SimHQ - Bubba does the ATI 9700 Pro.......
- Thread starter martrox
- Start date
- Status
Bigus Dickus
Regular
bah, screw it.
S#it someone must have said the three magic words for that fiend to appear on these forums.
Derek Smart
Derek Smart
Derek Smart
http://www.somethingawful.com/article.php?id=61-4
look at the bottom of the page.
Derek Smart
Derek Smart
Derek Smart
http://www.somethingawful.com/article.php?id=61-4
look at the bottom of the page.
noko said:I am no developer nor am I a programmer, just thought I get that out. Anyways I think Derek brought up some excellent points such as no W-Buffer and a 24bit Z buffer on the Radeon 9700. Now I wish someone explain, maybe somebody already did (signal to noise ratio was low) what games need these features and what happens when they are not available. How it affects the clip plane from near to far and what the hell is the W-buffer used for and if it was ever used? Allright maybe I have nothing to contribute but those questions are bugging the $hit out of me
Here's some long winded points that might help you out and explain why some developers find it a pain in the a** to develop games on PCs:
1) W-buffer can benefit vast-large open scenes where the viewing distance is "huge" because more precision is maintainted at a greater distance from the viewer.
2) D3D apps using W-buffer must query for hardware support, and if there is non, use standard Z buffering... too bad so sad
.3) An example of Z fighting occurs when a game renders a shadow to a texture and then places the shadow texture _right on the ground_ AND is doing a Z compare. This, unfortunately, was about the only way to hack shadows before stencil shadows and shadow map hardware were available (which they are now). The "place shadow texture on ground" method fails to look good at all times because it essentially has the same Z value as the ground on which it is cast.
4) To alleviate the problem in #3, D3D has a ZBIAS renderstate that can be set to help the video cards decide if the ground or the shadow texture has priority and will hopefully display the shadow on top of the ground at all times (if it is set to have a higher priority). Unfortunately, ZBIAS is illdefined in the D3D standard (i.e. there is no standard way of implementing it and its consistency cannot be guarenteed across different video cards) and should not be relied upon.
5) Some games do decal texturing in a stupid fashion by placing a texture mapped polygon right on top of (on the same plane as) other polygons, to, for example, put a skull picture on a wall. This is the same problem as the example in #3 and is always destined to fail... _even_ if the skull picture is placed slightly out from the wall. The correct way of doing this is to produce a "hole" in the wall and put the skull picture in the "hole" and it is up to the programmers to relay this information to level designers.
6) One should never rely on ZBIAS in D3D the same way one should never rely on colorkeying (colorkeying is also illdefined in D3D and different cards produce different results). Colorkeying should be replaced with texutre alpha operations instead. Btw, no modern game uses colorkeying anymore... if they do, they haven't done their research.
7) As a previous poster said, higher accuracy closest to the viewer is more important than 100 miles from the viewer _in most apps_ and that is why Z buffering is sufficient most of the time. Note: OpenGL has no W buffer, and plenty of apps get by just fine without it.
Hoped this help cause some more confusion
crystalcube
Newcomer
noko said:
I think it has been highlighted earlier too..but...
W-Buffer support was optional and seems it has been dropped in DX9 so ATI decided not to support it in newer hardware.
As far as fallback & informing developer , the DX API provides the option to query the availability of W-buffer so if a developer has indeed programmed according to the API then there is no need for informing the developer.
So I really fail to see how ATI broke any atandards or API in their new hardware by not supporting a feature which was optional.
Bambers said:Derek Smart [3000AD said:
The G400 is a prime example.
well in D3D which is all DS cares about (I think) yes it was a prime example - it was the best card for purely D3D games at its launch time. Admittedly a slow OGL ICD must have been a bit of an oversight
He didn't have any problems with the drivers, so what?Derek Smart [3000AD said:
This is your opinion, stated as fact as usual.For one thing, the 775 drivers that ship with the card are as buggy as hell. A common trait for ALL ATI drivers that ship with boards. So, I have no clue wtf he's talking about.
That's funny as things are working better. Games like "The Thing" and "Battlefield 1942" are no longer having problems. There are always going to be some people who encounter problems no one else does. This is almost always due to user error.
A driver bug does not constitute "busted functionality". As Dave and others have posted, TnL on the 7.75 drivers was working fine with the vast majority of applications.
Maybe you should replace the "Ph.D." after your name with "Seer" as it's just as valid.
W buffer was never required for Direct 3D. It's not even well-defined. Direct 3D supports CAPS bits so that programs can see what features are available. If a certain feature, like W buffer, is not available, then the application has to deal with it. If the application can't deal with it, then the application is broken.
All I know is that the vast majority of games I have bought from companies such as id, Valve, Raven, Red Storm, Bungie, Activision, Blizzard, Sierra, etc. didn't need major patches to make them playable or to add basic functionality. Why can't I say the same about your games, Derek? You've released countless patches for your games, yet they are still rubbish.
You just can't stand it when someone has a difference in opinion, can you?
You keep saying "9xxx series". Don't you know that the 9000 and the 9700 aren't even related? The 9000 (aka RV250) is a derivative of the 8500, which is the R200. The 9700 is the R300. And you can say "suck" all you want, but it's still just your opinion.
I have yet to see one example of a multitexturing problem.
Z fighting implies multipass rendering (i.e. coplanar geometry), this is not multitexturing unless at least one of the passes uses two texture stages or more. And even if one of the passes were using multitexturing, that still wouldn't necessarily make it a multitexture bug.
Let's take this one piece by piece.
So, because you haven't noticed any changes, that means the drivers aren't improved?
You keep pulling links out of the air in a vain attempt to prove your case. Of course you neglect the links that show where people aren't having problems, even with the same games that others are. You neglect the links where people get their problems sorted out due to a configuration error. You neglect links where people are happy with their new card. It appears that, in your mind, one person complaining about driver problems, whether confirmed or not, means that the driver "sucks".
You know, the world doesn't revolve around Derek Smart: Kicking, screaming and throwing temper tantrums won't change that.
-FUDie
Why isn't this thread closed since page 2 ?
I'm sure we can use Beyond3D space and bandwidth in far more interesting discussion than this huge bashing and 10 year old boys comments.
If Derek Smart really want to say something he can make a new thread with meaningfull, precise informations, including screenshots and proofs.
So what's the point in leaving this thread open ?
[I'm not against arguing, I'm against bashing, and useless comments.]
I'm sure we can use Beyond3D space and bandwidth in far more interesting discussion than this huge bashing and 10 year old boys comments.
If Derek Smart really want to say something he can make a new thread with meaningfull, precise informations, including screenshots and proofs.
So what's the point in leaving this thread open ?
[I'm not against arguing, I'm against bashing, and useless comments.]
Is venting a good reason for leaving it open?
-DC Phd, MD, MPH, JD, Rear Admiral, and RN (Registered Nurse)
Project Lead, Product Manager, CEO, Senior Vice President of Engineering, Receptionist, and Mail Room Boy for DemoCoder industries, creates of the FlameCruiser2003 series.
Next patch for FlameCruiser will include ability to use new words like Discombobulate, Eviscerate, and MORON synonyms like ament, cretin, feeb, half-wit, idiot, imbecile, simpleton, dullard, dullhead, dumbbell, dummkopf, dummy, and ignoramus.
-DC Phd, MD, MPH, JD, Rear Admiral, and RN (Registered Nurse)
Project Lead, Product Manager, CEO, Senior Vice President of Engineering, Receptionist, and Mail Room Boy for DemoCoder industries, creates of the FlameCruiser2003 series.
Next patch for FlameCruiser will include ability to use new words like Discombobulate, Eviscerate, and MORON synonyms like ament, cretin, feeb, half-wit, idiot, imbecile, simpleton, dullard, dullhead, dumbbell, dummkopf, dummy, and ignoramus.
Randell said:well in D3D which is all DS cares about (I think) yes it was a prime example - it was the best card for purely D3D games at its launch time. Admittedly a slow OGL ICD must have been a bit of an oversight
With the first drivers it had major stability issues (esp at high agp rates) with quite a few systems.
Derek Smart [3000AD]
Regular
Its a little past 5.30am and am at work. I check my email to get a notification of several PM in my inbox. I check the PMs. Check the threads.
....and run into the same shit. I don't even know WHERE to start.
