Follow Slashdot stories on Twitter

 


Forgot your password?
Close
typodupeerror
×
Quake First Person Shooters (Games)

Linux Q3Test 1.07 128

Posted by justin++ from the woohoo! dept.
ozium writes "Q3test 1.07 out for Linux. blah blah get it blah blah lose sleep. www.quake3arena.com " Finally! I've been dying, since all the different Q3's are incompatible. For the last week or so, Linux users have been unable to play on just about all the public servers out there. I wonder how much faster this one is...
This discussion has been archived. No new comments can be posted.

Linux Q3Test 1.07 More

Linux Q3Test 1.07

Comments Filter:
  • Thanks, pulled it in at 50kbyte/s :)
    FunOne

  • Visual quality (Score:1)

    by Anonymous Coward
    In addition to the framerate being essentially the same, I noticed the visual quality was a little better under Linux on my V2 than under Win98. In the first Windows release, the 3dfx driver had major problems with portals, light halos, and explosions while the Linux driver had no such defects. With the current Windows drivers, things are much better but I still get some ugly patterns in smoke that are less noticeable under Linux.

  • Have you loaded the 3dfx module? If so, is the permissions on /dev/3dfx ok? (Be warned, a malicious user could crash the machine by poking into it)

    Glide will fall back to mmap of /dev/mem for control of the 3dfx card if it can't use /dev/3dfx

    /AE
  • If I had a nickel for each time I've responded to this question on 3dfx.glide.linux..heheh. If you've downloaded and installed Darryl Strauss' libraries for the V2, then all you need to change is your q3config.cfg file. There's a line in it which says seta_indga bla bla=1. Change the one to a zero, save, and start it up again. Should work. Also double check to make sure your resolutions/color bit depths are defined for whatever you want to game at. If your X11 config doesn't have modelines for 1024 and that's what you're gaming at, you'll have a big problem.
  • Daveo has had a record of posting off-topic and annoying posts (which is why he now starts out with a score of 0), so some moderators routinely mark him down. I would assume that's what happened here--they just saw "Daveo" and marked him down--even though it was clearly unwarranted in this case.

    To be honest, this is the first intelligent post I've seen from Daveo (IMHO: though I haven't exactly been keeping tabs on him, either). If you take a look at his posting history, you might understand better what I mean. His quality of posts seem to have gone up since when he first started, but his spelling is still atrocious. The spelling is bad enough that it is hard to understand what he is saying and makes him look careless and sloppy. On the whole, the moderators have been following the guidelines by marking down posts that don't make sense (i.e. are hard to read) and detract from the discussion (not always the case, but his earliest ones definately did--enough to get him a base score of 0). Specifically, this may have been tagged as flamebait because his speaking in the third person draws flames.

    PS to Daveo: is there some reason you post in the third person?!?

  • Actually, I happen to like my ISP. They provide a static IP, their uber-linux friendly (Unlike BellSouth ADSL) and encourage me to run servers behind my connection (unlike BellSouth, who run screaming if you even hint at running multiple machines). The 2GB/mo limit is fine with me -- Im a casual user who doesn't so much online gaming. If there's still servers for it, if I want to play a game, I'll just go on the net with good old Quake (the first one). I was just curious as to how much data a session of gaming like this would pull...

    --
    Dave Brooks (db@amorphous.org)
    http://www.amorphous.org
  • Check out:

    http://glx.on.openprojects.net/

    The FAQ mentions how to get anonymous CVS
    access to the nVidia GLX module.
  • Zoid,
    Is there perhaps a way you could list some id-suggested optimizations that people can do on their linux boxen to improve performance? There are a few 'q3a on linux' pages out there, but you guys have a unique perspective...

    d

  • Re:Er... (Score:1)

    by greg ( 1058 )
    Your comparison of ADSL to @home cable modem service is innaccurate. @home's service let users consume as much bandwidth as they wanted until the 128K download cap was instituted. DSL specifies guaranteed transfer rates and charges a monthly fee based on the level of service in the contract. This means they can put limits or fees on volume (i.e. >2GB per month) but they can't lower your download or upload cap like @home did because its specified in the service contract.
  • eh? All my tests show 1.07 is very nice about using CPU. I have a server running on my gateway Linux firewall box. With eight players on it, top reports:

    21369 zoid 12 0 14640 14M 816 R 0 69.4 20.3 473:39 linuxquake3

    So it's using 70% of the CPU for eight players (the box is a lowly P2-233). When there's no one on it, it consumes very little CPU.

