bloom+filtering+fov+simple items+quality sky+marks on walls, etc. etc.) that it isn't easy to compare two or more systems. Anyway, players could use it to test which settings work better on their machine.

Anone that wants to try to record another demo with many bots?

You can create a baseline benchmark by moving your q3config.cfg and any autoexec, and running the demo with default settings.
Here are some of the relevant bits from my config. (3398 frames 8.0 seconds 426.1 fps)
//Screen
seta r_fullscreen "1"
seta cg_fov "105"
seta r_customwidth "1920"
seta r_customheight "1080"
seta ch_drawspeed "1"
seta cg_drawTimer "1"
seta cg_drawStatus "1"
seta cg_scorePlums "0"
seta cg_drawFPS "0"
seta cg_drawGun "0"
seta cg_zoomfov "40"
seta cg_drawCrosshairNames "1"
seta cg_crosshairPulse "0"
seta cg_simpleItems "1


//visuals
seta r_picmip "0"
seta r_textureMode "GL_LINEAR_MIPMAP_LINEAR"
seta r_detailtextures "1"
seta cg_draw2D "1"
seta color "1"
set cg_centertime "1"
seta cg_oldRail "0"
seta cg_bobroll "0.000"
seta cg_bobpitch "0.000"
seta cg_bobup "0.000"
seta cg_runroll "0.000"
seta cg_runpitch "0.000"
seta cg_shadows "0"
seta cg_marks "1"
seta com_blood "1"
seta r_ext_compressed_textures "1"
seta r_ext_compress_textures "1"
seta r_dynamiclight "0"
seta r_displayRefresh 100
seta cg_lagometer "0"
seta cg_crosshairHealth "0"
seta cg_drawRewards "0"
seta cg_drawAttacker "0"
seta cg_drawAmmoWarning "0"
seta cg_gibs "0"
seta com_blindlyLoadDLLs "1"
seta cg_trueLightning "0.5"
seta r_vertexLight "0"
seta s_mixahead "0.14"
seta r_intensity "1"
seta r_mapoverbrightbits "3"
seta s_khz "11"
seta sv_fps "125"
seta com_maxfps "125"
seta cg_drawCrosshair "4"
seta cg_brassTime "0"
cg_noprojectiletrail 1
cg_altLightning "0"
seta cg_smoke_sg "0"
seta cg_smokegrowth_gl "0"
seta cg_smokegrowth_rl "0"
seta cg_smokeradius_gl "0"
seta cg_smokeradius_rl "0"
seta cg_muzzleFlash 0

With the default config, I got:
3398 frames 9.7 seconds 350.2 fps

I did DO NOT LINK[/b]) h t t p s : / / openarena . wikia . com/index.php?title=Manual%2FDemos&action=historysubmit&diff=9206&oldid=9152]this update (http://([b) to the wiki.  :)

That's a very strange console for a 0.8.8 test.

I can't exceed 200fps in this demo, for some strange reason. That strange reason seemed to be snd_useOpenAL. Now I enter 230fps.

I should note, OA's legacy rendering path isn't exactly the most optimal thing for modern DX10+ level cards. It does not scale well. IF you want faster, get some GL 2.0 action and make fun use of vertex shaders, vertex buffer objects and other stuff.

I should test the benchmark on a P2 system.

A lot of the slowness experienced in this demo come from player models. They use many surfaces. Kyonshi uses many surfaces AND many textures. She causes a lot of dip...even more than Angelyss. Hence, the whole push for OA3.....

personally I think a benchmark demo should be on a small good looking map (oa_shine, slimefac, etc) with a few "best looking" bots (skelebot, angelyss, grism, gargoyle and sergei are all models that look great) - the probable situation, maybe in spec mode going across them. After all, four dm was on a small map (q3dm6) with a few of the id crew.

I did it in that map (that Neon suggested), and with many bots, to give more challenge to computers, especially those recent (you see what high fps sittingduck got, unlike me). My problem is that my computer is quite old...

If you wish, I can try something different... or maybe someone more skilled than me could...

And another problem is that the ioTech3 engine is 10 years old. Newer PCs won't find a big challenge in any OA map, regardless of the size.

My 0.8.8, wtih default settings, did 65.2 fps once, and 64.6 another time, with the demo above.

My system is not top of the line at all, but it's pretty decent.

CPU = AMD Phenom II X4 3.7 Ghz
GPU = GeForce GTX 550 Ti (Fermi) 1GB 192-bit GDDR5
RAM = CORSAIR XMS 4GB (2 x 2GB) 240-Pin DDR3 2000 (PC3 16000)

Done another test.
This runs quicker than the previous one (another map, maybe some less bots)... but gameplay shown sucks.

I think I will try to do one while spectating bots, like Fromhell suggested... but if in the meanwhile you wanna try it.
I scored something around 91.3 fps here, with default configuration...

NEVER SPEC BOTS!
you will get mad ;)

Tested only with my normal config:
2083 frames 3.1 seconds 664.2 fps

I couldn't see much more than a brown blur at that speed, so the gameplay shown is pretty irrelevant.  ;D

Okay, there is a third demo here... it is in slimefac map, spectating bots playing (in spectator mode, not in firstperson!).

I scored 100.2 with default settings, here.

All three of them are released under GPLv2. Excuse me if, for time reasons, I haven't included the "copying" file. UPDATE: Files re-uploaded, "copying" file now included.
@Fromhell... then, do you think you could include a pre-defined demo in 0.8.8?

@Sittingduck: I hoped in a nice looking demo, considering that a pre-defined demo could be played also without the "timedemo" enabled.

2571 frames 3.5 seconds 725.2 fps (my config)

If you want to see something fun, just go watch Gerbil's videos! ;)

http://www.youtube.com/user/GerbaIity

jep, but this another type of demo.
searched our side, but no file found, which could fit the needs :(

@sittingduck

We need a .dm71 file, released under GPLv2 license, done with "vanilla" (stock, base...) OA 0.8.8.

Then, what do you think?

Should I do another one, in the graphic-heavy map of the first demo, but as spectator like in the third?
Or do you think one of these three demos could already be ok and end up in 0.8.8?

Fromhell, you are the leader... what would you like?

Something like that last demo would work just fine. Lots of action, basic map, not too long.
It could just be bots.

Hi! I've done a fourth demo. Again, it is GPLv2 ("copying" file included). What do you think about it?

About demos, a question:
- Settings g_synchronousclients 1 before recording a demo allows DO NOT LINK[/b]) h t t p s : / / openarena . wikia . com/wiki/Timedemo#Recording_a_best-quality_demo]to record a "better quality" one (http://([b). But while playing a demo? I noticed that if you have synchronousclients enabled when playing a demo, "s n c" letters appear inside the "lagometer", and they do not appear if I play the same demo with synchronousclients disabled. Any info about that?

Another question: why does here (http://www.quake3world.com/q3guide/demos.html) they say to set g_synchronousclients 1, then record, then g_synchronousclients 0 and then stoprecord? :-/ In practice, to begin recording with it enabled, and then to continue recording with it disabled. Is sounds strange to me...

g_synchronousclients 1 will will let you record a smoother demo.
It is not required for playback.

I think I read somewhere that this "setting it back to 0" trick doesn't work since even if it was set to 1 when starting the record. Probably can't cumulate the advantages.

Are you referring to the sentence DO NOT LINK[/b]) h t t p s : / / openarena . wikia . com/wiki/Timedemo#Recording_a_best-quality_demo]here (http://([b)?

Si senor