|
| T O P I C | |
|
Walk Into Wall. Bleed. Repeat.
September 4th 2000, 04:09 MSD by andy NVIDIA owners -- rejoice! Epic have learned from their past mistakes. Owners of other video cards -- sigh in disbelief. Epic are going to make the same mistake again... Epic's Unreal engine was originally designed with 3DFX cards in mind, with its hardware rendering being based around the Glide API. This was because 3DFX cards were expected to be the de facto standard for gaming. It didn't work out that way, though, and as many other 3D cards became popular, so Epic had to adjust their engine to work with Direct3D and OpenGL. Although things got better with Unreal Tournament, and have continued to improve with patching, people still complain of poor (by comparison) frame rates under Direct3D, especially on NVIDIA cards. So do you think Epic have learned by now that it's a bad idea to target specific cards? (Cue the 'Crap masses: No!) Believe it or not -- but you really should believe it because it's true -- Epic have decided to focus their next engine on a specific manufacturer's cards. Again. This time it's NVIDIA cards, as announced in this press release:
Amazing, isn't it? Epic put themselves through a world of pain by focussing on one manufacturer when they believed it was going to dominate the market, and now four years later they're doing the same thing for the same reason. Gee, I wonder if it will all go horribly wrong... |
| C O M M E N T S |
|
Home »
Topic: Walk Into Wall. Bleed. Repeat.
|«« - Previous Page - Next Page - »»| |
|
<b>Andy</b> (#9): <QUOTE>BTW, we've got another Epic topic coming up soon. So if the Epic boys turn up here and start flaming me, and then another Epic topic pops up tomorrow, that's just the way it goes, it's not revenge. Not that the next topic is in any way critical of Epic, but just so you know. </QUOTE> Heh, cool. :) <b>Matthias Worch</b> (#13): <QUOTE>This deal will help getting the D3D performance of Unreal as good as it gets, and ALL D3D cards will profit from it (as long as they got a T&L unit, that is). Making Unreal better for NVidia cards means that it will automatically become better for all other D3D cards. Everybody should be very happy about this, but I'm sure some people will find enough hair-splitting techniques to make this a 300+ post thread ;) </QUOTE> Huzzah!! -- Warren Marshall - Professional Nuisance<I><B></B></I><I></I><I></I> |
|
Methinks andy now has a new whipping doll.....Epic. ---------------------- I used to be conceited, but now I am perfect.<I><B></B></I><I></I><I></I> |
|
Chris: Nah, the Radeon already has a T&L unit, and the next Voodoo cards are supposed to get one, as well. Of course the capabilities of each chip might differ, but that should just result in increased/decreased performance. So for cards with half-assed T&L implementations you might have to turn down a few details. Pretty much like having a slow CPU. <I><B></B></I><I></I><I></I> |
|
I'm just wondering if this means that nVidia will actually implement funtional palletized texture support for the geforce line. If they do, and Unreal engine performance gets reliablly good because of it, I just may ditch my V5 5500. ---------------------- I used to be conceited, but now I am perfect.<I><B></B></I><I></I><I></I> |
|
Hmmmff. This is a moot topic. Let's do some time travel, 'kay? Way back when, when Unreal was first being written, there were no 3d accelerator cards. Then, there WAS a 3d card, or rather a chipset. 3dfx. Have you forgoten? Nothing could touch the performance of a VooDoo back then. So Unreal was updated/recoded to take advantage of what was at the time, the supreme 3d card. The way you did it back then was to write for Glide, since DirectX/Direct3D was very flaky. At the time, coding your entire engine for Glide seemed to be a good idea, since there wasn't any other vialble API around. So we arrive back at the present day. Epic has an engine that is still very Glide-centric. And 3dfx is borderline on the ash heap of computer history. This needs to be fixed. Now, nVidia is the current 3d card of choice, but you know what? It HAS NO API. No hardcore to-the-metal API like Glide is. So... As Apache stated in <B>#10</B>, this whole thing is moot. It doesn't matter if Epic is coding their nextgen engine to do D3D, or OGL. Both are set standards. D3D works on any Windows box, OG works on any box with supported drivers from hardware vendors. Bah. This has taken me 20 minutes to type. Having a broken collar bone is a pain in the ass. Mark/Mad_Dog<I><B></B></I><I></I><I></I> |
|
