There s certain degree of objectivity editors have to keep to retain their stature of trusted source when they start to get more and more readers. There are only a handful of exceptions in the world of IT hardware and right now one of them is much too active to miss him in what we decided to be a simple true/false review of current NVIDIA position. Charlie from semiaccurate started to sound more than angry lately and while he still keeps the politeness in his forum posts, most of his articles started to look like a regular forum troll postings. In fact – an ATI fanboy postings, and one of those that don’t have much technical knowledge and substitute it with pure hate. There is also certain degree of envy in our words – not everyone can sound as deluded as Charlie Demerjian while keeping his readers (he even gets more and more) but we certainly prefer to stay objective instead of taking the one way road. Because you know that once you take that road there is only a couple of times you get to fail before you get lost – even Neo can only dodge a few bullets before he gets hit, right? We won’t run Charlie’s article analysis here, but all in all we think he has someone over TSMC, as most of his accurate postings seems to be either about chip manufacturing, or anti-NVIDIA PR analyzes. The rest… well, the rest of the ATI/NVIDIA stuff  is just too much crap to be analyzed. Most of it is included in one simple form with 10 points below:

As promised – truth as simple as we can say it:

1. NVIDIA doesn’t have DirectX 11 hardware!
It is obvious, no need to remind us every now and then! In fact NVIDIA was never supposed to have new hardware by the time ATI released HD5xxx. Those two companies never release their products by the same time. Still there would be delay of future NVIDIA graphics cards as indicated by TSMC inability to get good yields with their 40 nm process. NVIDIA chose to make a very complicated GPU, which makes things harder.

2. But they said they have it and it was fake!
So what? ATI screamed about r600 for months before they had a real product, Intel released like 2 benches about their Core architecture 6 months before we could get one, and NVIDIA wanted to make themselves look something more than chip engineers. There are thousands of products which sit at their early development state and are still advertised as finished. It was epic fail, yet a one that you will forget once you get a Fermi board.

3. They will fail to make GF100 better than ATI cards!
Yeah, right. Get an HD5870 then. Or better – a couple of 5970… :)

4. It won’t have tessellation and it won’t be DirectX 11 compatible this way!
The simple truth is that NVIDIA said they will have a hardware tessellator, so there is no point arguing about it. But if NVIDIA wants they could run it on software level, if they manage to squeeze enough speed. GF100 is expected to be programmable at levels that allow NVIDIA to actually make most things at software level if they want to, and we are not talking about CPU-software level. Of course ny ATI fanboy would tell you this is not the same, but actually it is – NVIDIA is known to have good software support, so if there is no difference in speed you won’t care what levels it runs on.

5. NVIDIA has bad practice renaming cards
As well as ATI sometimes, although NVIDIA’s much more pronounced lately. They seem to have some marketing issues which everyone hates, but they all seem to be OK with it. NVIDIA actually made a good step in renaming mobile chipsets, but they managed to make quite a mess from it too. If you want to see “bad practice in renaming” just take a look at processor market. Intel and AMD make new names for every new step in technology and nobody gets angry. We don’t like it too, but there is no point crying about it all the time.

6. CUDA is now obsolete now with open standards!
CUDA will continue to develop, and it already features OpenCL support, while NVIDIA cards support OpenCL at same level as ATI cards. There is nothing that will make CUDA obsolete as it far more advanced by now and it’s widely used by software developers nowdays. Just look at those video encoders and decoders with CUDA suport. Stream was around for years but you cannot find even simple plugins that will make use of ATU GPU programmability. Featuring OpenCL in 2.0 ATI Stream SDK final release won’t make Stream more popular, but compatible, so there wont be any difference once NVIDIA get Fermi out. A common quote for NVIDIA that explains it: “The OpenCL standard was developed on NVIDIA GPUs and NVIDIA was the first company to demonstrate OpenCL code running on a GPU”

7. They say Eyefinity is better than 3D Vision!
Don’t know who are “they” but Eyefinity is nothing like 3D Vision. It’s comparing apples and oranges. And by the way NVIDIA just announced 3D vision Surround (still unofficial) which is essentially the same as Eyefinity plus S3D, but both technologies would suffer from compatibility problems, as games were never meant to be played this way. Also there is no real FOV expansion in those in different planes, they all use one flat plane in front of you, which makes them impressive in size, but awkward compared to simple human peripheral vision.

8. 3D Vision is not something NVIDIA invented!
They never said that. Before NVIDIA started to work in the S3D niche almost nobody knew about S3D at all. Now everyone have heard of it and most people have tried it. They just got the edge – it is branded, it has compatibility certification and amazing driver support. It is widely supported and manufacturers of displays and even those of projectors also like it and started to make more and more compatibles screens. It’s easy to make even 2D movies to look like 3D for watching with 3D Vision glasses. NVIDIA also used well the fact 3D cinema raised in popularity lately. To put it simply – 3D Vision only problem (for ATI) is that it only works with NVIDIA graphics cards and nowdays, at least until we see fermi based boards, they are a bit obsolete in terms of performance.

9. ATI has best architecture!
They’ve said that before. Even the biggest failure of NVIDIA in near past – the FX 5×00 series, made a good step forward to what eventually became some of the most advanced NVIDIA graphics cards. Which ATI couldn’t answer for years, if you remember, although they had good competitors in most cases and pushed NVIDIA to lower their prices. To say one or other architecture is better just because they have similar performance while one of the cards being a fifty percent (or why not more) bigger in terms of transistor count is, again, like comparing apples and oranges. Some people like to do it, we call it stupid.

10. NVIDIA pays to game developers, reviewers and important decision makers!
Considering ATI has AMD behind its back and NVIDIA has a few drops of business here and there, it is strange to claim that NVIDIA pays for almost everything that sounds like a rock in the ATI’s garden. Like we said – ATI didn’t manage to make much use of Stream, while NVIDIA has put lot of efforts to make CUDA work for professional or even home users. Then with their game compatibility program came the assurance that give game runs flawlessly on NVIDIA hardware (or not so sometimes). What some of the developers said is that they just got the best support they’ve seen. Why call the white  black and say it has money involved? Because ATI never had such software support. Oh, well, Batman is NVIDIA advertisement, but that’s rare stuff, you know it.

There is more but it gets boring at some point. There is no point arguing ATI has the performance edge today and that their software development team is getting better (at last, we never thought we’d say that), but that is today. As for Charlie – there is always a dog barking around the caravan, but we have a suggestion for him: get a life.

Tags: 3d vision, ATI, ATIStream, Charlie Demerjian, CUDA, Eyefinity, Fermi, GF100, NVIDIA, NVIDIA 3D Vision, Open CL, semiaccurate

Related articles:

• NVIDIA is going to use Optimus with all new mobile Fermi
• Kaspersky Lab is using NVIDIA Tesla S1070 to quickly identify new threats
• Adobe and NVIDIA made it… or not?
• Acer releasing new 17-inch Aspire 7740G with DirectX 11 graphics and using all new Core i3, Core i5 and Core i7 processors
• First ATI Mobility Radeon HD5650 benchmarks!

Quick navi: