More ATI Driver News from Derek Smart

[fresh]

Don't make up crap you know nothing about. The driver bug that was causing the hangs in certain games (The Thing, BF 1942 and Mafia(?)) is fixed. Software emu? Give me a break.

I sent you a PM yesterday when you asked about the bug. Did you read it? If so, then why make up this garbage?

If you are trying to hurt ATi with slanderous comments, then you need to step back and take a break.

It was a simple error in a code path that didn't get hit by our testing. End of story.

P.S. Now that I am looking at this "multitexturing" bug, I can report all sort of "interesting" things that Battlecruiser Millenium is doing. Do you want me to post a bunch of stuff about your programming skills, Derek? Or your lack of knowledge of D3D specs?

Actually, please DO!! I've dealt with ATI devrel a lot and they've always been helpful and quick with fixes. I used a 8500 is my main dev machine since it came out, and now I'm using a 9700 with 0 problems.

Who the hell sets the near plane to 0.001?!?!? No wonder he's having z fighting issues.

 

[Dave Baumann]

So, if I were to give an answer off the top of my head, I would say that when 3D 2001 runs on a DX8.1 part, it strictly uses Vertex Shading code, and not hardwired T&L code. One way to test this out would be to load up 3D Mark 2000 and try the software vs hardware T&L (as it is strictly using the old hardcoded T&L)

Josh, look at the tests I ran again - Max Payne does not use vertex shaders, but 'T&L and that clearly shows a difference with hardware and software T&L. AFAIK Dungeon Siege is also a DX7 core game as well.

 

[demalion]

]

From there, you can wherever you want, but I hope it is someplace productive.

I just read your post.

I SWEAR if you do that again, I'm gonna......oh wait, you don't live next door. Never mind

No seriously though, if I ever saw a really good fuzzification job, you just posted it. Good job. At the end of the day, I'm going to refer you, once again, to the definition of what BROKEN is, m'kay?

Hmm...I've looked up "fuzzification" in the dictionary. Perhaps not finding it is the source of my problem in seeing what you are talking about?

Perhaps it would have been more helpful to just simply clarify what is wrong with my discussion instead of using a made up word and requiring me to spam the thread to ask what you mean. If you think some aspect of my post is "fuzzy", I'd question what, as everything seems clearly stated. If you think my post "obscures the issue", I'd ask how, as the points seem just as valid as when I posted them, mainly because you haven't demonstrated any fault in them as of yet.

I still am missing your point, I guess, since quoting "the definition of broken" (which I did) and then being pointed to it again seems a bit circular, don't you think?

 

[fresh]

]In case there was any doubt, according to Dictionary.com, here is the definition of broken.

Thanks. Here's one for you .

 

[Derek Smart [3000AD]]

Who the hell sets the near plane to 0.001?!?!? No wonder he's having z fighting issues.

A person who obviously knows what he's doing?

Besides, which part of the word seeding eluded you in my post? Or should I go dig up a definition of this for you from dictionary.com

And get your facts straight. OpenGL_Guy is not in dev support. He's in driver development. Two COMPLETELY DIFFERENT camps. m'kay?

Now be a good boy and siddown unless you have something worth reading (apart from jabs at me) to add to the discussion.


ps: What game are you developing, again? Anyone, hiding behind an alias can post crap on the board. Being a developar and all, you should be able to provide this info. I'll be waiting.

 

[Joe DeFuria]

Thanks. Here's one for you .

Wow...compared to the definition of "Broken", that particular definition is very straightforward and precise!

 

[Derek Smart [3000AD]]

Hmm...I've looked up "fuzzification" in the dictionary. Perhaps not finding it is the source of my problem in seeing what you are talking about?

I still am missing your point, I guess, since quoting "the definition of broken" (which I did) and then being pointed to it again seems a bit circular, don't you think?

There you go again, with the accusatory tone. Just because you don't know what it means, doesn't make it a made up word. *sheesh* Whats it gonna take?

You obviously didn't look hard enough. I come from an AI background (primarily), so I applied the term as it related to fuzzy.

My point was that broken by any other name, is still broken.

I then posted the definition of broken.

You counted with circular diatriabe in a pitiful attempt at confusing and reducing the effects of my argument. In fact, you played right into the whole Clinton Syndrome I pointed out on the previous page.

 

[Derek Smart [3000AD]]

Thanks. Here's one for you .

Wow...compared to the definition of "Broken", that particular definition is very straightforward and precise!

OK, so we're going to reduce the discussion, once again to name calling?

Want to get me started?