From reading what Epic have been saying over recent months, I'd be inclined to say that they will be supporting Nvidia cards by default more than in any specific manner - and that this agreement is little more than mutual advertising bunk. My reasoning for thinking this is that Epic have said a lot about focussing all future engine development on Direct3D, and Tim Sweeney has repeatedly commented upon using DirectX 8 and 9's features. Since DirectGraphics (the Direct3D/DirectDraw merger in DirectX 8) is pretty much a hardware implementation of the GeForce 2 (and later NVx hardware), its no surprise that their future projects work best on Nvidia's chips. ATI and other manufacturers have some wonderful hardware - but its not quite as close to the DX8 specification, and therefore probably won't work as well. |
|
I would not read into this very much. It is not a dig at 3dfx or ATI. It is really just a PR statement that nVidia will pay full attention to Epic when Tim Sweeney calls them up saying he wants to test UT2/Next Big Game with nVidia's line of cards. I know that 3dfx, ATI, and Matrox have similar deals with developers, possibly even Epic. It is called developer relations, plain and simple. I've dealt with the ATI and Matrox DR in a small way (not for 3D, but for their Video in a Window cards) Aside from some marketing noise, I don't understand why this is a big deal. Greg |
|
I'm not too sure we can draw any conclusions from this. I tend to agree with Apache, who said it's "PR CRAPOLA" It is probably prudent to wait it out, and see how things develop. Until we see the code in action, it's a bit counter productive. Until then, the Bills are up 16 - 13 with not much time left.. can the titans do it again? Paul Shrinkweb.com |
|
Gnargh. As long as Epic don't use a new proprietary API by nVidia, I don't see much of an issue here. - Morn <I><B></B></I><I></I><I></I> |
|
Hey! Great news! Especially since my video cards don't have any problem running D3D. Unlike some... other... cards. It's gonna take more than some advertising gimmick to get me to drop my 2xVoodoo^2s :)<I><B></B></I><I></I><I></I> |
|
While all video cards nowadays support D3D and OpenGL, there is a world of difference in the way you would optimize for certain hardware. In the case of the GeForce series and the V5, code which will speed up performance on one card will have no effect on the other, or perhaps even slow it down. The fact is that right now you would have to write two different rendering cores to get the best speed out of both chipsets. So unless we see a similar announcement from 3Dfx & Epic, we can pretty much count on them releasing a basic engine that will work on any hardware, and a very optimized one for nVidia's chips. And given that MS said Sweeney would be involved in DX8 development, I think it's pretty clear what Epic will be spending their time on: X-Box. |
|
<b>#17</b> "Chris Johnson" wrote... <QUOTE>Are the T&L unit, etc, a specifically nVidia thing? Even if increased D3D support does help other cards/chipsets as well, wouldn't the code for thsi sort of thing still be more of an advantage to nVidia? </QUOTE> Yes and No. Nvidia was the first out the door with a hardware T&L engine for consumer level cards, since it was the first and only one avalible at the time Microsoft basicly had to use it as the base for DirectX hardT&L. There are some changes here and there, but at it's core it's Nvidia's setup. On a side note ATI would have more reason to get excited about a pure D3D T&L out of UT. The Rendon is builted around some of DirectX 8 (T&L engine mostly). -- None-1a. O forget it.<I><B></B></I><I></I><I></I> |
|
<i>Thinking...</i> |
|
<i>Thinking...</i> |
|
Epic's not doing one thing wrong here Andy. D3D is really the industry standard. Glide is dying and really dead now. Most people will not support OpenGL because it has driver issues across cards, and requires the card people to commit to an OGL driver. That leaves D3D. May not like it. But MicroSoft isn't going anywhere and they own the driver market now. Period. D3D is the future of PC games. Get used to it. All Epic is doing (I'm sure) is making sure their news stuff runs on nVidia cards with all the bells and whistles. If Epic isn't using D3D then that's another issue, but I suspect they are. George Broussard, 3D Realms |
|
This is the change I've been arguing should take place for a LONG time. This is NOTHING but good. The problem with targetting 3DFX is that they use a <b>proprietary</b> API. Designing for that API automatically limits your target audience and makes it complicated to "reverse engineer" the core rendering engine to work with other APIs...thus the problems nVidia owners are currently seeing with UT-based games in D3D. By targeting nVidia, they're not latching on to a proprietary API. They'll have full OpenGL and D3D support, and it's a relatively simple matter to design for those APIs and <b>still</b> include support for the hardware-specific features on systems that support it. In my opinion, four years ago, Epic made a logical choice (though arguably not the best choice). At the time, it looked like a good idea, as 3DFX was poised to take over the 3D-chipset world. 3DFX dropped the ball. Now Epic is in a position to switch their focus, and I believe they're making the right decision. 3DFX probably won't roll over and die, but they need to do some restructuring to ensure that their hardware and drivers fully support D3D and OpenGL at a competitive performance level. IMO, GLIDE has had it's day. We've moved beyond what can be accomplished with a proprietary subset of OpenGL features, today's games (and tomorrow's) will require much more than the 3DFX/GLIDE combination can provide. -Valeyard<I><B></B></I><I></I><I></I> |
