Create an account


Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[SUGGESTION] Xonotic system requirements needed

#26
My specs:
AMD Athlon X2 4800+ @ 2 x 2,2GHz
2Gb RAM
Nvidia GTS250 (1024Mb) @ 1440 x 900

I can set everything to maximum except ´Relief Mapping´ and the game runs all fine. Smile I got about 40-90 fps then. But that also depends on the map: Some are really horrible and slow down my machine even with lower specs, so I would say that it also depends on how good the map is VIS-optimised!
Reply

#27
Maybe it would be a good idea to have some sort of benchmark function included with Xonotic that automatically runs like 3 demos on different maps (official ones only) that show action with several other players? Then the automatic benchmark could perform several runs at different detail settings and measure the achieved fps (I assume writing fps to a log and calculating the average value from that per run would be possible?).

That way we could have comparable performance reports on different hardware, I suspect a lot of posts would come stating hardware and benchmark results. Also, it might get Xonotic more publicity, if it can be used as a standalone benchmark.

I have absolutely no idea about how hard it is to code something like that, it's just an idea I got from all this system requirements discussion.
My Xonstats Profile
Latest track on soundcloud: Farewell - to a better Place (piano improvisation)
New to Xonotic? Check out the Newbie Corner!

Reply

#28
(01-03-2012, 09:14 AM)Halogene Wrote: Maybe it would be a good idea to have some sort of benchmark function included with Xonotic

Yeah, that sounds good.

As for my Xonotic-high loading times, well I don't really know. I'll test it some other time (it doesn't really matter as I'm using Xonotic-regular at the moment and it works perfectly).
Reply

#29
I can run Xonotic on normal settings(lowered particles)average 60 fps(can dive to 30 on certain maps). Here are my laptop's specs:

AMD Athlon 2 P360 dual core 2.3ghz
3 GB DDR3 ram
ATI HD4250 256mb
Windows 7 64bit/Linux Mint 13 64bit (MATE)
Reply

#30
Here's mine. I would consider it to be below the minimum specs, unless you only want to Duel. I get 40-70 fps in a Duel, but that number quickly drops on larger maps, and with more than 2 characters on screen it's 30 fps or less. 10-20 fps is about average with 5v5 CTF. This makes most gametypes unplayable.

Intel Atom 330 1.6 ghz dual core
Nvidia ION
4 gb ram
WinXP 32 bit

The GPU is fine for Xonotic, but not the processor. I play with low settings and picmip.

I'm definitely considering quitting PC gaming because I hate being at a disadvantage all the time. I might just stick to BlazBlue on 360.
Reply

#31
(06-11-2012, 03:18 PM)W4RP1G Wrote: The GPU is fine for Xonotic, but not the processor.
I'm surprised a dual core 1.6GHz Atom is the real bottleneck there. Have you tried anything other than Windows on it? I can get Xonotic running playably on a 2x1GHz P3 system if there is a big enough graphics card in it.

I'm somewhat surprised also with the poor performance some people get on some rather high end hardware. There is a lot to be said for how a system is setup but when we also have ~3GHz systems being dismissed in this thread as being too slow, is it just a slow config or hypersensitivity to FPS?

You might want to have a look through this thread I did some time ago and see if it helps:
http://forums.xonotic.org/showthread.php?tid=2685

What makes a set of specs minimum? The lowest to run the game at all? The lowest to make it playable? What is recommended? It's all down to interpretation and I have got the game running on some very old stuff.

The absolute minimum right now would be a Pentium II 233MHz (or worse, an AMD K6 166, Cyrix 6x86 PR166 (MMX I think?) or IDT Winchip 180 (anyone remember?), I think these are all i686 and MMX), as little RAM as your OS will boot with and something that more or less is able to draw polygons on screen... Lowest I've tried is P3-550 with 192Mb RAM and a Radeon 9000 although seperately I have tried only 128Mb RAM and Via graphics.

To be recommendable, well I have an XP2400+ with a 6600GT and 1Gb RAM which plays on normal at 1600x1200 very well. I wouldn't quite push this to recommended though as a 128Mb graphics card can struggle in some situations, hence 256Mb video memory I would recommend. I'm sure a 7600GT 256Mb would fly.
I'm at least a reasonably tolerable person to be around - Narcopic
Reply

#32
Quote:I'm somewhat surprised also with the poor performance some people get on some rather high end hardware.

Support for multi-gpu setups would be awesome as my single card (Asus Ares) only uses 1 gpu core in Xonotic atm.

Also, animations need to be primarily gpu-bound, instead of cpu-bound.

Bots really need to be optimised.

And reflections too.

t
Reply

#33
Also, solarium has some oddly poor performance for me. I mean inside, so it's unlikely to be water's fault.
Reply

#34
@edh I'm pretty certain the bottle neck is at the CPU. I play with very low settings, lower than the low setting in-game. Things like texture quality don't affect performance at all, but player models and bot AI kill it, and those are known to be handled by the CPU. I've already done all the optimization I can do, so I guess I have no choice but to suck it up and play like that or not play.