 

[fresh]

Ok Derek, here goes.

I don't like your tone (I doubt I'm alone!). You take your driver complaints to public forums and slam the IHV (ATI in this case). What exactly do you hope to accomplish with this? I really love your "I dont give a shit what the IHV thinks about me" - attitude, but come on. Without the hardware, there would be no games - and without the games there'd be no hardware. So calm it down, M-KAY? I've had plenty of run ins with driver issues, but I don't go and make a big deal about it on public forums. I email devrel and wait for a response. Most of the time it's something that I did wrong. The other times the response is "Yeah, it's a bug, we'll fix it" - in which case I'll wait or have a temporary work around. Or it's "It's a hardware issue", in which case I put in another code path. That's your job as a programmer. You don't see the developer writers post on public forums about your shitty use of indexed prims, or how you're doing 10 state changes per triangle. M-KAY? So have some respect. It's called professionalism, and you seem to have none of it. Oh I'm sorry, you're Mr Badass Programmer who is out to set the world straight on how awesome your code is and how screwed up drivers are. It's like you're out to prove yourself. WHY?

Secondly, why the hell would you even seed the near clip plane with 0.001? Obviously if you're trying to model an entire universe from human scale to planetary scale, you're gonna run into z fighting issues. It's up to you to fix those by partitioning the zbuffer, doing multiple passes with multiple z-clears, or whatever. W-Buffering is dead. It's been long dead, give it up. Stop your whining and move on like the rest of us have.

And yeah, I still think you're an egomaniac. In one of your posts you went off about what a GREAT programmer you are. In my experience, I've found those people who claim to be great programmers to be the shitty programmers. That combined with your "Better Than Thou" attitude makes for a superb combination!

 

[demalion]

]

Hmm...I've looked up "fuzzification" in the dictionary. Perhaps not finding it is the source of my problem in seeing what you are talking about?

I still am missing your point, I guess, since quoting "the definition of broken" (which I did) and then being pointed to it again seems a bit circular, don't you think?

There you go again, with the accusatory tone. Just because you don't know what it means, doesn't make it a made up word. *sheesh* Whats it gonna take?

Well, for me to correct my statement that it is a made up word, it would take a link to a demonstration that it isn't one. I even provided a clear and exact reason as to why I thought it was made up "Hmm...I've looked up "fuzzification" in the dictionary. Perhaps not finding it is the source of my problem in seeing what you are talking about?".

You obviously didn't look hard enough. I come from an AI background (primarily), so I applied the term as it related to fuzzy.

Hey, a link! <- this means I recognize that it isn't a made up word.

Hmm...fuzzification seems to just literally mean an application of "fuzzy logic". I still don't see how that even remotely applies to my post, however. Let me try to find a specific definition now that you've got me started in the right direction...

"In other words, fuzzy logic handles partial truth. Instead of just considering fuzzy logic as a single theory, we should look at it as a process - a process of fuzzification!!! With this, you can generalize any particular statement. During the fuzzification process, you go from discrete (or crisp) to continuous (or fuzzy)." Ah, so after all this my guesses about "fuzzy" and "obscures the issue" were right, I guess context is useful. But I offer you the observation, again, that it is not on "www.dictionary.com" so perhaps your application of the word in this context is not entirely appropriate at this time? I assume you agree with this since you didn't post a link to it on www.dictionary.com...
Perhaps it would have been clearer to use a phrase instead of something from neural networks terminology, however, since we aren't talking about neural networks...it really is a made up word as far as common English usage goes, though maybe it won't be in a few years.

My point was that broken by any other name, is still broken.

Did you even read my initial post? I discussed this pretty clearly and in detail.

I then posted the definition of broken.

Well, you posted a link to a definition of broken, that scrolls down into several pages. I took several specific definitions of broken and discussed them, which I think demonstrated that I have read the definition of broken.

You counted with circular diatriabe in a pitiful attempt at confusing and reducing the effects of my argument.

Hmm...I quote a definition of broken, and discuss it clearly, and then you point me back to the definition of broken, and your conclusion is that I am circular? This confuses me...if my application is incorrect or you disagree with it, quote it and illustrate why...pointing me to a definition that I've demonstrated I've read by quoting seems pretty useless.

In fact, you played right into the whole Clinton Syndrome I pointed out on the previous page.