|
<b>#Main Post</b> "andy" wrote... <QUOTE>Gee, I wonder if it will all go horribly wrong...</QUOTE> I don't see how. When you weigh the available hardware performance, nVidia comes out on top. When you weigh the APIs, D3D comes out on top. Unless another vendor comes out with a radically new hardware design <b>AND</b> it requires a proprietary API <b>AND</b> it can garner support from developers and consumers....the situation isn't going to change for a while. When you consider game development time of 18-36+ months, the decisions made today will affect the industry for at least that long. That doesn't imply that the industry will stagnate...the key is scalability. Even if someone came up with a 3D chipset that pushed twice as many poly/pixels as a geForce2 Ultra, it'd <b>have</b> to support the open-standard APIs <i>just</i> to establish a competitive edge. Developers simply aren't going to risk jumping ship to a new, untested product with a small user-base. Considering that, if this wonder-chip popped into existance tomorrow, and DOES support the open-standard APIs...we've lost nothing. With proper LOD support, the games can be made to scale up to the new hardware or remain playable on the average user's system. This is a win-win situation for Epic and Gamers. It may also give 3DFX the wake up call they need to leap back into competition. Open-standards are good, scalability is good. Mmm'kay? :) -Valeyard<I><B></B></I><I></I><I></I> |
|
Okay, either you've all misunderstood, or I've misunderstood, or God help me I've been suckered big-time by a press release. I took the press release to mean that Epic will be designing their games to work best on NVIDIA cards. I don't mean that Epic games would be designed for some new, as-yet-unknown NVIDIA API, but that features would be implemented that would look/work great on NVIDIA cards, but not necessarily work so well (if at all) on other cards. Stupid example, just to illustrate what I mean: NVIDIA cards can render photo-realistic skies using one D3D command. Epic uses these skies in their next engine, and people with other cards get some dodgy mid-90's DOOM sky. So yeah... me, suckered by a press release. I bow my head in shame. |
|
<QUOTE>So yeah... me, suckered by a press release. I bow my head in shame. </QUOTE> Well, its that or you could just assume Epic and/or Nvidia is full of shit and that what this really means is that D3D will work like it should, but with a few specifc and meaningless Nvidia graphical features. Ah damned if you do and damned if you don't Andy. -- Rambar |
|
Andy, I took the press release to mean that Epic will be designing their games to work best on NVIDIA cards. I don't mean that Epic games would be designed for some new, as-yet-unknown NVIDIA API, but that features would be implemented that would look/work great on NVIDIA cards, but not necessarily work so well (if at all) on other cards. What other chipsets are worth a crap besides nVidia and 3Dfx? Sorry, they all pretty much such outside of those two. The point is to do things in one API and let it use whatever features the cards can. Sure, nVidia cards will likely look the best for now, and if you have some cheap ass card it's going to be less acceptable. But how is that a bad thing? The days are gone of trying to hand tune a game to look good on several cards at once. The hardware people better get in line with features if they want games to look good on their cards. There will always be a couple of APIs out there, but anyone that thinks D3D isn't going to be the primary one is mistaken. George Broussard, 3D Realms |
|
I've said it before and I'll say it again: Quake 3 engine for OpenGL Lithtech engine for D3d Unreal engine for GLide So, just code the entire game in each of these engines, and you're set! <I><B></B></I><I></I><I></I> |
|