Nevertheless, I was going to start with this post - taking it apart at the seams, as only I know how - until I got to this bit. Then I went ...screw this. No way I'm bothering with this git So, I'm just going to post the bit and move on.
This is yet another example of assassination and fanATIcs deliberately taking my posts out of context, twisting my words etc etc. NOWHERE in this forum - or anywhere else, have I EVER - EVER used this word. What a damn lie.
Look, I don't mind taking on the world. I have a box of toothpicks. But if you're going to engage me, BE FAIR. I'm only ONE man you know. As such, do you folks have to sink to the depths seen in this thread and others, in order to get ahead (which, in itself, isn't bloody likely) of me?
Do yourself a favor and just ignore Fresh. If you've read the other thread, you'd know why. He doesn't have a CLUE what he's talking about. And I clearly described what I'm doing, how I'm doing it and why.
An example? He says that most devs use 1.0 ft as the nearZ and why should I set it to 0.01m? That right there, is the sign of someone who has no frigging clue what he's talking about - but likes to think so.
Even though I painstakingly gave a description of why its seeded at that value - he is the only one (surprise!) who choses to ignore this and continue to post the usual drivel. He must think I'm developing a shooter, in which all the pickups in the world as the same size as a 6' human. e.g. the TDU (Target Designator Unit) used to illuminate targets for precision munitions guidance, one of the smaller items in the game, is as tiny as 0.18m H x 0.12m W x 0.03m L
As for the W buffer issue, just TWO days ago, I was in an email trail with Mike over at dev rel related to this W buffer. To the extent that I even sent emails to some of my industry contacts (including JC) to see if they knew wtf was going on and whether they were in fact advised by ATI that they had (a) removed W buffer support from the 9xxx series (b) that W buffer was in fact, NOT in the HW at all
Below is my email to the head of dev rel, in its entirety. His reply basically said that the W buffer is not (!) in the HW, they had no intentions of supporting it - and that one of the devs will come up with a sample app to do it via pixel/vertex shaders (I already tried it - and as I suspected, it does not bloody work!). Thats when I told him that if they say it works, they can bloody well write the damn sample app like nVidia would if I asked them to prove something to me. He agreed. And I'm in a holding pattern waiting for it.
Of course, note that this shader solution won't work for legacy apps (such as my previous games, which rely on a W buffer)
I mean, they took out W buffer WITHOUT TELLING ANYONE. To the extent that the card is DX backward compatible and a W buffer is accounted for in DX6 and higher. The jury is still out on DX9, but as far as I know, is still available in Beta 2.
I see. I think it depends on the reviewer. I for one am of the opinion that the drivers that come on the CD are those that should be used. In the same way that I described off-the-shelf products. Bubba was quite right in using the 775 drivers.
heh, OK. I see you skirting the realms of prodding and joining in the fan club.
Ah, OK. I agree with you then. Besides, Bubba would be hearing from me if he had gone an' run Sacrifice, Morrowind or such - being a hardcore simmer and all, that would be like sacriledge.
If you're going to jump on the Hey look at me!! look at me!! I have something interesting to say!! No really!!! bandwagon, at least gets you facts right. Who else is making personal attacks apart from the Usual Suspects?
I think that there are enough to go around on the web. Go to the Rage3D forum and take your pick. I'm not in the mood.
Besides, I'm working on a research article about this fiasco for two hardware sites. By the time I'm done and that propagates across the Net and in print, we'll see who has a louder voice - me or the bunch of goons making personal attacks. I'm going to bury them so deep, they'd need a spatula to get out of it.
Its obvious that on forums like this on the Net, the signal to noise ratio is too high for anything meaningful to be derived from my postings. Thats the plan though - they feel that by distorting the thread and taking it downhill, the message is lost. Fair enough, I'll just pick a different battlefield. The more they try to stiffle me, the louder I'm going to keep yelling.
You'll read about it soon. Be patient.
Well said. If you look at my postings. This thread in particular, what you describe is exactly what goes on. As soon as they want to stiffle me, they start the personal attacks, character assassination, post distortion etc etc To the extent that you'd think I was running for office or something and this was an election year.