Hmm...is that your way of saying my points about scope and context are meaningless? Since Clinton did exactly the opposite of my actual point, it sounds like you read only part of the post, and stopped before I got to the part where I said, to paraphrase, "I'm not advocating that you arbitrarily provide your own context, but I am using this to establish a starting point showing that it is important to use a context that makes sense not just to you but to the people with whom you are communicating". This seems to rather un-Clintonesque. Since my quoting of "a" definition of "broken" came after where I stated this, perhaps this explains this entire "definition of broken" circular discussion?

And why have we spammed the thread with several posts and the only thing established is the meaning of a word not in common use? Still nothing about why my post is "obscuring the issue."

 

[Sharkfood]

Derek-

Getting back to the MT problem you have been describing...

If you change observer Z, do the texture layers switch between top-most/lower like a toggle? If so, this may indeed be a reproducible bug that can be illustrated in multiple existing games, but is a bit tougher to reproduce in them.

If your problem is indeed related or the same issue, this might give the priority needed to help get this fixed in a shorter time window.

Cheers,
-Shark

 

[HogWash]

Pathetic

Ok, I just have to reply to this thread at this point.

Could you PLEASE let the damn definition of BROKEN go? This is pathetic. You go into some grand tirade about the definition of a word that needs no definition.

If something is not working, it is broken. If you fix something, then it was broken. The drivers were fixed, therefore they were broken.

It's as easy as that!

You don't fix something that isn't broken, right? LET IT GO!

Do you have nothing better to do than discuss the definition of a word that anyone with half a brain should know the definition of. Yes, the definition can vary in CERTAIN CIRCUMSTANCES.

Derek is using the word in the way everyone uses it, so just STFU already and let it go.

This "discussion" is very interesting, but I'm not interested in hearing your drivel about the definition of "BROKEN". I want to read what Derek, OpenGL, Dave, and other knowledgeable people have to say about the topic.

EDIT: Specific reference for this post goes to demalion

 

[LittlePenny]

Perhaps if some are having software issues, they could be settled publicly in the Architecture forum. It would be an easier read for those of us still trying to learn.

 

[Bigus Dickus]

HogWash, as much as you might wish it to be the case, definitions are rarely so simple.

An engine with a bad O2 sensor is clearly broken, right? I mean, there's something to fix, so the engine must be broken. But wait, the engine still runs, so perhaps it's not broken after all. Further testing reveals that only under very specific circumstances does the O2 sensor result in a malfunctioning engine.

So do you tell your friends that your engine was broken, or that the O2 sensor was broken? WAIT! Bait and switch you scream, we were talking about broken engines, not broken O2 sensors... right?

And there's the heart of the debate. An engine is not a single component, but an assembly of numerous components. When one component fails but the engine continues to work, it is common English to refer to the component being broken--not the engine. If you tell someone your engine is broken, it means it doesn't run, period.

Newsflash: TnL functionality isn't just a single bit of code, but an assembly of functionally different code structures. If one obscure part isn't functioning correctly, is the whole TnL unit broken? Does it still work for all but a few rare cases? Isn't that precisely the same as a vehicle engine?

Ahh... but that's demalion's whole point. Context decides what relative meaning should be applied to a certain term. Mr. Smart refuses to accept that context has any bearing on his definition. Ironically, if that were the case, all drivers for all cards would be broken. If he is unwilling to accept that there are contextual lines drawn between parts of the whole, and the whole, then he must accept that all drivers contain bugs somewhere, and as a result, the driver as a whole must be broken.

So which is it? Does Mr. Smart get to define where the contextual lines should be drawn, or does common sense? I'll side with common sense.

I apologize HogWash, since you emphatically stated that you didn't want to see any more discussion about the definition of 'broken.' But, this being a public forum, and me having my own useless opinion, oh well...

 

[demalion]

Re: Pathetic

Ok, I just have to reply to this thread at this point.

Could you PLEASE let the damn definition of BROKEN go? This is pathetic. You go into some grand tirade about the definition of a word that needs no definition.

I'd ask you to look at how many pages of this text are based on precisely a demonstration of the opposite of the following statement of yours:

Derek is using the word in the way everyone uses it, so just STFU already and let it go.

My problem with your post is it ignores statements I made in my post, mainly by agreeing with them in conclusion and repeating the gist of them, yet condemning me for addressing the same issue without using colorful abbreviations such as "STFU". Also, you seem to not realize that every person who you name by name was discussing this prior to my post being made.

This "discussion" is very interesting, but I'm not interested in hearing your drivel about the definition of "BROKEN". I want to read what Derek, OpenGL, Dave, and other knowledgeable people have to say about the topic.

In case you didn't notice, I was responding to the issues some of the people you mention have with the word broken, not creating the issue myself. This makes your post seem a bit contradictory.