Hey Warren? I just thought I'd let you know that the Linux port of UT is fucking <b>awesome</b>. For those of you who didn't catch it the first time, I said <b>awesome</b>. S3TC support, an OGL renderer that runs faster than the Windows D3D, beautiful smooth stutter-free playing (hello Linux memory managment ;), faster load times, volumetric lighting that looks ass kicking, blah blah blah... you guys did the right thing hiring Loki. I could rant for a while on this one... When a friend of mine saw it, he said "Holy shit, it's like Quake3 now, only it includes fun."... UT's back on my harddrive, on it's own 2.5GB Linux partition (yeah, I installed it just to see if it was better). Thanks for the support Epic. :) As for you Andy, I think it was just a simple misinterpetation mistake on your part... Epic may very well support features only NVIDIA can do, <B>at the time</b>... but I doubt they'd support proprietary technology avalible and applicable only to NVIDIA. Another thing to consider is: These days if a game can run efficiently on an NVIDIA card using D3D or OGL, it should run well on other cards as well as long as their drivers are in good shape, right? NVIDIA so far has avoided proprietary technology and proprietary APIs, so if a game is optimized for a card with fairly standardized features, it should do well for most cards. |
|
<b>George Broussard</b> (#33) wrote: <quote>D3D is the future of PC games. Get used to it.</quote> I beg to differ. Unfortunatly, OpenGL won't die because it's multiplatform and not OS-specific. D3D is going to eat some market share, but there will still be a large section using OpenGL because it's what's supported under Linux, Unix, BeOS, and Mac. Honestly, the only major thing I see pulling for DX here is the fact that D3D will be integral to the design of the Xbox. With titles being easily portable between PC and Xbox if D3D is used, that's obviously going to sway a huge amount of developers over. But unless and until you see MS produce DX/D3D for other platforms (which is doubtful... but hey, they <b>are</b> porting Office to Linux :), you're not going to see D3D take over. |
|
Carmack is only interested in Doom 3 on the X-Box if the X-Box has OGL implementation. So it will have it. Andy's italics-source in the other thread (regarding 'Molehill Erectus') is spot-on, from now on all George Broussard has to do is say <b>Meep Meep</b> whenever he wants to reduce Andy's arguament to rubble. |
|
Even without proprietary API, a programmer can use hardware-specific functs of nVidia cards directly (surpassing D3D). Call it a driver hack, if you want. As far as I know, DX8 will support all this stuff anyway, so there probably will be no need to make any workarrounds, everything should work out of the box (...heard this so often before :-)). I believe nVidia just wants to ensure that the damn thing will work fine with the next incarnation of Unreal engine, not more than that. Also, I think Epic guys wouldn't be so stupid to do the same mistake again. My two cents... |
|
Andy, why don't you just ask Warren M. about it? |
|
<b>xero wrote in post #41:</b> <quote>D3D is going to eat some market share, but there will still be a large section using OpenGL because it's what's supported under Linux, Unix, BeOS, and Mac.</quote> "Large" section? Maybe this is one of those Linux things, but I really doubt that the market share is really being eaten up by that many Linux, Unix, BeOS, and Mac game users. Maybe it's just me. -- Dethstryk Damage Gaming |
|
One thing that i've been thinking about, especially since NVIDIA's involvement with the X-Box is how much control they will have over the content of DX. Given how much is riding on the X-Box i'd have thought that MS will have given NVIDIA as much knowledge as they need for designing and implementing a DX8 chip. Also MS may well be asking NVIDIA 'how do you want us to implement function Y?' I think that NVIDIA will have the best DX8 chip out there because of this and i'm just wondering about the knock-on affects of this. (mindless speculation admittedly :), but its something i keep thinking about) Dev |
|
In a recent thread, Hulka popped in swearing because Unreal Tourney was hosing his machine. It seems he had an Nvidea card. I have an Nvidea and seemingly had the same show stopping horrible bug. When the program ran, the screen was garbage. In my case, the game's INI files were pointing to the unused and inactive inboard video chip instead of the Creative TNT card. Before I read the other comments that have probably addressed this, the odds that this horrible show stopper only happened to two people are slim. My guess is both Nvidea and Epic have taken a lot of heat and are getting together to clear both their names. -Steve |
|
George Broussard wrote: <quote> What other chipsets are worth a crap besides nVidia and 3Dfx? </quote> Guess you haven't kept up with the ATI Radeon George, which compares favorably to the V5 and the GeForce chipset families. However much you may wish for one OS, one solution, one API, one company, it won't happen. Choice is a <b>good</b> thing. |
|
I will give Andy some credit on the topic title. I can't read it enough. :) -- Dethstryk Damage Gaming |
|
#49" When I first read the topic I seriously thought it was gonna be a thread about Daikatana :) <I><B></B></I><I></I><I></I> |
|