I can run Quake Live at a constant 120fps on almost maxed settings though. Too bad I don't care for it. UT2K4 also does very well except for onslaught. I can run that game in DM maps on holy shit settings and still average 50-60 fps.

I'm guessing the reason those games run better is dual core optimization.

Also, I would say if normal matches can't be played without dipping below 30 fps, then that is below minimum specs. Obviously, if someone bought a game because they met the minimum specifications and then went online in a normal match and got below 30 fps, that would not be acceptable.
Reply

#35
(06-11-2012, 10:59 PM)W4RP1G Wrote: I'm pretty certain the bottle neck is at the CPU. I play with very low settings, lower than the low setting in-game. Things like texture quality don't affect performance at all, but player models and bot AI kill it, and those are known to be handled by the CPU. I've already done all the optimization I can do, so I guess I have no choice but to suck it up and play like that or not play.
Have you tried recompiling from source and targetting your architecture? On a CPU bound system there is more gain to be had, especially with something like an Atom which has a lot of instruction sets, not much speed. I'm guessing yours is a Cedar Trail-M:
http://en.wikipedia.org/wiki/List_of_Int...2832_nm.29

Also have you tried anything other than Windows XP x86? Your CPU does support x86-64 which may give a small improvement itself but also as I have found, Linux can substantially outperform Windows especially on the lower quality settings where CPU is more important.

(06-11-2012, 10:59 PM)W4RP1G Wrote: I'm guessing the reason those games run better is dual core optimization.
I think UT2K4 actually doesn't make use of SMP. We already have some SMP use by offloading sound onto a second core and graphics drivers tend to have some parallelisation so it's not like it won't be used at all.

(06-11-2012, 10:59 PM)W4RP1G Wrote: Obviously, if someone bought a game because they met the minimum specifications and then went online in a normal match and got below 30 fps, that would not be acceptable.
There's a lot of commercial games that perform terribly on minimum specs and even when running on a normal level of hardware getting small dips in framerate are pretty normal. We do have the one second averages which if you take a look at may show a min of less than 30fps even when averaging in the hundreds.

One very quick way to boost framerate for a wide section of players on older systems 'out of the box' would be to change the release build -march flag to Pentium 3. This would make a Pentium 3 or Athlon XP absolute minimum, the game would NOT RUN AT ALL on anything not supporting SSE but might give 1-2% improvement for everyone else. I suggested it here and have now added a poll:
http://forums.xonotic.org/showthread.php?tid=3024
I'm at least a reasonably tolerable person to be around - Narcopic
Reply

#36
So, how hard is it to move animations and bot calculations to the gpu? I made a thread a couple of months ago about optimising two cores under OpenGl.
Reply

#37
I bet you it is very very hard. If it wasn't, divVerent already would've coded it.
[Image: 561.png]
"One should strive to achieve; not sit in bitter regret."
Reply

#38
Very hard I would guess. Plus why not put the bot AI in another thread so it can run on a different CPU core if available? Wouldn't that make more sense?
I'm at least a reasonably tolerable person to be around - Narcopic
Reply

#39
AI certainly very hard if bot data needs to be sent back and forth between CPU and GPU for other stuff (like calculating if you've hit or bot hit you).

I don't know about animations. But surely, judging by RPC in Overkill, performance can't be that bad?
Reply

#40
Quote:But surely, judging by RPC in Overkill, performance can't be that bad?

It's mainly animations like swaying trees or objects in the environment that will cause slow-downs. Jump to this post and the post below it for more info.

Quote:Why not put the bot AI in another thread so it can run on a different CPU core if available?

Yes, that would be a good idea. I suppose the CPU and the GPU having to communicate with each other will pose quite a few complications.
Reply

#41
Animations ARE a big cpu hog, this is a known problem. One thing to try, tough it will guarantied be reaaaaly fugly, is typing in: "mod_alias_force_animated blah; r_restart" in the console. This will disable animations altogether and load meshes in their reference pose. Note that you wont see when a player duck, shotty slap etc (as those are animations). So really not recommended but possibly interesting for performance analysis. For me the difference is hardly noticable at normal effects:
With animations: 184 327 689
Without animations: 189 341 705
(timedemo demos/the-big-keybench.dem)
I do however have allot more cpu then gpu. @ W4RP1G It would be interesting to see how the numbers look in your case.

When you play on a remote server, the bot AI obviously does not run on your computer. So the most straight forward way to "run the AI on another thread" at the moment is to launch a server, locally, and connect to it. Not the smoothest of solutions, but possibly helpful for *local* games.
Reply

#42
tZork, I gave that a try and I noticed about a 10 fps average boost. The only thing is, when turning off models animations, the game will get really choppy for no reason at times, and at other times when it would normally drop down to 10-15 fps, it would be at about 30.

Also, I tried setting r_sky to 0, and I noticed a 15 fps boost. This was the one thing from all the fps tips I've read that I've never tried bacause it looks awful.