As I've suggested before, one shot posters on BOTH sides of the "Derek Smart" fence, please stay away?

EDIT: Specific reference for this post goes to demalion

Hmmm...well, for what you are criticizing, you're addressing everyone you list by name including DS, except you are saying only DS is correct to do so. I won't be addressing this tangent of yours anymore here, but feel free to make a rebuttal here (no reply from me to save spam) and/or continue the discussion in PMs.

 

[Derek Smart [3000AD]]

Derek-

Getting back to the MT problem you have been describing...

If you change observer Z, do the texture layers switch between top-most/lower like a toggle? If so, this may indeed be a reproducible bug that can be illustrated in multiple existing games, but is a bit tougher to reproduce in them.

If your problem is indeed related or the same issue, this might give the priority needed to help get this fixed in a shorter time window.

Cheers,
-Shark

Yes, that is exactly what its doing to a certain degree. And you've already seen the results in the shots I presented. If not, here they are again.

Compare these two identical scenes taken from within my terrain editor

MT normal on 8500
MT broken on 9700 Pro

I'm still baffled because on 09/10, I got email from dev support that they had identified the problem and were working on a fix for release asap.

Between yesterday and today, OpenGL_Guy can't find anyone who has a clue what exactly this broken MT is about and he is investigating it himself.

BUT, instead of telling me what it is he is investigating, where its happening and why, he shows up, does a drive-by-pot-shot at me and disappears.

I'm still none the wiser. Don't get me wrong, I'm sure they're working on it fixing it, but I'd still like to know wtf is going on and what is broken. It could be another one of those incorrect code path jobbies OpenGL_guy was talking about wrt the broken hardware TnL. If thats the case, I'd like to know. And I'd like it fixed

 

[OpenGL guy]

]ok, lets not start THIS again. OK? Lets take this ONE step at a time. ZERO fuzzification, OK?

This is my last post I'll make. Do I give up? No, but I've realized I am wasting my time.

And as far as "fuzzification" goes, you should stop it as well.

As I understood it, that 09/10 ATI patch fix was for issues with those games *and* something to do with TnL.

I already KNEW about the TnL disabling issue, thats why I asked about it.

Ok so far...

So, I do NOT know what else you are talking about.

Really? How about your comment on a different forum that Doom 3 was probably running with SW TnL? See what I mean about fuzzification?

I sent you a PM yesterday when you asked about the bug. Did you read it? If so, then why make up this garbage?

See above

How does this answer my question?

If you are trying to hurt ATi with slanderous comments, then you need to step back and take a break.

Don't be silly. I'm shocked that you would even make such statements!!!

Is that so? Let me remind you of what you said earlier:

Thanks for showing up. I am going to, quite literally, rip you to shreds for all the aggravation you bastards in driver development have caused me.

Now as I am a representative of ATi's driver development team, I take this to mean you are attacking the whole team. In other words, you are badmouthing ATi employees, without good cause really. I guess I'll let Jeff explain it all to you. I'll be sure to tell him to use small words so you'll understand it. You know, when I was at SGI, we would charge customers for help debugging their apps, too bad ATi doesn't do the same, as I bet we could make some extra money that way.

Quite frankly, you should be ashamed of yourself for going there. Thats just OUT OF LINE.

Funny comment coming from you.

It was a simple error in a code path that didn't get hit by our testing. End of story.

I said ZERO FUZZIFICAITON, right? NOW, with that in mind, PLEASE explain EXACTLY what that simple code path was and WHAT it broke. My guess is, it has to do with TnL.

HWTL was never broken. Was the driver broken? Yes, under certain circumstances. This was my point. Surely someone with your extensive PCB design experience knows the difference between HW and SW.

Which invariably means, I'm right and you're wrong and pissed. (I'll use this now, thanks.)

Why do I need to say "EXACTLY" what was wrong? What does it matter? The bug is fixed, end of story (again).

So, I would tread very, very carefully if you decide to go along this path. I'm good at what I do. DAMN good in fact. As such, the end result might NOT be what you are expecting.

Just what is it you are good at? Shooting off your mouth?

Bring it on.

You don't want to go there, trust me.

As I promised, this is my last post responding to Derek. I'll just go back to the interesting threads.

 

[Derek Smart [3000AD]]

HWTL was never broken. Was the driver broken? Yes, under certain circumstances. This was my point. Surely someone with your extensive PCB design experience knows the difference between HW and SW.

You're KIDDING me right??!? Isn't this and ALL these discussions about DRIVERS?!?!?! Since when was hardware, as it relates to board design and specification, have ANYTHING to do with this discussion?

