Pixel shaders... Nv specific titles on the way?

Just wanted to show these...this is how it looks on my Radeon 9700 PRO
with 4XFSAA 16X ANISO 1280x1024



number2:




Cool huh ? pixel shaders rules

Tompa

 

And how is what I said totally wrong?

1. They used nVidia extensions for PS water effects - agreed
2. Bioware said they tried to implement ATI PS effects in OGL but the driver didnt work well and the effect was buggy initially - I dont know if it is true or not (which was why I said MAYBE) but both ATI and bioware worked togther to get it to work for the latest beta patch and Humus here had said for ages the ATI Fragment shader extension wasnt working properly for ages in driver revisions.
3. The AA slider in NWN didnt work on my 8500 IIRC and yes the quincunx option was a dead giveaway this was for nVidia. We agree.

Blimey if I was TOTALLY wrong there, I'd hate to be wrong elsewhere

 

Well, it's "totally wrong" in the sense of implying there's some sort of hardware feature the nVidia card has which the 9700 Pro doesn't which is going to require specific versions of programs for specific chips, and then using NWN as an example to prove it.

Both of the cards can do "shiny water" without a problem, so in the case of NWN it had nothing to do with the hardware being different in terms of capability, and had everything to do with the developers writing an OpenGL program which made use of extensions the developer knew beforehand were not useful with other cards. Basically, the developers of NWN did exactly what 3dfx used to get so roundly criticized for with Glide--and that was they deliberately wrote their code to support a single hardware type--when they didn't have to. I mean, the idea of putting "Quincuxx" actually inside the game, as a hard-coded choice for FSAA, was just...well, I don't know what they were thinking... Were they thinking that people who played 3D games used nothing but nVidia 3D cards? I suppose...but anyway--that's a developer flaw--and you have to admit a pretty rare one these days. It used to be fairly common back in Glide days--but this tendency has pretty much disappeared these days (which is nice.) I understand the reasons the developers did this--and although I'm not really upset about it--it is, however, a mistake and hopefully one these guys will learn from. Doing it right the first time would have generated a lot less work for them after the fact.

Well, from what I read that BW did--they essentially handed ATI nVidia's extensions and said "Here--use these--that's all we can do." Of course, there's little ATI can do there since that's not an honest attempt to correct the developer's problems with the game (which, again, is slight.) Apparently, though, at least they say--BW is now making an honest effort to fix the game. Again, all of this could have been avoided if they'd done it right the first time. I think they are very talented people but this attempt at 3D really shows me how in a sense they are novices at supporting it--so far. Still, NWN is a pretty cool game, I think.

I suppose, too, I may be a bit sensitive to this because I heard so many "bugs in the ATI drivers for shiny water" remarks early on--and I had a 9700P which was terrific in giving me "shiny water" in a number of other games, and because of the hard-coded "Quincuxx" FSAA option in the drivers inside the game itself, it was easy to see what had happened.

Well, I figured you might have had an 8500 or another set of drivers or something that caused the FSAA controls not to wrk for you, but I just wanted you to know they actually did work for me with the 9700 Pro! Hai! But I didn't actually play the game until after the first couple of patches were out--which I installed previous to playing--so they may have fixed the FSAA thing relative to ATI in one of those patches--and you might have started playing before the patches were released. Anyway--no biggie...

 

Guys like you are insignificant. Guys like you or any other programmer who isn't making a game or software that sells.

Guys like you are basically guinea pigs (hope you know what that means).

It is all a matter of priorities Humus... just like if and when you decide to make sure your demos run on all cards and all kinds of drivers. Either you want to spend the time, or you don't, for the reason only you know.

That's just me telling it like it is.

 

Surely guys like Humus should be supported just the same as game developers, however? After all, if he discovers driver bugs and reports them it can only lead to better drivers for all in the future. Perhaps a 'hobbyist' like Humus would discover bugs which might later have turned up in games being developed by a major company. If these 'would-be bugs' have already been fixed, it is less work for the game developer which, in theory, should give them more time to improve the quality of their product.

Therefore, it would seem to me that technically able hobbyists like Humus are, in fact, a very good thing and should therefore be supported properly.

By the way, I do realise that Humus doesn't want to remain a 'hobbyist' indefinitely.

 

No no you misunderstood me, or I wasnt clear. I was using NWN as an example of nVidia getting a developer to make it look like the game only worked best with their cards, even though other hardware is more capable (ATI/Matrox).

What I find funny over at the Bioware forums are the why dont PS effects work on my Gf4MX

 

Sure guys like me doesn't have the same priority. Still, you'll find my demos on tens of thousands computers around the world. Nowhere near as many affected as on real commerical games though. If we were talking about getting help with using certain features or other kinds of developer support, then sure, I fully agree with you. But driver bugs should always be regarded as significant, and just writing an email back saying, "we'll take a look at it" takes about 5 seconds. I can't really imagine that they don't take a look at it. It's like they don't want to admit they have bugs, it's a stain on their "gold standard".

 

I think like Humus. Right they are not on top of the list, but they need feedback when they point at something wrong. It's not in dismissing programmers that you will be able to improve.

 

I tried the Rallisport demo on my R8500 and the ice looks just as nice as in the first screenshot

 

Well do you think M$ would really not write a game to their own API standard?