They WASTING THEIR TIME because I'm going to keep voicing my opinions as I see fit. The day ATI cleans up their act, they'll get a clean bill of health from me, from devs and from gamers. But I'm not holding my breath. There are STILL bugs in the 7xxx and 8xxx driver series, yet, we have a 9xxx series of boards on the shelves - with their own set of problems. ONLY in this industry like this will crap like that occur.
I for one, won't be surprised to see the first ATI class action lawsuit. From what I know, its coming. You'll see.
All good points. But show me one OpenGL program which has large outdoor scenes.
There IS a reason I chose to stick with D3D instead of adopting OGL several years ago when I migrated from DOS.
Every feature in DX is optional
W Buffer support has not been dropped from DX9
Listen to me, closely. No, not that close - back off the monitor a bit....
EVERY line of code that tries to do something such as this, checks the card's CAPS to see if it is supported. OK?
Good
Now, if the program checks for a W buffer, doesn't find it, then checks for a Z buffer, finds and uses it - thats normal. OK?
Good
If the program RELIES on W buffer (hence the check in the first place) in order to alleviate artifacts and use that as a preference, the first time the dev (such as me) is going to know that the W buffer no longer exists, is (a) when I notice artifacts (as I did in the case of the 9xxx series) or someone reports it or (b) if I knew this beforehand
NOBODY knew that W buffer was absent on the 9xxx series until the artifacts in their games started showing up. ATI never told anyone. It doesn't appear in ANY of the docs/bulletins on their dev site.
What? Do we now have to trace through every line of our code each time a new card comes out, in order to find out which CAPS report is failing?
You're kidding, right?
....and run into the same shit. I don't even know WHERE to start.
Nevertheless, I was going to start with this post - taking it apart at the seams, as only I know how - until I got to this bit. Then I went ...screw this. No way I'm bothering with this git So, I'm just going to post the bit and move on.
rubank said:
This is yet another example of assassination and fanATIcs deliberately taking my posts out of context, twisting my words etc etc. NOWHERE in this forum - or anywhere else, have I EVER - EVER used this word. What a damn lie.
Look, I don't mind taking on the world. I have a box of toothpicks. But if you're going to engage me, BE FAIR. I'm only ONE man you know. As such, do you folks have to sink to the depths seen in this thread and others, in order to get ahead (which, in itself, isn't bloody likely) of me?
noko said:
Do yourself a favor and just ignore Fresh. If you've read the other thread, you'd know why. He doesn't have a CLUE what he's talking about. And I clearly described what I'm doing, how I'm doing it and why.
An example? He says that most devs use 1.0 ft as the nearZ and why should I set it to 0.01m? That right there, is the sign of someone who has no frigging clue what he's talking about - but likes to think so.
Even though I painstakingly gave a description of why its seeded at that value - he is the only one (surprise!) who choses to ignore this and continue to post the usual drivel. He must think I'm developing a shooter, in which all the pickups in the world as the same size as a 6' human. e.g. the TDU (Target Designator Unit) used to illuminate targets for precision munitions guidance, one of the smaller items in the game, is as tiny as 0.18m H x 0.12m W x 0.03m L
As for the W buffer issue, just TWO days ago, I was in an email trail with Mike over at dev rel related to this W buffer. To the extent that I even sent emails to some of my industry contacts (including JC) to see if they knew wtf was going on and whether they were in fact advised by ATI that they had (a) removed W buffer support from the 9xxx series (b) that W buffer was in fact, NOT in the HW at all
Below is my email to the head of dev rel, in its entirety. His reply basically said that the W buffer is not (!) in the HW, they had no intentions of supporting it - and that one of the devs will come up with a sample app to do it via pixel/vertex shaders (I already tried it - and as I suspected, it does not bloody work!). Thats when I told him that if they say it works, they can bloody well write the damn sample app like nVidia would if I asked them to prove something to me. He agreed. And I'm in a holding pattern waiting for it.
Of course, note that this shader solution won't work for legacy apps (such as my previous games, which rely on a W buffer)
I mean, they took out W buffer WITHOUT TELLING ANYONE. To the extent that the card is DX backward compatible and a W buffer is accounted for in DX6 and higher. The jury is still out on DX9, but as far as I know, is still available in Beta 2.