So of course in sayin that TnL is broken, I'm talking about the frigging drivers being broken. *sheesh*.

As for my comment on VE/MO, it was in response to someone asking how come come Doom3 was running just fine at E3 if it wasn't using hardware TnL. And myself, and SEVERAL people straightened that particular git out in this regard. In that (a) OGL and DX drivers are different (b) even if Doom3 was running DX (which is wasn't), the TnL problem could've crept up between driver releases ----> LIKE ALL THE UNEXPECTED (fog, specular, text buffering, TnL etc etc) PROBLEMS THAT CREEP UP BETWEEN DRIVER RELEASES BY ATI

Now as I am a representative of ATi's driver development team, I take this to mean you are attacking the whole team. In other words, you are badmouthing ATi employees, without good cause really. I guess I'll let Jeff explain it all to you. I'll be sure to tell him to use small words so you'll understand it. You know, when I was at SGI, we would charge customers for help debugging their apps, too bad ATi doesn't do the same, as I bet we could make some extra money that way.

If you think that stupid crap is going to scare me, you have a LOT to learn about me, my friend. I don't care how YOU or any of folks at the top of the ATI foodchain feel about my thoughts - they are MY THOUGHTS. You should go chasing down each and every person, including the media, who badmouth ATI and give them a righteous spanking over this abysmal driver development. Trust me, you DO NOT want to go down this path. It can only lead to a LOT of trouble for a LOT of people. Naturally, you have NO idea what I'm talking about. But, do carry on.

As for charging for tech support, ATI should try that. I'm sure the size of the class action lawsuit would be enough to create some new jobs in the legal business. And have you out, once again, looking for work.

Comparing SGI to ATI is just ludicrous to begin with. But, do what you must in defense of your FAILING and EMPTY arguments.

As I promised, this is my last post responding to Derek. I'll just go back to the interesting threads.

aw, c'mon. Thats the easiest copout I've seen all day. I'm not even going to respond to the rest of you post because it is pointless.

ALL I want to know is, WHAT IS GOING ON WITH THE BROKEN MT, m'kay?

 

[Althornin]

So muhc for the temporary clarity and maturity in your posts DS.
That lasted all of 4 posts.
Did your meds wear off?

You cant seem to comprehend simple ideas like the concept that your sweeping generalizations about issues cause problems (flames, misunderstanding, etc - i mean, if you were gonna buy a ATI 9700 and you heard someone say HWT&L is BROKEN you'd assume it didnt work at all.....) and though you claim to hate "Fuzzification", you in fact propagate it with your continued rudeness, your lack of compromise in your posting style (this closely correlates to your rudeness) and your lack of comprehension that YOU are causing more "fuzzification" here than anyone else.

If you cant discuss without merely discarding others reasoning "because you say so" (IE, no use of logic in response to posts, just trying to bull your way through by sheer force of assholishness) then please, go away.
Or, you could moderate your tone, and be more specific.
Which many people have asked you to do, and your responses have been "Nice fuzzification of the issues, blah blah BS" - probably because you lack the actual ability to rationally and coherently post and find it much easier to just scream worthless rhetoric at them about how "broken is broken" etc.
I mean, come on. You damn well know that words have a CONNOTATION and a DENOTATION.
The dictionary defenition covers one. You do the math about which one people are ahving a problem with, ok?

All anyone wants is
1) And end to your supremely prickish behavior (IE rudeness)
2) You to be more specific and end the problems with what you mean/what you say/what everyone else would mean by what you say.

Thanks,
Looking forward to a highly unlikely polite and rational reply.

 

[Sharkfood]

Yes, that is exactly what its doing to a certain degree. And you've already seen the results in the shots I presented. If not, here they are again.

Great, I can illustrate it in many existing games. It's ellusive as some games use blends with multitexuring which hides the issue fairly well (looks like a flicker in the blend), but others use direct layering like yours does.

I've got a handful of games with good reproduction steps, and it is Z/angle related. Perplexing at best. Two frames as an example, with the observer point moving closer (drew a line for reference so you can see where the viewer point moved closer):
Original viewing spot:

Move forward a bit, and notice the top layer texure on a full polygon (upper left) completely goes away:

It's visible (albeit not as apparent) in games that use MT for cheap reflections- where the reflectivity layer is overlayed with a model skin texture. The effect is certain values of Z where the reflectivity is gone and back again. Interestingly enough, I see this in both OGL and D3D so might be some 8500 code that needs reworking to handle texture layering on the 9700 better.