Xero, <quote>I beg to differ. Unfortunatly, OpenGL won't die because it's multiplatform and not OS-specific. D3D is going to eat some market share, but there will still be a large section using OpenGL because it's what's supported under Linux, Unix, BeOS, and Mac. </quote> Differ all you like ;) I never said OGL was going to die, but the reality is that a mere handful of developers will ever use it. There are too many reasons to use D3D and it's updated every year - while OGL is essentially frozen in time. <quote>But unless and until you see MS produce DX/D3D for other platforms (which is doubtful... but hey, they are porting Office to Linux :), you're not going to see D3D take over </quote> Whos' talking about (or cares about) other platforms? ;) We're talking PC games here. Crusader: <quote>Guess you haven't kept up with the ATI Radeon George, which compares favorably to the V5 and the GeForce chipset families. However much you may wish for one OS, one solution, one API, one company, it won't happen. Choice is a good thing</quote> The Radeon is fine. Different hardware is ok, although only the top 2 at any given time will be popular and normally one will dominate. I was mainly talking about API's. And Glide has been great in the past, but now 3Dfx's market share has dropped and will likley continue, forcing developers to abandon Glide and go to D3D or OGL. And the major majority will choose D3D because it's updated once a year and MicroSoft (love em or hate em) dominates. Realize that only a couple of companies like id have the clout to pretty much force card makers to create goot GL drivers. You think the hardware people want to do that? Hell no. They'd all likley choose to support D3D only if given the choice. Makes their lives easier. George Broussard, 3D Realms |
|
Anyone else see the irony of GB posting how DX is going to dominate with OGL a supported second, while Glide has died a death, <b>as he's working on a Glide-heavy game</b> after he dumped an OGL engine? Really, anyone else see that as funny? <i>Anyone?</i> <I><B></B></I><I></I><I></I> |
|
#52 Hmm, maybe the whole fatbabies rumour about the new engine had some minor basis in fact that 3DR was going to switch engines, to the Unreal 2 engine? Would that be an engine switch though, or an upgrade? Tough call. |
|
<b>xero</b> (#40): <QUOTE>When a friend of mine saw it, he said "Holy shit, it's like Quake3 now, only it includes fun."... </QUOTE> LOL! :) <b>xero</b> (#41): <QUOTE>I beg to differ. Unfortunatly, OpenGL won't die because it's multiplatform and not OS-specific. D3D is going to eat some market share, but there will still be a large section using OpenGL because it's what's supported under Linux, Unix, BeOS, and Mac</QUOTE> It depends what you define as "large". The only company that really pushes OpenGL anymore is id. Seriously. Take a look around at recent games ... any game that isn't a Quake engine license is generally in D3D. And even some Quake engine licenses add a D3D renderer on their own ... I used to be a huge OpenGL supporter. I still love the API -- I think it's clean and fast and it's VERY nice to program in. But I also accept that it's losing the war. D3D is the future. :) <i>NOTE : this does not necessarily reflect the opinion of Epic Games ... I'm just a cog in the machine. :)</i> <b>godZero</b> (#44): <QUOTE>Andy, why don't you just ask Warren M. about it? </QUOTE> I'm tempted to give an opinion on this that I believe is Epic's official position, but I'm not going to be the one writing the next renderer ... that'll be Tim. I CAN say that I would be <b>VERY</b> surprised if we didn't go with Direct3D. <b>Crusader</b> (#48): <QUOTE>However much you may wish for one OS, one solution, one API, one company, it won't happen. Choice is a <B>good</B> thing. </QUOTE> One API IS happening. You just haven't noticed. :) -- Warren Marshall - Professional Nuisance<I><B></B></I><I></I><I></I> |
|
Hmm Kyro Chipset (tile based rendering) Matrox Chipset, g400 runs UT in d3d just fine thanks S3 Chipset (on the way out, allegedly) 3dfx/gigapixel Nvidia Via have an onboard graphics solution ATI (radeon and god alone knows how many furys and rage 128) Trident, the Dimarge cards run UT in d3d Intel, the i740 runs UT ...and of course the onboard i8x chipsets... BitBoys Oy (well, maybe its not a hoax) see, there are _lots_ of video cards that run d3d .... Some of them even have a decent GL implementation, unlike 3dfx who still resort to that mini-gl wrapper Ds <I><B></B></I><I></I><I></I> |
|
<b>[@~]MizuGami</b> (#8): <QUOTE>I am glad that Epic has decided to go with the VASTLY superior Nvidia cards over the 3DFX cards. </QUOTE> Nvidia anal sex happening! <I><B></B></I><I></I><I></I> |
|
Can anyone explain to me why an engine can't work with Glide and D3D with good performance in one game?! I would love to hear reasons why developers (read: Programmers) why they can't seem to work with both Glide and other API <I><B></B></I><I></I><I></I> |
|
