#pentagram@irc.freenode.net logs for 21 Sep 2003 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage

[04:36:41] --> Kirben has joined #pentagram
[04:36:42] --- ChanServ gives channel operator status to Kirben
[06:07:02] <-- Kirben has left IRC ("System Meltdown")
[09:35:53] --> Dark-Star has joined #pentagram
[11:05:07] --> wjp has joined #pentagram
[11:05:07] --- ChanServ gives channel operator status to wjp
[11:51:45] <-- Dark-Star has left IRC (orwell.freenode.net irc.freenode.net)
[11:52:42] --> Dark-Star has joined #pentagram
[12:55:09] --> Colourless has joined #Pentagram
[12:55:12] --- ChanServ gives channel operator status to Colourless
[12:55:31] <DarkeZzz> Greetings.
[12:55:34] <Colourless> hi
[12:56:37] <DarkeZzz> Looks like Daikatana was slapped silly again in the latest GameSpy "25 Most Overrated Games Of All Time" article. *grin*
[12:58:39] <DarkeZzz> http://www.gamespy.com/articles/september03/25overrated/ <- Here for those who haven't seen it. *grin*
[12:58:48] <Colourless> i would seriously not place it in most overrated games of all time
[12:59:04] <Colourless> because by the time it came out, everyone had written it off
[12:59:17] <DarkeZzz> Heh.
[12:59:39] <wjp> hi Ryan
[13:00:59] <DarkeZzz> I'd put it at #1 myself, rather then #3 that gamespy put it. Simply for me because it defines hype-overkill to the point of "It can *never* be this good!" unbelievability.
[13:08:02] <wjp> I never even heard of most of the games on there
[13:08:24] * DarkeZzz thinks you need to play more games then. *grin*
[13:08:55] <wjp> more 'new' games, you mean? :-)
[13:09:40] <DarkeZzz> I wouldn't call Battletoads, Mortal Kombat, Battle Arena Toshiden, or Donkey Kong Country 'new'. *grin*
[13:09:55] <wjp> I know Mortal Kombat of those, but that's it :-)
[13:10:36] <DarkeZzz> Blinx is the only one I hadn't heard of, but I'd seen it and it (didn't remember the name) and it didn't look too bad.
[13:11:25] <DarkeZzz> LC: Tomb Raider: $(whatever the current rehash is called) I didn't really have to know much about to know it was overhyped and it stunk. *grin*
[13:12:17] <wjp> I 'heard about' B&W, Enter the Matrix, Daikatana, FF8, Unreal 2, Dungeon Siege, BC3000AD, Morrowind
[13:12:21] <DarkeZzz> BC 3000AD has Derek Smart as it's programmer and lead... err... 'evangelist'. You simply assume it's overhyped. *grin*
[13:12:33] <wjp> and 'know about' Quake 2, Mortal Komat, NWN
[13:12:39] <DarkeZzz> That's FF7, BTW, not 8.
[13:12:44] <wjp> um, yes
[13:12:52] <wjp> too much U8 :-)
[13:13:25] <DarkeZzz> 8's the one that everyone thinks is inferior to 7, but I find was far nicer to play in terms of both storyline and gameplay. *grin*
[13:13:54] <DarkeZzz> Morrowind I never found overhyped, but then again I never played Daggerfall or it's predicesors so I never paid much attention.
[13:15:27] <Colourless> quakr2... was a far better game than quake 1 ever was. if anygame should be in the 25 most overrated games it should be quake 1.
[13:16:18] <Colourless> quake 1 was a design mess. it's single player game was an embarrassement.
[13:16:18] <DarkeZzz> Halo seems to also define overhyped. Being originally an X-box release exclusive and the fact the gameplay is a bit dull and repetitive.
[13:17:48] <DarkeZzz> I dunno. I never really liked the single player game of Q1, but quite like'd Q2's, and I didn't mind either's multiplayer gameplay, though I always found Q1's faster. Q2 certainly seemed to get lots more hype before release then the original though.
[13:18:53] <Colourless> quake 2 suffers from revisionist history. it was a damn good game when it came out. Quake 1 fanboys hated it because it really was a different game. And then Half-life came out later in the year, so quake 2 always gets compared to hl, even though IMO q2 should be compared to the games that can before it, not almost a year after it.
[13:19:01] <DarkeZzz> The Starfleet Command series of games certainly falls into the same catagory as that Tomb Raider one. Bad design, buggy as anything, derivitive and unimaginative, and making millions simply because of drooling fanboy's.
[13:19:07] <Colourless> Quake 1 came out pre 'internet' pretty much
[13:19:28] <Colourless> it was hyped pretty heavily by magazines
[13:19:30] <DarkeZzz> Yup. So most of it's hype would have been in the compter mags, which I never read.
[13:21:00] <Colourless> IMO q1 really should have been slammed by the magazines. it came out after duke3d, and other than being true 3d really didn't have anything compelling about ut.
[13:21:09] <Colourless> s/ut/it/
[13:21:15] <DarkeZzz> I'm not really sure why 'Alice' is in that list too, there was lots of hype about how really really pretty and well designed it was, and it lived up to that. I don't remember any hype about how well it played really, and it's gameplay wasn't brilliant, but certainly compariable to FAKK2's or any other semi-adventure Q3-engine game.
[13:22:30] <DarkeZzz> I'm one of the weirdo's who thought Duke3d would have been better off staying 2d. *grin* Didn't really much like it's multiplayer gameplay at all.
[13:24:42] <DarkeZzz> Still, again, the problem is that's all pre-internet hype.
[18:01:47] --> Fingolfin has joined #pentagram
[18:01:47] --- ChanServ gives channel operator status to Fingolfin
[19:55:15] <wjp> hm, an SDL 1.3 branch has just been created
[19:56:05] <wjp> with offscreen opengl render targets, it seems
[19:56:35] <Colourless> bah, aka pbuffers
[19:56:43] <Colourless> arr/
[19:56:48] <Colourless> s/bah, //
[19:56:53] <Fingolfin> interesting. but there has been a CVS module for SDL 2.0 for quite some time, too, so I wouldn't overrate it =)
[19:57:07] <Fingolfin> of coruse SDL 2.0 is the one with a completely newly designed API
[19:57:08] <wjp> yeah, that contains a readme or something, doesn'it it? :-)
[19:57:19] <Fingolfin> it has some code actually
[19:57:22] <wjp> ok, I'm sure I should be able to spell "doesn't" correctly
[19:57:44] <wjp> code already? wow :-)
[19:57:51] <Fingolfin> http://www.libsdl.org/cgi/cvsweb.cgi/SDL20/
[19:58:06] <Fingolfin> (code != useful code, though =)
[19:58:43] <Colourless> thing about pbuffers is they will require explicit opengl driver extension support. If the driver doesn't support pbuffers, things will probably die pretty badly if you attempt to use the offscreen surfaces
[19:59:14] <Colourless> that said... useful for what i want to do with pentagram :-)
[19:59:35] <wjp> it sounds pretty hard (if not impossible) to emulate without driver support
[20:00:03] <Colourless> not entirely impossible
[20:00:23] <Colourless> there are ways around it, but with severe limitations, and potential speed problems
[20:00:37] <Colourless> you can always just render to the back buffer and copy to a texture
[20:00:51] <Colourless> the copy though might be 'really' slow
[20:01:28] <wjp> clueless question: so rendering and flipping are separate?
[20:01:41] <Colourless> yes
[20:02:16] <wjp> any idea which calls triggers rendering in SDL's GL mode?
[20:02:21] <wjp> s/calls/call/
[20:02:40] <Colourless> when rending in gl, you use gl calls :-)
[20:02:56] <Colourless> normal 2d rendering and GL do not mix :-)
[20:03:05] <wjp> oh, the various calls cause the object to be rendered immediately?