    I'm curious as to how you are running the server. i have seen problems where people use nohup to run it and it gets confused when stdin is closed (you log out) and ends up spinning on the select for stdin. Thought I fixed that tho.

    I run all my servers under screen so I can get quick console access to them from anywhere.
  • I have SMP code written (and it seems to work well) but there isn't any SMP support in the 1.07 binary.

    When adding thread support, linuxquake3 became dependant on libpthreads. I didn't want that dependancy since I found a binary linked with glibc 2.0+libpthreads doesn't run on a glibc2.1 system+libthreads.

    The SMP code is still a tad buggy and I need to do more testing on it. :)

    I may distribute a seperate linuxquake3-smp binary in future builds. I'm just worried that I can't build a universal SMP binary that will work on glibc2.0 and 2.1 systems. Bummer.
  • Is there hardware OpenGL support on the Alpha? I am planning on supporting it as a server platform, but unless you can get hardware OpenGL working on it, there's not much point in compiling a client.
  • I'm sorta shaken by ID's delay in releasing builds of q3test for linux and mac. I hope this doesn't become an ongoing thing with updates once the retail product comes out, considering it's paid for.
  • To my knowledge, the earliest version of Quake2 we saw had version 2.92 or something, so the number sare not quite as bad as they seem.

    Quake2 has had 4-5 patch levels since release, QW has had about 10 versions (but it was always labeled as an unsupported work in progress).
  • ...well think again; apparently not all of them are fragging away all the time. Check it out. [q3center.com] I guess everybody does have to sleep from time to time, contrary to popular geek wisdom. Or is it that Quakers' CPUs are so brawny that they can run SETI@Home and Q3A concurrently without problems ?
  • I just finished downloading it from ftp.cdrom.com and found it pretty slow (~15 k/s). Those of you living in Sweden or nearby with reasonably fast connection to the net should be able to get it much faster here [sprit.nu].
  • Do I need a voodoo or 3D acceleration card for this? I've got my 8MB AGP Rage thats a 2D card that I'm just happy with. No, I wont be getting 30,000fps or whatever it is...

    Also: If I were to play quake3 for, say, an hour, how much bandwidth would I be expected to pull? My ISP limits my ADSL useage to 2GB/mo without imposing $.01/MB afterwards.



    --
    Dave Brooks (db@amorphous.org)
    http://www.amorphous.org
  • that sucks! whos your ISP? what you do with your
    bandwith is your buissiness, not thiers.
  • When running ./linuxquake3 with the +set dedicated 1 option, it uses 99.9% CPU time, according to top. I had hoped that it'd been reduced, as at the moment the q3test server seems to interfere with the quake2 servers running on the same machine.

    Does anyone have a good solution for this? Would renicing the process help or will the performance drop when running on nice level 4 or something?

  • q3test 1.05 runs fine with glibc 2.1.

    OK, well, q3test 1.05 runs fine on my machine, which includes glibc 2.1 (Debian), so that suggests it isn't a problem.
  • OIC - I read more into your post than I guess you intended, sorry about that.

    That is an interesting point, I never noticed that before (quake 2 shipped out at 3.05 if I remember). I assume you are correct that q3 is derived from the same code going all the way back to quake 1, so it probably would make sense.

  • I think the G200 GLX module is somewhat working on the Alpha, check the g200-dev list for more info...
  • Except that the average person is not going to
    be able to put this together into their own
    system. I myself could probably could, but
    I'd bet about 95% of the computer users out
    there would go out to get some dip for their
    new computer chip...
  • All this talk of SMP had my hopes up too high.