Whatever the case, a 10-15 fps boost will hardly get me close enough to where I think I should be. I just don't like playing at 40-50 fps when my opponent is at 125. Throw in my 60ghz monitor vs 120ghz, and I just don't even want to bother.

I'm definitely open to any magical tips that will put me at 120 fps if anyone has some.
Reply

#43
To be perfectly honest i don't think Xonotic will ever be able to do 120fps on a system that limited without very, very major update of darkpalces. Yes Xonotic is slower then many commercial game's of a similar visual quality, since we simply do not have the resources to optimize things the way they can. One thing to test, if you have not already, is using vid_dx9 1 mode. Its mainly intended for the terribad intel gpus iirc, but perhaps the ION driver likes direct X better then OGL. Also mane sure you use the SDL binary as it offloads sound to a second core, helping a little with the CPU load issues.
Reply

#44
If you look here, I tested DX9 vs OpenGL and OpenGL is far superior on my PC.
Reply

#45
dx9 crashes the dp engine for some reason. it's the latest version too. Also, idk why, but when I used the sdl exe before and ran the benchmark, I got a lower fps.

Oh well, thanks for the suggestions.
Reply

#46
My laptop:
CPU: Intel Atom 1.70 Ghz
RAM: 1 GB
GPU: Intel Graphic Accelator 3115

Even on omg mode im lagging like hell..... :/

My bro's pc:
CPU: Intel I7 2.80 Ghz
RAM: 4 GB
GPU: Geforce GT 220

This pc is powerfull enough to have 120 fps with Ultimate at a resolution of 1080x7(something).
Reply

#47
I run Xonotic on 2 different laptops. Xonotix runs smooth on both, with video set to 1280x720 and all effects on. Resolution turned on higher will stutter a bit on big arenas (like Newtonian Nightmare CTF).

One is a Lenovo ThinkPad T410:
CPU = i5 2.4GHz dual core
RAM = 4GB
GPU = Nvidia NVS 3100M
OS = Windows 7 64bit

The other is a Gateway:
CPU = AMD 2.7GHz dual core
RAM = 4GB
GPU = Radeon HD 7520G (512MB)
OS = Linux Mint 17.3 64bit


Attached Files Thumbnail(s)
   
Reply

#48
i`ve tested nexuiz on rk3066 with mali400mp4 1Gb ram tablet(android 4.0.X).
on own screen 1024x600 in DM server it was able to give 40-60 fps with 4 players dm on fogged silvercity.
effects settings was default, i.e. normal iirc.
Reply

#49
@rocknrool237 are you using the latest Xonotic releases, 0.8.x? The Git versions have quite a bit longer loading times due to uncompressed data.
Reply

#50
(07-03-2016, 05:21 PM)kaadmy Wrote: @rocknrool237 are you using the latest Xonotic releases, 0.8.x? The Git versions have quite a bit longer loading times due to uncompressed data.

please read the posts.
You are repling to a 4 year old post.
<Samual> I am the most unprofessional developer ever
<bluez> halogene, you make awesome music, but you have no clue about ctf.
<Halogene> I didn't know mappers include some mysterious waypoints so members of the BOT clan can navigate a map?
<divVerent> if you don't pay for a premium account, your movement speed is limited to 100qu/s
Reply



Possibly Related Threads…
Thread Author Replies Views Last Post
  [SUGGESTION] A game package should contain all needed components kanatov 2 4,321 03-10-2015, 10:58 PM
Last Post: kanatov
  [SUGGESTION] Matchmaking system? machine! 14 13,561 09-21-2014, 03:20 PM
Last Post: Mr. Bougo
Exclamation [SUGGESTION] Better Singleplayer AND A better pause system satuim 13 14,049 08-27-2013, 04:47 AM
Last Post: satuim
  [SUGGESTION] A small addition to the spawn system Dodge 25 23,929 07-05-2013, 06:48 AM
Last Post: Mr. Bougo
  [SUGGESTION] XP system for stats cccxd 10 10,334 09-04-2012, 07:43 AM
Last Post: hutty
  [SUGGESTION] Kick idlers only when needed! unfa 12 14,235 05-04-2012, 05:15 AM
Last Post: unfa
  [SUGGESTION] extended gravity system hutty 19 18,893 04-11-2012, 02:58 AM
Last Post: unfa
  [SUGGESTION] some kind of system to schedule matches/meetups jammo 5 6,603 10-09-2011, 04:54 AM
Last Post: lda17h
  [SUGGESTION] Unified icon system to give info about players unfa 18 22,249 11-06-2010, 01:05 AM
Last Post: Rust Heart
  [SUGGESTION] Permanent ladder system, raking in-game PGP_Qz 9 14,384 09-29-2010, 11:27 AM
Last Post: PGP_Qz

Forum Jump:


Users browsing this thread:
1 Guest(s)

Forum software by © MyBB original theme © iAndrew 2016, remixed by -z-