Carmack speaks

[glen]

http://www.webdog.org/cgi-bin/finger.pl?id=1&time=20020211165445

 

[NFS4]

Damn, Carmack just unloaded on the GF4 MX and NVIDIA's naming convention

 

[Adul]

ouch, maybe I will decide on the 8500 still yet.

 

[nortexoid]

i think everyone agrees....

 

[GTaudiophile]

<< A GF3 or 8500 will be much better performers...As usual, there will be better cards available from both Nvidia and ATI by the
time we ship the game. >>



Hence, I'm not upgrading until the game comes out. Sounds like ATi is working hard on their driver issues, but still has a little ways to go. Wonder how much more potential the Radeon 8500 has?

 

[Soccerman]

I just have say that only with developer feedback like what Carmack seems to provide will ATi produce near perfect drivers. ATi seems to fix nearly everything he points out quite quickly.

also, either ATi's driver team is in no-way related to the hardware team (which could be where some of the bugs come from), or they just don't follow/test OpenGL and DirectX API standards/calls right off the bat, but let the developers find the problems... that kind of bugs me..

 

[SteelyKen]

<< A high polygon count scene that was more representative of real game graphics
under heavy load gave a surprising result. I was expecting ATI to clobber
Nvidia here due to the much lower triangle count and MUCH lower state change
functional overhead from the single pass interaction rendering, but they came
out slower. ATI has identified an issue that is likely causing the unexpected
performance, but it may not be something that can be worked around on current
hardware. >>



If he is referring to the "high poly bug" that many people have reported on the 8500, it doesn't sound like he believes that this can be fixed with a driver update. I wonder if the problem with Doom is related to the high polygon slowdown experienced in other games.

 

[MrBond]

<< Damn, Carmack just unloaded on the GF4 MX and NVIDIA's naming convention >>

And rightfully so if you ask me.

I had no idea the GF4MX was so neutered. And to think, they're going for $179. For a Geforce2. Good GAWD.

 

[John]

I expected GF4 to be the speed bumped GF3, but calling the NV17 GF4-MX really sucks.

 

[Mltsao]

Nvidia naming a faster GF2 a GF4MX is worse than AMD's Preformance Rating in my opinion.

People have told me that they wanted to buy a GF4MX thinking that it was a slower version of the upcomming GF4 Ti series which is why i think this.

 

[AGodspeed]

Nvidia's OpenGL drivers are my "gold standard", and it has been quite a while since I have had to report a problem to them, and even their brand new extensions work as documented the first time I try them. When I have a problem on an Nvidia, I assume that it is my fault. With anyone else's drivers, I assume it is their fault. This has turned out correct almost all the time.

Nuff said....

 

[GL]

Sounds like ATI relies on Carmack for their quality control. I feel for Carmack. I know first-hand how annoying it is for your code to work well except the underlying third-party code you rely on is riddled with bugs.

 

[billyjak]

Maybe ATI should hire Carmack

 

[BD231]

Maybe ATI should invest more time in creating a completely compatible 3D chipset and driver code that they actualy understand the in's and out's of insted of fixing little glitches on poorly implemented drivers. The RAW specs on the 8500 are awsome, but the drivers hold the thing back like nobodys buisness. I bet N-Vidia's driver team could do a better job of devoloping drivers for the R200 core...

 

[bluemax]

Another problem here is that Carmack has been sleeping with nVidia for so long all of his routines are based on the Geforces.
So *naturally* when *anything* else comes along, it's not "as good" because he has to do things differently.
While it is nice to see ATI working on their drivers to suit the Doom3 engine which will wind up powering a dozen or more games (none of which I'll like) perhaps the drivers aren't as flawed as he says they are...

How about reports from OTHER software designers?

 

[NFS4]

<< Another problem here is that Carmack has been sleeping with nVidia for so long all of his routines are based on the Geforces.
So *naturally* when *anything* else comes along, it's not "as good" because he has to do things differently.
While it is nice to see ATI working on their drivers to suit the Doom3 engine which will wind up powering a dozen or more games (none of which I'll like) perhaps the drivers aren't as flawed as he says they are...

How about reports from OTHER software designers? >>


Is it Carmack's fault that it took ATI so long to come out with a competive driver in terms of performance and relative "cleanliness" compared to NVIDIA drivers?? I think not. Carmack was simply stating that when he works with NVIDIA drivers, he doesn't have to deal with bugs and inconsistencies at the driver level. When he gets to ATI, he has to call them up so that they can fix their bugs.

That's just like saying that the American public has slept with Japanese cars for so long and not paid attention to Korean cars. The fact of the matter is, Korean cars were known for crap quality and poor resale value until just recently. NOW they are starting to get closer and closer to the Japanese in quality and features at a lower price point (sound familiar?). And their sales are rising as a result.

The point is, don't blame Carmack b/c he is used to a programming with bug-free drivers; i.e. Don't blame the American public for being used to driving reliable cars.

Everyone will get their turn...the Koreans are now getting more attention from the buying public b/c of their increased reliability, low price, and long warranties, but they still have a bit further to go to make a full swipe across the Japaneses' bow. ATI is starting to get better with their drivers, but they are still not quite up to Carmack's standards; but he is giving that chance by working side by side with them.

 

[bluemax]

But is it really a matter of bug-free drivers or the fact that he has to do something different from his pre-programmed routines?
Like: I use C++ for programming all the time, but these guys give me (insert OTHER language here: Pascal, COBOL...) so I'm gonna' say they SUCK until they send me C++ code so I don't have to work around it.

 

[Leon]

<< But is it really a matter of bug-free drivers or the fact that he has to do something different from his pre-programmed routines?
>>



If you read again, you will see that he originally used his standard ARB codepath, which any OpenGL 1.2 capable card should run 100% perfectly. No GL_NV, EXT_NV, of any kind.

Leon

 

[Alex]

<<

<< A GF3 or 8500 will be much better performers...As usual, there will be better cards available from both Nvidia and ATI by the
time we ship the game. >>



Hence, I'm not upgrading until the game comes out. Sounds like ATi is working hard on their driver issues, but still has a little ways to go. Wonder how much more potential the Radeon 8500 has? >>



damn the 8500s drivers have matured sooo much since their release... now with all these leaks im sure 8500 will destroy ti500 @ doom3.... now gf4... welll thats a force to be reckoned with... at least until r250 and r300 come out

 

[bluemax]

Still not bad for a video card less than $200.

 

[NFS4]

<< Still not bad for a video card less than $200. >>


Makes you wonder how fast the 8500 would be now if ATI had got the drivers at 99% at launch:Q

 

[bluemax]

You know, if the Geforces ever got a RADICAL new redesign (instead of rehashes of the old core) we could probably expect glitches from nvidia as well. Even the GeForce3 isn't THAT big of a change over the GF1/2/MX/etc... but probably the biggest change since the GeForce was released.
I see what you mean though... if the drivers only continue to improve, they'd be 10% faster now. (Or something.)