    Well, anyway, 1.07 runs much smoother then 1.05 did on the same system. I can't wait to see what SMP does for it.

    Thanks and keep up the good work!
    hypnotik


  • well its not the glibc thing I have rh6.0 and it runs fine for me. Either 2 Voodoo2 SLI or under TNT.
  • My 120Mhz i80586 runs Q3 very well compared to win32 boxes @ 200Mhz. I (re)nice q3 to -17 and use a 3dfx Monster3d to the graphics hardware. When I play on the net I kill the web and ftp servers for even more speed.

    They need to change system "requirements". You don't need a 3dfx card to play in linux. You can be on a fast machine and do software emulation.

    Damn win32 centric specs!
  • Actually it *could be a problem. I've seen some q2/q3 so's built in 2.1 that will seg fault running with a 2.0 execuable.

    Please email me if you noticed this as well, I've alerted Zoid that he may have to build a glibc 2.0 and a 2.1 release. If this is true and you have mod authors building 2.1 sos for a 2.0 executable it would be bad. =)
  • I have been compiling the tnt drivers from the CVS server, and even with MesaCVS, it is slower than the drivers that came with the rpms pointed to on mesa3d [mesa3d.org]. About 15% slower with q2. They however do fix one bug: xscreensaver gl screensavers don't flicker.
  • The GLX CVS drivers aren't meant for MesaCVS, they are only (really) supported with Mesa3.0 .. try it with mesa3.0. I'm not saying its gonna be faster, but it might be.
  • Perhaps you should try mesa 3.1 beta, it has
    many acelerations for 3dNow!
  • With all the supposed improved net code, i still can't get on a server with 60 ping on my cablemodem. =(

    maybe the CPU overhead is too high.
  • Yeah, okay. Well 1/2 of them were still LINUX servers with Q3Test 1.05

    Given id's track record with patches, where the last Quakeworld one was 2.33, Quake2 was 3.20, do you think the last Quake3 one will be 4.x?
  • Well obviously your cpu power isn't enough.

  • I'm frankly amazed...K6-233, 64Mb RAM, stock Voodoo1... q3test1 lags in large areas and with many players, but the smaller areas run at a quite playable speed. And tried q3test2 in a one on one deathmatch against the roomie on his super spiffy Windows box, and except when he was up really close to me, again quite playable. I forget what the "suggested requirements" are, but I'm really impressed. With all the money I've saved using free software, I can probably afford a new video card for Christmas!
  • You must have a slower CPU.
    I got a PII-450 here and i get 25ms on my Cablemodem.

  • Re:Framerates and such? (Score:1)

    by Anonymous Coward
    Nope. ATI is a no go.

    For now, Linux users are best off with a Voodoo3 -

    I have an AMD K6/2 300 with Voodoo3, and it is quite nice. X windows is faster than it was with the Diamond Stealth S3-968 I had before, and I get a good 30 FPS in Q3 test. It would be higher if I had an Intel chipset, or if the Linux GL driver gets 3DNow! support.

  • Re:"Just about all the public servers?" (Score:3)

    by Jburkholder ( 28127 ) on Tuesday July 20, 1999 @06:22PM (#1792896)
    Just what exactly are you getting at? Yeah, so id releases a lot of patches for their games, so what? Since this is a *test* and they are making an effort to get feedback and patch it before its a released game instead of pushing it out the door, maybe we can expect a better product when it ships?

    Post-releases are a fact of life, unfortunately. The only games I can think of I've ever bought that didn't have any patches were some of the LucasArts games.

    God I bought Unreal almost a year ago and there have been how many patches and its still not as playable as it should be. I think id probably has a better track record than most when it comes to adressing problems after the game is out.