<b>#34</b> "Valeyard" wrote... <QUOTE>By targeting nVidia, they're not latching on to a proprietary API. They'll have full OpenGL and D3D support, and it's a relatively simple matter to design for those APIs and <B>still</B> include support for the hardware-specific features on systems that support it. </QUOTE> you are kidding right ? Nvidia has unbelievably kewl stuff in their OGL drivers that are not in other cards - these features if chosen significanlt enhance the engine. So if they end up using these features they end up using a proprietry version of OGL specifically targetted for a set of cards. And as Nvidia own the IP to some of featuers we will only see them on other cards if other manufacturers license them or Nvidias death grip is released. Do you see either happening ? <I><B></B></I><I></I><I></I> |
|
<b>#35</b> "Valeyard" wrote... <QUOTE>Considering that, if this wonder-chip popped into existance tomorrow, and DOES support the open-standard APIs...we've lost nothing. With proper LOD support, the games can be made to scale up to the new hardware or remain playable on the average user's system. </QUOTE> err that is a common misconception. It really depends on where bottleneck is - in some cases it is faster to throw full set of tris at card, in other cases LOD will work and in still other cases it will have 0 effect. Put basically it depends on power ratio between card/cpu and the way the game uses geometry/textures etc <I><B></B></I><I></I><I></I> |
|
<b>#41</b> "xero" wrote... <QUOTE><B>George Broussard</B> (<A href="spy-internal:Load/148#33">#33</A>) wrote: <quote>D3D is the future of PC games. Get used to it.</quote> I beg to differ. Unfortunatly, OpenGL won't die because it's multiplatform and not OS-specific. D3D is going to eat some market share, but there will still be a large section using OpenGL because it's what's supported under Linux, Unix, BeOS, and Mac. </QUOTE> <QUOTE>Honestly, the only major thing I see pulling for DX here is the fact that D3D will be integral to the design of the Xbox. With titles being easily portable between PC and Xbox if D3D is used, that's obviously going to sway a huge amount of developers over. </QUOTE> Again it will not be a simple port from PC to XBox - there are so many differences in content/memory/disk requirements that any port will be a effort. Sure having a consistent API may make it a little bit easier - but still it is an effort (and OGL will be on XBox too). The reason I see DX taking off is because it is still going forward, where in OGL standardisation goes through the ARB which is **************SLOW**************. This is a shame because I am a OGL fan :(<I><B></B></I><I></I><I></I> |
|
<b>#57</b> "BloodKnight" wrote... <QUOTE>Can anyone explain to me why an engine can't work with Glide and D3D with good performance in one game?! I would love to hear reasons why developers (read: Programmers) why they can't seem to work with both Glide and other API </QUOTE> different philosophies. Glide is all raw, to the metal style programming where you can do virtually anything with the card. It makes a lot of assumptions on style of hardware and thus can not take advantage in other styles. In this case programmers are god. DX is a mixed environment. You can code to the metal with a little hackery and also program via high level apis. This is relatively delicate architecture and known to brake. However I believe it is much much much more stable now and when DX8 comes out I believe I will be switching to it :( OGL is high level and in this environment Driver writers are god. So Glide is aimed at bit pushers/garage hacks that go to the metal. OGL is aimed at people who actually understand differnt design choices (it is design heavy). DX allows almost any style of programmer. One of the problems with OGL is that it is easy to program with and a lot of less skilled programmers can use it - and they generally use it wrong way - and get bad performance - and then blame API. DX has a much larger entry requirement and thus people who get that far usually understand a little bit about underlying structures. <I><B></B></I><I></I><I></I> |
|
Its marketing bullshit. I don't see wtf Epic could get from nVidia, that me, or any other developer, can't get via the nVidia developer program. If they get something that I don't, I'm going to make a big stink about it, make a couple of calls and either get someone fired or cause so much ruckus, you'll know I'm involved. --- "Game developers are just human beings who happen to make games for a living. If you want to hold us up to higher standards of conduct, then go ahead ... but don't be surprised if we don't uphold them. - Warren Marshall" <I><B></B></I><I></I><I></I> |
|
