State of the Graphics Industry Rant

Sabastian

Regular
Seems there are more then just a few "fanb@ys" upset with the industry. Here is a snip.

http://www.hardwareaccelerated.com/articles/reviews/graphics/stateofind/

"The most concerning is when a reviewer neglects to state the difference between two graphics chips and their differing anistropic filtering methods or their different antialiasing methods. I remember when the Radeon 8500 came out; many called its anistropic filtering method flawed. While it did have problems, which are said to be fixed with the Radeon 9700, many claimed it was flawed because it strayed from the norm. This mentality isn’t good for the industry. Many reviews I have seen show one video card with multisampling while another used supersampling for antialiasing but never stated how one may be better while another may be faster. Sometimes when both cards have identical settings available, a different setting is used by a reviewer due to lack of research on his part. At the same time the point of benchmarking itself is becoming less and less important. Sure, you always need to know what really is faster, that’s a give in, but there is only a certain point where that needs to go. Quake3 can only be benchmarked so many times."
 
However I can't comment on the rest of the rant as I haven't read it yet. How is that for laziness? ;)
 
Yeah .. it is a fairly good "rant" but when the writer suggest nvidia is going the way of 3Dfx as a result of the NV30 being late well... I beg to differ. For the most part though I do agree with most of his "rant".
 
delivering the Radeon 9700, containing over 100 transistors

Wow. 100!

I remember when the Radeon 8500 came out; many called its anistropic filtering method flawed. While it did have problems, which are said to be fixed with the Radeon 9700, many claimed it was flawed because it strayed from the norm.

So, a problem is not a flaw if its simply was different from the norm, yet was fixed? ;)

I question calling it a rant. Maybe a ramble. The only thing he seemed to be ranting about was poor quality of reviews, which I certainly agree with, but I'm confused about what exactly he was calling into question with reviews.

But back to the anisotropic thing. Was it a problem in his eyes? Or was it acceptable but different? Is he advocating "so and so card does it differently, so its OK to have inferior results", or ????
 
Is he advocating "so and so card does it differently, so its OK to have inferior results", or ????

It's plainly obvious to me that he is advocating "so and so card does it differently...the different approach has pros and cos when compared to the way we've seen it implemented in other hardware."
 
Wow. 100!

LOL yea messed that up I'll edit that later. I didn't catch that one.

But back to the anisotropic thing. Was it a problem in his eyes? Or was it acceptable but different? Is he advocating "so and so card does it differently, so its OK to have inferior results", or ????

It was flawed, but when the Radeon 8500 was released many complained about things that turned out not to be flawed with their methods. I'm just saying that if any company does something different, it is seen as negative many times.

nvidia is going the way of 3Dfx as a result of the NV30 being late well

Yea, I seem to have emphasized that a bit to much. All i meant was that if that product cycle delays/misses are important and do have effects. NVIDIA's not going anywhere anytime soon, and will most likely be the market leader for some time to come.
 
I'd just like to say that I really do not remember many reviewers calling to attention the drawbacks of the Radeon 8500's anisotropic filtering method. Not many even noticed the issues, apparently. In fact, there's only one that sticks out in my mind, and that was in Tom's review of the Parhelia, where he didn't want to benchmark the 8500 against the Parhelia/GF4 with anisotropic filtering enabled (As a side note, he also didn't benchmark the Parhelia against the Radeon 8500 with FSAA enabled for that review...which makes good sense, as the GF4 and Parhelia's FSAA methods are more similar).

I'm not sure any fewer described the drawbacks of MSAA.

But, I personally agree with the description that the 8500's aniso was flawed, while MSAA works properly.
 
Chalnoth... :LOL:

IS their FSAA the same, FAA vs MSAA..NO...your logic is flawed bigtime..
Every card does things differently, so now all IHV's have to do the exact same thing before reviewers are allowed to benchmark it :rolleyes:

The 8500's AF is not flawed..its different live with it....the same could have been said about texture compression on GF1-GF2-GF3-GF4, they were not implemented properly and relied on tweaker hacks to lower the amount of compression...
In fact the texture compression issue on those cards will affect overall image quality far more than the loss of filtering on a 45 degree plane.

I'd also like to point out that Joe Schmoe doesn't automatically download Rivatuner after buying his $400 card from Bestbuy to fix those issues.
 