    Yeah, it sucked when Q2 had so many patches right after it came out, but it wasn't the end of the world. Q3 will have patches, probably not as many as Q2 did, since they are doing a much more extended 'test'. And hell, we're getting to play the game and give them feedback. How many other game companies do that?
  • Does anyone have any comparisons between Quake3 on Linux and in Windows 9x or NT 4/5? Im not talking Linux is faster because its better comparisons.. Hard facts comparisons [preferably with a p350 and ati rage pro chipset.. :-)]

    Heh.. thanks
    Stan "Myconid" Brinkerhoff
  • My 386-33 gets 1.2 ms ping over 10base2 (16 bit isa ne2k) and my 300MHz celery gets 0.6ms ping (16 bit isa SMC Ultra) on the same network (three machines: 386-33, 486-66 and dual celery 300 (haven't oc'ed yet)). CPU speed does NOT account for 35ms difference in cable modem pings. Either you've got very good connectivity through or provider, or he's got very bad through his. I'll bet if I had cable modem the difference in ping times for my 386 and dual celeron boxen (and my 486 too) would be lost in the noise (ie +/- 1-2ms).

    Much greater than 1ms differences in ping times will be caused by bandwidth (or shoddy modems, had that problem), not by CPU spead.

    DISCLAIMER: this is for regular 56(? default, anyway) byte pings. I know that cranking up the packet size slows down the response (indefinite for ~65536 (don't know the exact size) to certain boxen:)

  • Good point. I'd thought of that after I'd hit `submit'. I was getting the impression the original poster was talking about normal unix pings, not pings from within Quake (didn't even realize you could do that).

    Anyway, wouldn't the latency be more a function of server speed rather than client speed, at least for a ping? Probably depends on the implementation of the client side.

  • The problem is, there's latency caused by framerate. Quake draws a frame, processes packets, draws another frame, etc... So it is very heavily CPU bound in that regard.

    This is why Half-Life uses the word Latency instead of Ping.

  • Re:Slow CPU - Not likely (Score:5)

    by Zoid ( 8837 ) <zoidctf@gmail.com> on Tuesday July 20, 1999 @06:49PM (#1792902) Homepage
    An important thing to know is that the ping time as shown in the game does include your frame rate. If you are getting 20fps, that's 50ms added to the trip time to the server. Add a good 30ms to get over a WAN link and you're sitting at 80ms.

    But if you are getting 100fps (not unreasonable, I get 100fps pretty consistently on my Voodoo3 on a P2-450), that's only at most 10ms added to the trip time.
  • With Quake 1, 2 and 3, ping is how long it takes for a packet to go from the client to the server, back, and processed. Even over a LAN with 0.2ms ping times, a P133 with a Voodoo1 can get 100ms pings in heated action.
  • (This is for Rendus as well)

    Thanks for the explanations, guys. I'll just go find a crowbar to remove this foot from my mouth. However, I did learn something new (and got a response from a net.demi-god:), so I'm not in the slightest upset. A little embarrassed though...

  • Is there any offical word on when SMP support will be included in the linux builds? I can understand
    that its probably something that no doubt requires
    some linux specific code and thus is probably
    sort of backburnerish after getting most of the
    win32 issues sorted out but still painful to see
    SMP support for NT and not linux.

    -N

  • Re:controlling frame rates (Score:3)

    by armybob ( 70831 ) on Tuesday July 20, 1999 @07:02PM (#1792906)
    thats because Q3s netcode isn't connected to the players framerate. Just use the /rate command to set your proper speed. Typical

    33.6 = /rate 2800 or 3000
    isdn 64k = /rate 5500 or 6000
    isdn 128k = /rate 10000 or 12000
    t1/dsl/cable = i think you get the trend on /rate figure out your sustained bandwidth & set it up.
  • This is a perfect time for me to pleadingly request help with getting my matrox g200 card to work with quake 3 / linux. It works fine under windows, but in linux I get a sig 11, if anything at all, with redhat 6.0. Does matrox opengl work at all in linux? I downloaded the rpm of q3test, and it installed alright.

  • glide? (Score:1)

    by DAVEO ( 61670 )
    where is daveo to download the dependencie 'glide2x.20'? he didn't see a link on the download sight. thank you to anyone who helps daveo, as he is *very* anxious to play his new quake3!!:-) # rpm -i q3test* error: failed dependencies: libglide2x.so is needed by q3test-1.07-glibc-4
  • Just out of interest, why was this post marked down as flamebait? It seems like a valid question for someone who has never had to deal with 3dfx's glide libraries (maybe because they don't have a voodoo card).