[20:03:23] <Colourless> to the programmer that is how it appears
[20:03:25] <wjp> (like I said: clueless :-) )
[20:03:37] <Colourless> often there is some latency involed, but you never normally see it
[20:03:48] <wjp> they're async?
[20:04:02] <wjp> (the rendering calls)
[20:04:09] <Colourless> each call doesn't wait for the hardware to complete the operration before returning
[20:04:43] <Colourless> in fact, if the hardware hasn't finished the previous operation, the new operation will be buffered untill it can be executred
[20:05:14] <Colourless> so, when you call to flip the surfaces, you might actually have to wait till the hardware has actually finished rendering before the flip actually occurs
[20:05:35] <wjp> flip is probably synchronous?
[20:06:27] <Colourless> you'd think so... but it's not always (can get higher performance if it's not)
[20:06:52] <wjp> hm, weird
[20:07:19] <Colourless> if you have vsync enabled, then it is most likely going to be syncronous
[20:08:37] <Colourless> the way if all works is pretty simple. the hardware has a largish fifo buffer (no idea of example sizes, and it will differ on all hardware)
[20:09:15] <Colourless> the biffer itself can be big enough to store data from multiple rendering frames, so it could have multiple buffer flip commands in the fifo
[20:09:27] <Colourless> s/biffer/buffer/
[20:10:42] <Colourless> it's as simple as creating latency can improve performance, since you can more ensure that the hardware is always working
[20:10:55] * wjp nods
[20:11:10] <wjp> I see
[20:11:54] <Colourless> there are always certain operations though that require the fifo be flushed
[20:12:14] <wjp> things like projection changes?
[20:12:17] <Colourless> there would pretty be any operation that attempts to lock the cards memory
[20:12:26] <wjp> or is that queued too?
[20:12:39] <Colourless> projection changes are just changing the matrices
[20:13:19] <Colourless> locking the memory would be the app reading or writing to the framebuffer (reading is really expensive), and uploading new textures
[20:14:34] <Colourless> and modifying vertex buffers/arrays. However, vertex buffers/arrays have a number of optimizations, that the programmers needs to be aware of so they can minimise the performance impact
[20:16:07] <wjp> I assume the vertex buffer contains, uh, buffered vertices? :-)
[20:16:28] <wjp> do all operations that need coordinates use the vertex buffer or can they use raw coords too?
[20:16:29] <Fingolfin> wjp: what a wild theory!
[20:16:48] <Colourless> you can use raw coords
[20:17:30] <Colourless> but, doing that isn't really friendly to your performance. in opengl you have at least 2 or 3 function calls per vert, which as you can imagine get pretty expensive, pretty fast
[20:17:51] <wjp> I expected as much :-)
[20:18:01] <wjp> (otherwise the vertex buffer wouldn't be particularly useful)
[20:20:43] <Colourless> raw or 'immediate' mode as it's often known came first
[20:22:40] <Colourless> this was back in the days of IrixGL and OpenGL 1.0 which were often used to send commands to renderes over networks
[20:23:53] <wjp> so I imagine reducing the amount of data to be sent would really improve performance
[20:29:38] <wjp> hm, I think I'll go play BG for a bit and then go to bed
[20:29:43] <Colourless> cya
[20:29:45] <wjp> thanks for the explanation
[20:29:49] <wjp> night :-)
[20:29:52] <-- wjp has left IRC ("Zzzz...")
[20:29:53] <Colourless> i'll be going too
[20:29:55] <-- Colourless has left IRC ("casts invisibility")
[20:40:29] --> Dark-Star|afk has joined #pentagram
[20:49:21] <-- Dark-Star has left IRC (Connection timed out)
[21:31:26] <-- Fingolfin has left IRC ("42")
[21:49:16] --> mm1 has joined #pentagram
[21:50:25] <mm1> Hello, anyone around?
[23:19:33] --> wjp has joined #pentagram
[23:19:33] <-- mm1 has left IRC (Read error: 104 (Connection reset by peer))
[23:19:33] --- ChanServ gives channel operator status to wjp
[23:19:49] <-- wjp has left IRC (Client Quit)