LittlePenny said:
I see. I think it depends on the reviewer. I for one am of the opinion that the drivers that come on the CD are those that should be used. In the same way that I described off-the-shelf products. Bubba was quite right in using the 775 drivers.
LittlePenny said:]Just a question, I wasn't prodding.
heh, OK. I see you skirting the realms of prodding and joining in the fan club.
LittlePenny said:
Ah, OK. I agree with you then. Besides, Bubba would be hearing from me if he had gone an' run Sacrifice, Morrowind or such - being a hardcore simmer and all, that would be like sacriledge.
Althornin said:Yeah, It would be nice to see that from BOTH sides here...Nagorak said:Can't we just keep the personal attacks to a minimum?
If you're going to jump on the Hey look at me!! look at me!! I have something interesting to say!! No really!!! bandwagon, at least gets you facts right. Who else is making personal attacks apart from the Usual Suspects?
Reverend said:
I think that there are enough to go around on the web. Go to the Rage3D forum and take your pick. I'm not in the mood.
Besides, I'm working on a research article about this fiasco for two hardware sites. By the time I'm done and that propagates across the Net and in print, we'll see who has a louder voice - me or the bunch of goons making personal attacks. I'm going to bury them so deep, they'd need a spatula to get out of it.
Its obvious that on forums like this on the Net, the signal to noise ratio is too high for anything meaningful to be derived from my postings. Thats the plan though - they feel that by distorting the thread and taking it downhill, the message is lost. Fair enough, I'll just pick a different battlefield. The more they try to stiffle me, the louder I'm going to keep yelling.
You'll read about it soon. Be patient.
geo said:
Well said. If you look at my postings. This thread in particular, what you describe is exactly what goes on. As soon as they want to stiffle me, they start the personal attacks, character assassination, post distortion etc etc To the extent that you'd think I was running for office or something and this was an election year.
They WASTING THEIR TIME because I'm going to keep voicing my opinions as I see fit. The day ATI cleans up their act, they'll get a clean bill of health from me, from devs and from gamers. But I'm not holding my breath. There are STILL bugs in the 7xxx and 8xxx driver series, yet, we have a 9xxx series of boards on the shelves - with their own set of problems. ONLY in this industry like this will crap like that occur.
I for one, won't be surprised to see the first ATI class action lawsuit. From what I know, its coming. You'll see.
IT said:
All good points. But show me one OpenGL program which has large outdoor scenes.
There IS a reason I chose to stick with D3D instead of adopting OGL several years ago when I migrated from DOS.
crystalcube said:
Every feature in DX is optional
W Buffer support has not been dropped from DX9
crystalcube said:
Listen to me, closely. No, not that close - back off the monitor a bit....
EVERY line of code that tries to do something such as this, checks the card's CAPS to see if it is supported. OK?
Good
Now, if the program checks for a W buffer, doesn't find it, then checks for a Z buffer, finds and uses it - thats normal. OK?
Good
If the program RELIES on W buffer (hence the check in the first place) in order to alleviate artifacts and use that as a preference, the first time the dev (such as me) is going to know that the W buffer no longer exists, is (a) when I notice artifacts (as I did in the case of the 9xxx series) or someone reports it or (b) if I knew this beforehand
NOBODY knew that W buffer was absent on the 9xxx series until the artifacts in their games started showing up. ATI never told anyone. It doesn't appear in ANY of the docs/bulletins on their dev site.
What? Do we now have to trace through every line of our code each time a new card comes out, in order to find out which CAPS report is failing?
You're kidding, right?
crystalcube said:
Derek Smart [3000AD said:]This is yet another example of assassination and fanATIcs deliberately taking my posts out of context, twisting my words etc etc. NOWHERE in this forum - or anywhere else, have I EVER - EVER used this word. What a damn lie.
I beleive that was one of those 'enter a name' and generate a generic psycho-babble attack things. I dont beleive it was meant seriously.
I’ve seen lots of screenshots of issue with the like of Morrowwind and a number of other titles nobody is saying they don't exist. But I’ve yet to see any issues with all of the games Bubba has been looking up, from either Rage3D, or any forum, or from yourself – you are talking about issues with an absence of fact, we are only asking you to back them up.
Tribes?
- Status