    I also wondered why the package seemed to require the glide library when I downloaded an older version, since q3test was programed with GL rather than glide (why do they include a 3dfx mesa library anyway?).

    I guess some moderators don't follow the guidelines.
  • I haven't downloaded the latest version, but I got 1.0.5 to run on my g200.

    You will require the matrox glx driver. There are RPMs of it at ftp://ftp.sus.mcgill.ca/pub/glx/. You will then need to add the following to your XF86Config file:

    Section "Module"
    Load "glx.so"
    EndSection

    After restarting X, you should have glx working. Now delete or rename /usr/local/games/q3test/libMesaVoodooGL.so.3.1, so that q3 will not try to use it. Now start q3test. It will probably be a bit slow and the sound will not work correctly, so set the detail down in the settings. You should be able to quake now. It is probably best to play around with the settings to find what works best for you.

    Note that there seems to be a few memory leaks in the GLX driver, so after playing, you might want to get out of X, and maybe even comment out the module section in the XF86Config file.

    There is more information about the GLX driver at http://glx.on.openprojects.net/faq.html

    BTW, you should also be able to play glquake and quake2 in GL mode with the driver (they both run very smoothly).

    One last thing, turn off xscreensaver if you are using it -- it will stuff things up graphics wise if it starts while you are in q3test.
  • I just finished downloading it from ftp.cdrom.com

    And I wondered why I got miserable rates for some software download from ftp.cdrom.com 8 hours ago despite having a 8kb/s ISDN connection .. that were you hordes of quake players!

    They should consider putting a warning in their banner ('Beware new Linux XY arrived') like a traffic warning. ;-)

    To make you snigger more: Yesterday I downloaded the old version - looks like I have to suck 22 megs again! :)

  • guess you weren't watching too closly which came first. Releases in order for 1.03:

    MAC,
    Linux,
    then
    Windows

    with 1.03 I had a good week or so of gloating to my office mates... heh. Be happy it's here at all.

    -free software's biggest threat is when people start expecting it.
  • I feel pretty honoured as I logged on to his server and played him one on one.

    I did not get a frag. Not one.

    But on the bright side, I did not finish in the negatives!
  • Check out Kingpin for Linux -- recently released.
    (its an unsupported beta patch, but it's a Linux binary, ok?? :)

    Oh, and Max Payne will be dope as well. Hopefully we'll see a Linux port.
  • Well, actually I played against a few people, and it only got real bad just when someone was entering (well, after I turned some of the effects down.) I was overly suprised at how well it did hold up, and the fact that after I installed the v2 drivers it came up the first time (it didn't do that on my celeron 300 + tnt.)

    matguy
    Net. Admin.
  • Either that or put some throttling on certain directories (now that I think of it, is that possible?)

    matguy
    Net. Admin.
  • While reading through the comments already made, and see ppl brag on their 400+ MHZ machines with Voodoo3, and cable modems and etc, I'm wondering if the current trend in minimum requirements in game computers is surpassing what the average game buyer will be able to afford or willing to upgrade readily.

    Quake 3 is not over this curve; I've got a 200MHZ machine with 64 megs RAM and a TNT card with 16megs video ram and a 56k modem; sure, I'm not getting 2 ms pings (more like 200-300) and frame rates are in the 20s, not 100s, but *IT IS STILL QUITE PLAYABLE*. The same goes with games like Half-Life and Unreal.

    However, as game designers continue to want to add more features, I'm concerned that they are going to forget that they actually have to make a sale to a consumer, and are hurting themselves in the end. While Unreal, for example, is playable on my system above, and given that only recently has the support for TNT cards been added, it's still well known that the game plays several magnitudes better with one of their suggested configurations: a Voodoo2 board with a 300MHZ PII processor and 128Megs of ram. Unreal came out just more than a year ago, and yet, I doubt that this configuraion can be claimed as average by consumers.