<b>#2</b> "None-1a" wrote... <QUOTE>Some how I see this as nothing more then supporting Nvidia's hardware t&l and the perpixel shaders, not really a try to get it running the best on Nvidia hardware thing. -- None-1a. O forget it.</QUOTE> Exactly. And knowing Epic, if they can't get their D3D ducks in a row, there's a good chance they'll fuck this up too. T&L is no brainer, any dumb-ass can support it. Per pixel shaders is just one of those fancy thingies nobody is going to use, anytime soon. Here is an interview I recently did, regarding all these new buzzwords. <quote> <b>Would you care to give me your thoughts on the following terms, as well as whether you would support them? Do any of them excite you? Make you wish you had time to implement them? Do you think they are useless marketing ploys, or viable tools? In other words, would you use them in BC and, if you did, how would you? Time is, unfortunately, dwindling, so a prompt reply would make my (and my editors) day. ;)</b> <i>sure thing :-)</i> <b>ATI's Charisma engine/Pixel Tapestry engine</b> <i>This engine on this card, well, lets just say, its about time ATI entered the high-end fray. Because the development of my upcoming game is in the advanced stages with 3-4 months more of development time, I doubt that Battlecruiser Millennium will take advantage of most of these features. T&L support was already planned for the GeForce boards sometime ago, so, we have no issues there. Though the Charisma engine provides some new advancements in multitexturing and blending, they are features that a rendering engine has to take advantage of internally. Therefore, I don't expect very many early adopters of this board's features. I mean, this card just breathes advancements. For one thing, we've had bump mapping support in the BCM engines in lieu of the Matrox G400 boards, for some time now. But due to the huge amount of texture content required for our model database and the limited resources available to us, this feature won't enabled in the shipping version of BCM. However, if the demand is there once BCM ships, I'll have the special bump map texture content done and thus we'll be able to automatically support bump mapping for both the G400 and the Radeon board, via a patch release. And as for key frame interpolation is concerned, I doubt that most of us will be supporting it anytime soon (remember that whole S3TC farce?) due to the fact that its not compatible with any existing content creation methods. To support this requires a tremendous amount of content effort. Vertex skinning is also a sexy feature which I expect to catch on, but again, like key frame interpolation, requires quite a bit of effort. Having a third texture unit in Pixel Tapestry is also very useful in terms of fillrate savings if used in conjunction with several texture passes. When you have an engine that does four to five texture passes, having a third texture unit can very well cut that down to four or even three passes, depending on implementation. The third texture unit certainly allows more scene detail while saving bandwidth. Other features which will most likely go unexplored but which will probably make their way into some products are support for Npatches, depth of field and motion blur, all of which the Radeon supports. I can't say enough good things about this board and to delve into each and every high-end feature, would require an entire article. Being a member of the ATI dev support program, gives us access to all these cool toys, including these Radeon bad boys. Rest assured, as time and resources permit, we'll be making full use of as many features as possible, in future products. It would be such a waste of this card's potential, not to do that. </i> <b>Full-scene anti-aliasing</b> <i>Every power gamer wants a rig with the fastest graphics board in it. But, it becomes a pointless venture if that board comes with a feature that you can only make use of, at low resolutions or one which you have to wait for a developer to support. The Voodoo5 has excellent FSAA support with varying results. This is one of those features that ships with a board, ready to work, right out of the box. But, like everything else, has its limitations. Anyway, ready to work out of the box is a good thing. And wouldn't it be nice if all high-end features in the next-gen boards were like that? :-)</i> <b>Geometry Acceleration</b> <i>Well, what can I say, its the Holy Grail of rendering and the more geometric calculations offloaded to the hardware, the better. One thing to remember though is that most and all of these new hardware initiatives, rely on Microsoft to come through with their DirectX 8 promises in most cases. And since they have a unique penchant for spec changing in midstream, it would interesting to see just how much catching up the hardware vendors have to do, this time around. Being on the DirectX 8 dev program and having had access to it since it went Beta, I can safely say that its very promising and most likely to be the most important DirectX initiative from Microsoft.</i> <b>3DFX's T-Buffer</b> <i>This one is not going to be widely popular because it is only useful on a per application basis.Right now, I'm thinking of this in pretty much the same way I regarded S3TC. It either catches on or it doesn't. And on the off chance that it does, it won't be anytime soon. Its the oldest graphics trick in the books. Lets put a bunch of less than useful features on a board, make it sound cool and hope someone actually finds it cool enough to actually implement it. I wouldn't bet on that anytime soon.</i> <b>Thanks again for your help in this Derek.</b> <i>No problem. And sorry for the late reply, its been one helluva week, especially with that noise last week about my licensing the Serious Engine for a new Xbox project.</i> </quote> <I><B></B></I><I></I><I></I> |