Doomtrooper is right, but what's the real issue here? Quality of the filtering is a definitive proof, and a tangible argument. Method is irrelevant compared to results.
 
Chalnoth...

IS their FSAA the same, FAA vs MSAA..NO...your logic is flawed bigtime..
Every card does things differently, so now all IHV's have to do the exact same thing before reviewers are allowed to benchmark it

The 8500's AF is not flawed..its different live with it....the same could have been said about texture compression on GF1-GF2-GF3-GF4, they were not implemented properly and relied on tweaker hacks to lower the amount of compression...
In fact the texture compression issue on those cards will affect overall image quality far more than the loss of filtering on a 45 degree plane.

I'd also like to point out that Joe Schmoe doesn't automatically download Rivatuner after buying his $400 card from Bestbuy to fix those issues.

My point exactly.
 
Doomtrooper said:
The 8500's AF is not flawed..its different live with it....the same could have been said about texture compression on GF1-GF2-GF3-GF4, they were not implemented properly and relied on tweaker hacks to lower the amount of compression...

Hmm, i remember that the GeForce cards got a lot of complaints about their texture compression. Same thing with multisampling and alpha textures which was brought up a lot.

Anyway, i agree with Russ here. Why was the 45 degree thing fixed if it wasn't a flaw ?
 
Fixxing

Who's saying it's fixxed, ATi? or a bunch of websites? If, it's ATi the question is somewhat valid, if it's a bunch of websites, the question means squat.

Look at itthis way, TVR release the Tuscan Speed 6 (which some could argue is flawed because of it's steering) a few years back, just recently they released the Tuscan GT-R. Now does that mean the original is flawed? No, it means they've figured out a way to improve on the original.
 
Anyway, i agree with Russ here. Why was the 45 degree thing fixed if it wasn't a flaw ?

Qualify fixed.

ATI say:

Although this algorithm was introduced in the original SMOOTHVISION anisotropic filtering technology, version 2.0 uses an updated version that does a better job of accounting for polygons rotated about the Z-axis (i.e. into the screen), ensuring that all surfaces and textures get the full benefits of anisotropic filtering

So its an updated algorithm - yes, anyone can say that if they 'fix' something. However, you you've got to look at is what algoritm Radeon 9000 uses, which is more or less the same design as 8500. If it still has the same limitation as 8500 then that would seem more of an implemtation limitation than something thats 'broken'.
 
Re: Fixxing

Heathen said:
Who's saying it's fixxed, ATi? or a bunch of websites? If, it's ATi the question is somewhat valid, if it's a bunch of websites, the question means squat.

Hope this helps.

1.jpg
3.jpg
4.jpg


Seems fixed to me.

Fuz
 
However, you you've got to look at is what algoritm Radeon 9000 uses, which is more or less the same design as 8500. If it still has the same limitation as 8500 then that would seem more of an implemtation limitation than something thats 'broken'.

Not too sure about that.

What we don't know (which would help provide the answer) is how much did it "cost" (in terms of silicon) to implement the improved version vs. the old one.
 
But wait, I'm still going to needle in on the subpoint of the argument, which was:

Different algorithms with different results are OK

and the subtext of which was

The 8500 anisotropic filtering was OK because it was done differently

Regardless of whether its fixed now, it was definately less capable then when compared to the competition. It didn't filter at 45 degrees off the UV axis.

Shouldn't reviewers say that? Or should they say, "each card is special in its own way, and momma loves them all the same"? Doesn't that completely negate the purpose of reviewers? To me, they're there to compare as many different products as they can FOR ME and give me their feedback, both with objectivivity and with subjectivity so I personally don't have to see them all with my own eyes. Reviewing a card in a vaccuum is useless.

Personally:
The GF2/3 DXTC decoding algorithm was plainly not as good as others, and it got well deserved flak. (Especially the GF3 certainly they could have fixed that)
Quincunx is damn useless of games with HUDs, especially on LCD displays. I'm so used to razor sharp edges that the fubared text drives me frickin batty.

Both of those were different implementations achieving the same goal as others in the field(and the DXTC one was technically correct according to the specification), but the end result was "measurably" lacking compared to other solutions and I'd prefer to know that.
 
Back
Top