    Maybe the increasing requirements for hardware are why games like Q2 (nicely playable on a low end pentium machine) are more popular for netplay than these new breads.

    And what also gets me is that these game programs are getting more out of control of than more useful programs: the OS, office applications, and such. I would not be surprised to see some game in months to come to require more from the hardware than what Win2000 will require.... which really hasn't been the case until now.

  • We had this problem at the Beatdown [planetquake.com], and had to run our tournament on Q3Test 1.05 because of it. Silly id Software. You'd think they would have tested the dedicated server before releasing it. Ah well, hopefully it'll get fixed soon (i'm just suprised it didn't get fixed in 1.07)
  • No, it is not faster that I have seen (timedemo is broken in 1.05) but for some reason it *looks* a whole lot better on my machine in linux.

    I have a PII-333 running a V2 - dual booting W98 and RH 6 with kernel patched to 2.2.10

    Both OS seem to give me the same framerates (> 40 on average it seems), but the real difference seems to be in how it looks (subjective, for sure). I can't explain it but it seems clearer and crisper running under Linux.

    My brother has a nearly identical box to mine running NT, and he agreed when he came over and played both that the Linux game seemed to just *feel* better. I know this may not help much, there's no way to measure this - just my $.02

  • Your comments made me think of something, that the server could be waiting for some input. And yes, that turned out to be the problem. Thanks and keep up the good work!!
  • I was playing a couple of lan games on my P166 pre MMX and Voodoo1. Sure I had to turn off a couple of the options, but it ran smoothly. Congrats to id on that.
  • No it's ATI's fault for not releasing the hardware specs so we can write an opengl driver.

  • Some hardware sites have been releasing scores for Q3Test SMP on NT lately, such as Tresh's FiringSquad [firingsquad.com]. The video card and driver seem to rapidly become the major speed bottleneck for Q3 at resolutions above 640 x 480. The second CPU doesn't come into play properly except at low detail/textures. With video card drivers and OpenGL libraries for Linux at a relatively tender stage of development and no threads in the kernel, I'm curious if you expect more, the same or less of a speed improvement in a Q3 SMP Linux.

    I was running some timedemos myself on 1.07 earlier to gauge the effect of SMP. The results so far are somewhat underwhelming. I didn't take Linux results because I've only set that up to run with my old Voodoo and it has no SMP anyway. I expressly did also try a slower PCI video card to magnify the bottleneck effect.

    Scores 1, 2 and 3 were timed on a Diamond Viper 330 PCI with Nvidia's 128ZX chipset and 3.37 reference drivers. Scores 4 and 5 are a Viper 550 with Nvidia TNT and 2.08 ref. drivers. Score 1 is Win98, score 2 and 4 are NT 4.0 Server SP5 with R_SMP 0, score 3 and 5 with R_SMP 1.

    ]timedemo 1
    ]demo q3testdemo1.dm3

    W98, NT4, SMP, NT4, SMP
    43.7 , 48.0 , 51.8 , 73.3 , 92.0 , FASTEST setting
    33.9 , 34.3 , 36.7 , 65.5 , 76.5 , FAST
    21.8 , 21.2 , 21.8 , 48.5 , 50.1 , NORMAL
    14.7 , 14.2 , 14.2 , 20.3 , 20.3 , HIGH QUALITY

    Epox KP6-BS / 2x550 Mhz Celeron(366)
    128 MB PC100 / 3.2 gig DMA IBM HD
    SB32AWE / NE2000 ISA
    NCR875UW SCSI PCI

    I usually play at NORMAL graphics setting and with only a speed improvement of around 0.6 to 1.6 frames per second depending on video card, this particular second CPU is going back to cracking RC5DES...