|
<b>#33</b> "George Broussard" wrote... <QUOTE>All Epic is doing (I'm sure) is making sure their news stuff runs on nVidia cards with all the bells and whistles. If Epic isn't using D3D then that's another issue, but I suspect they are. </QUOTE> George, Question: what makes you think that Epic would be able to do <i>anything</i> innovative, without going thru D3D. Answer: They can't. (<i>but there is one dumb schmuck out there, who might think thats a good thing. After all, we did it for Glide</i>) Therefore, all I read into this marketing hype, is that Epic are going to, finally, do something about the crappy D3D performance in their engine. In much the same way they hunkered down and actually fixed the net code in UnrealT. This announcement with nVidia, being the perceived leader in the chips market, (<i>which, they're not, last time I checked</i>) is just a marketing gimmick that serves to pave the way for their nextgen engine, in order to alleviate the stigma associated with the crappy D3D support in the Unreal engine. Period. Two guesses what nVidia gets out of it. --- "Game developers are just human beings who happen to make games for a living. If you want to hold us up to higher standards of conduct, then go ahead ... but don't be surprised if we don't uphold them. - Warren Marshall" <I><B></B></I><I></I><I></I> |
|
Post #62 is why I like Derek so much. You go, boy. Your posts have already cheered up my broken little heart a little. :) -- Dethstryk Damage Gaming |
|
<b>#63</b> "Derek Smart" wrote... <QUOTE>Here is an interview I recently did, regarding all these new buzzwords. </QUOTE> I can see the 'Holy Grail' of any hardware feature turning into getting it put into directX, of course that'd require more open hardware standereds (of course we've seen in the nobody knows thread how fast ideas to get that have their detractors). Take S3TC for example, I see it going down in flames and every one begining to use DXTC instead on most games. <b>#64</b> "Derek Smart" wrote... <QUOTE>This announcement with nVidia, being the perceived leader in the chips market, (<I>which, they're not, last time I checked</I>) </QUOTE> Last I knew ATI was still on top for total market share (or course that's because of the shear number of their low end cheap 'o chips). -- None-1a. O forget it.<I><B></B></I><I></I><I></I> |
|
<b>#58</b> "RahvinTaka" wrote... <QUOTE>Nvidia has unbelievably kewl stuff in their OGL drivers that are not in other cards - these features if chosen significanlt enhance the engine. So if they end up using these features they end up using a proprietry version of OGL specifically targetted for a set of cards. And as Nvidia own the IP to some of featuers we will only see them on other cards if other manufacturers license them or Nvidias death grip is released. Do you see either happening ?</QUOTE> The point was that you can write your renderer to support OGL or D3D and <b>still</b> include support for hardware-specific enhancements. Other engines can render OGL or D3D and selectively support the hardware T&L on systems with the appropriate hardware. You can do this without significantly affecting performance...it's just that some people will get more bells and whistles. The larger issue was the fact that building a new rendering engine using an open-standard API instead of the proprietary GLIDE, is the best decision. As Derek pointed out, this is most-likely raw marketting. It's a peace offering to nVidia owners, letting them know that Epic has "seen the light" and is now actively working to ensure that their products function properly on the card with the largest installed user base. -Valeyard<I><B></B></I><I></I><I></I> |
|
<B>#66</B> "None-1a" wrote... <QUOTE> Take S3TC for example, I see it going down in flames and every one begining to use DXTC instead on most games. </QUOTE> *ahem* DXTC <I>is</I> S3TC, just with a different name. Perhaps you're thinking of 3dfx's FXTC? ~~ Ignie Ferroque, Ash. - "Non Facete Nobis Calcitrare Vestrvm Perinævm" |
| C O M M E N T S |
|
Home »
Topic: Walk Into Wall. Bleed. Repeat.
|«« - Previous Page - Next Page - »»| |
| P O S T A C O M M E N T |
|
|
| C R A P T A G S | ||||||||
|
|
| There are currently 66 people browsing this site. [Details] |
Powered by blah 0.9-dev •
PlanetCrap is © 1997-2004 Hendrik "Morn" Mans
|