 

NVidia has hundreds if not thousands of developers to support. Many of these people are working on big games right now for big publishers like EA. NVidia needs these games to work on their cards more than they need Humus's demos to work.

I'm sure his bug report was filed, but no way is an engineer going to follow up on him immediately if John Carmack and Tim Sweeney are calling in with problems and if no one knows you.


ATI doesn't answer everyone immediately either. I've seen questions go unanswered on the support board. Humus, you now have a rapport with them, there are guys there that when they see your email address, know you, and are more apt to respond. But if you were flooded with questions about one of your demos, I'm sure you would not be able to respond to every email right away and would respond to "friends" or "aquaintances" that you recognize first, especially if it was someone from the "press" like Beyond3d.

What ATI does for you, NVidia does for NWN. As a result, you feel that ATI has better support, and NWN feels that NVidia offers better support.

There is no objective measure of which is better, it's just your personal experience. Now, if you have data on the percentage of ATI emails and issues that go unresolved over X period of time vs NVidia, I'd like to see it, but somehow I don't think fighting anecdotes is very scientific.

 

Hey DemoCoder...have you told that to Derek Smart?

 

While everything you say is true DemoCoder there's also the fact that I experienced this good support from ATi pretty much from day 1. Sure, they have become more receptive to me too over time, but even my first emails to ATi in the days I was pretty much unknown to them, or if the knew anything about me that would be as the guy who wrote the tweaker that screwed up everyone's drivers, and then under the name "humus" and not my real name as shown in emails from me. Yes, my first email to them if I recall it right actually was related to the old Raid-on tweaker, and I got a reply on it.
For nVidia though, I don't think I have ever gotten a reply. I think that's the least thing you can demand. At least as a confirmation as someone has read the mail. While I don't neccesarily expect them to fix my bugs soon I think a simple reply should be the minimum. And as the bugs get filed one should expect them to file the email address to the guy who reported it too and send an email back when it's fixed. I call that simple organisation and should be performed by everyone. I have learned about my bugs being fixed through discussions on opengl.org, through friends or otherwise, but that's just through the wrong channel.

 

While it is true that games need more DevRel support than demos, many of the significant driver problems have been found in demos, not full scale games.

Hobbyists are much more likely to experiment with new features and functionality than professionals, and that is where many of the driver problems reside.

So support for both is important, but for who has the best developer relations I'm not too sure. Early this year I would have said nVidia, because I had a handfull of really great developer relations people at nVidia that I could contact and get a good (accurate) response back very quickly. Now, many of them have moved to ATi and response times from nVidia have been growing, while shrinking from ATi.

 

Humus, I'm just telling it as I see it.

Remember that I said that guys like you are guinea pigs - you may discover bugs that some hotshot developer haven't discovered yet... and perhaps the IHV you informed have indeed looked at your bugs, fixes it internally and when a game developer later then tells them about this bug (which you have the distinction of discovering), this IHV tells them "Yes, we know about this, we have fixed it, it'll be in the next beta drivers...". All without acknowledging the fact that a "stranger" discovered this bug, which means this IHV is saying they know about it (discovered it) themselves. Hey, a possible scenario right?

And don't take this to mean that I'm talking about NVIDIA specifically.

Also, perhaps the fact that some of your earlier demos were featured on ATI's site a while ago may have something to do with this lack of response from NVIDIA? And that you publicly stated that you were going to join ATI? Possible reasons, no? Is NVIDIA being totally unresponsive or being paranoid or just taking a cautious stance (particularly where you're concerned)?

 

Thats strange ...

What kind of BIG NVIDIA SECRET will be revealed to ATI if they acknowledged the bug that Humus found?

 

Just to give you a different spin...

I understand where Humus is coming from. During the time that I was using the Parhelia, I was quite shocked at the level of support I was afforded, in terms of relaying drivers bugs to Matrox, and then getting feedback on these "action items."

I know for a fact that they weren't just doing it because I was in the process of completing a review...

Anyhow, I strongly believe that nVidia should do a hell of a lot better in opening up channels to the "small fish" out there...be it guys like Humus, or similar.

On the flip-side, I can sorta' understand their position...There are only so many people @ nVidia, and there are literally millions of users. Things are a little bit different when you're sitting on top of the heap, as is the case w/ nVidia...though ATI is not exactly a small time outfit either.

 

Hehe, probably nothing really. Depends on the kind of questions asked by Humus and the type of answers Humus expects. It may probably be more PR related more than anything else - developers don't "blab" about bugs for PR reasons... IHVs cannot be certain regular forum participants (as opposed to publicly known developers) don't do that. Carmack perhaps can get away with this though!

You'd be surprised that I have found bugs in NVIDIA drivers (or maybe even hardware, dunno) while doing some work - I relayed them to NVIDIA, got no response after about a week, so I tell it to some known developer, who then (after a short discussion with me) agrees with me that it is indeed a bug (that developer acknowledges that it wasn't something he knew about until I brought it up, and that it could potentially be a problem to him in his work) and that he'll tell NVIDIA. After a while, that developer tells me NVIDIA has fixed it.

Just as an example.

 

I know you Humus, at least from the forum boards and the demos. Keep up the good work.