    Michiel
  • I tried chmod 5755 on every library and the executable linuxquake3 but you still need to be
    root to run the damn thing
  • I actually had a similar question when we first got our T1 at work, we wanted to figure out how many clients it would take to fill the 1.54mb/s bandwidth based on the fact that people run fine over standard 33.6 modems and are really mostly blocked by latencey and not saturation. It would take some 500 users to saturate our line if we had enough servers to support that number of users, then there would still be room since they still don't use all of their bandwidth. The main differences for you is that on adsl you have more bandwidth than a modem user and more bandwidth will result in more data moved, but you are a single user and should not have any problem. In all seriousness you should not even see the amount of data moved by much of any online game unless you are running a server, and even then it's about negligable. You probably consume more bandwidth surfing the web.

    matguy
    Net. Admin.
  • Something to keep in mind with Quake II patches is that the Quake II engine has been used over and over again in games from other companies (Half-Life, Kingpin, etc.) So when they patch Q2, they're not only patching the game you bought, they're patching the code for several games that have yet to be released.

    The same holds true for Unreal (Klingon Honor Guard, the next Duke Nukem), and will for Quake 3 as well.

  • Well I wouldn't call myself a net.demigod *Grin*

    Sorry, just had to do it.
  • The q2test RPM contains a copy of mesa compiled for the 3dfx cards, and hence requires the 3dfx glide library (I can't remember where to find it -- try searching for it with rpmfind or gnorpm).

    If you have an alternative video card, such as a TNT one or a G200 and have glx drivers or a mesa port for your card installed on your system, it should be safe to use the --nodeps flag to install the RPM. After installation, delete or rename the mesa shared libary from /usr/local/games/q3test, so that the correct libGL is used.

    If you do not have a 3d accelerator with linux support, don't even bother downloading the linux q3test.
  • Well, from what I have seen on my machine versus my friend's (both identical in everyway, we built them together) they are ALMOST the same:

    We are both running PII-400s with Canopus Pure3D IIs, he runs Linux, I run Win98. (I have a Linux box too, don't worry. :-p)

    As far as FPS, they only differ by around 1 fps here and there.

  • Re:darn it (Score:3)

    by matguy ( 7927 ) <matguy.oblivion@net> on Tuesday July 20, 1999 @07:49PM (#1792942) Homepage
    I have it on my lan Party server at

    http://www.aeigames.com/party/

    It's on our T1 here at work, so be nice. (I know the server is NT, I don't have a choice in the matter, it's at work.)

    matguy
    Net. Admin.
  • Well, how about this, I got it running on a p133, 32mb and a vodoo2. Running well mind you, playable even with the pretty stuff turned on, stable as hell too. That was v1.05, not sure what has changed for that since then. Seems to be easiest with a voodoo2.

    matguy
    Net. Admin.
  • I was mistaken about glquake. I had not actually tested it out. Quake 2 with GL works fine.

    On closer inspection of the glquake executable, you are indeed correct about the fxMesa stuff. To get it to work correctly, you would probably need a combination of LD_PRELOADing some special libraries, and creating a few dummy ones to satisfy some dependencies ("gcc -shared -o libglide2x.so -Wl,-soname,libglide2x.so" is a start).

    One preload library would have to convert the fxMesa* calls to the equivalent glX* calls. You would also need a replacement for libvga.so that uses X for input (it would probably also have to create the window for the game as well). Some of the GGI source code may be useful here.

    Of course, this is a lot of work just to get one game to work on other video cards -- it may just be easier to wait for id to release the quake source, or write one of those illegal glide wrappers that map to opengl/glx :)
  • http://glide.xxedgexx.com/3DfxRPMS.html [xxedgexx.com]
    go get what you need :) oh...and sometimes (a lot of times for me) the .tar.gz is easier than the rpm (esp true with quake2/3)

    -----
    If Bill Gates had a nickel for every time Windows crashed...
  • Posted by NJViking:

    What if you have a 1-way cable modem? I get download speeds of up to half a T1 thru a T1 at times, but sending goes through the phone line as a (maybe 28.8/33.6?) Kind of pitiful, but my cable co. (RCN) won't upgrade until 3Q/4Q of 2000.

    -= NJV =-
  • ... and I would appreciate some help. I was running XF86 under rh5.2, and q3test (1.05) ran very nicely there. I've since upgraded the software to rh6.0, and I'm running Gnome. q3test (either 1.05 or 1.07) fails, even though I have the same libs (I think) and definitely the same video hardware (Matrox Millenium, Voodoo2). Here are the final messages printed by linuxquake:

    ...loading libMesaVoodooGL.so.3.1: Initializing OpenGL display
    ...setting mode 3: 640 480
    Using XFree86-VidModeExtension Version 0.8
    XFree86-VidModeExtension: Ignored on non-fullscreen/Voodoo
    Using 4/4/4 Color bits, 16 depth, 0 stencil display.
    XF86DGA Mouse (Version 1.1) initialized
    GenuineIntel cpu detected.
    Received signal 11, exiting...

    Any suggestions ... THANKS!

  • I've been doing that...initially by accident, but that first time, I didn't notice until later. My (entirely uninformde) theory is that the video card (I have a TNT) is doing most of the grunty stuff.

    Although, from what I've read here, I suppose it may be adding to my ping. Maybe I'd be getting 350 instead of 400ms.. ;-l

    (33k6 modem _and_ I tend to play in Australia because the only local (NZ) server I know of is often empty)

    At any rate, with my P2-350, I'm certainly getting quite reasonable frame rates with SETI@Home running in the background. YMMV.

    --
    Repton.

  • Has anyone successfully run q3test on a nVidia
    based card?
  • That seems odd because the Viper 770 seems
    to have double the hardware power. Is the TNT
    driver just that far behind the 3Dfx driver or
    is it more of an architecture problem that XFree86
    4.0 will solve?

    Did you compile the TNT Mesa library yourself?
    Did you use the latest release or CVS archives?

    I found this educational:

    http://reality.sgi.com/ripperda_engr/glx/
  • I'm reasonably sure that it is supported now.

    I have a dual PII-333 at home, and as of this morning, with Q3Test 1.07 it was running both CPU's at max. I was averaging somewhere around 30fps on 800x600 with alot of detail.

    However.. I could be wrong.
  • ya... that's what I ment -grin-

    At any rate, the point is still valid... there were MAC/linux releases before any version for windows were available, and if they decide not to support any other platforms at all... it's their decision. People have the tendancy to become accustom to generosity of others and expect it, when they shouldn't. Habituation... it's going to screw up a lot of things.
  • Right, I know about the ps2rate issue. In fact, I got a USB mouse to solve the 40hz sampling issue b4 the ps2rate program became well known. I went back to a PS/2 mouse after I installed ps2rate so I wouldn't have to swap mice back and forth when booting between linux and windows. (wanna buy a used USB intellimouse?)

    Funny thing is that Quake2 plays and feels almost identical between the two OSes. Its Q3Test where I see a big difference. It has gotten better in windows with the 1.06 and 1.07 versions, but it still just looks better in Linux for some reason.

  • I didn't say do software emulation on a x86 cpu. I said "You can be on a fast machine and do software emulation. ".

    That when I say fast machine, I'm impling SMP. Have a cpu or two just do rendering. =)

    Renicing q3 helps a hell of a lot, so does giving it a larger core space. I also have a 120Mhz i80586 @ 120Mhz and 66Mhz bus and 48MB EDO RAM with a 4MB 3dfx card. I bet I can blow the doors off a win32 machine anywhere near those specs.
  • A shot in the dark, you have glibc2.1 with RH6.0 and not glibc2.0.

    I don't know if it changes things. Try downloading glibc2.0, put it in a separate place like /usr/local/quakelibs/ and issue something like
    LD_LIBRARY_PATH=/usr/local/... quake3 or whatver the command is.
  • Also, (from what I've heard) Q2 was just about forced to release early when most of the game was leaked and Id din't want to deal with compeeting against itself. At least that's the story I heard. I forsee this as being a highly organized and deliberate testing process, I applaud the guys at Id. Always have.

    matguy
    Net. Admin.
  • no one erased the post, it's rating was lowered. If you turn your restrictions down you should see it.

    matguy
    Net. Admin.
  • well, i've the same configuration and i've to laun ch q3 test as root. If i launch it as a user i got sig11 to.

Slashdot Top Deals

I've noticed several design suggestions in your code.

Close