I realise you're desperate to draw a parallel with 3DFX somewhere, but things have moved on a bit since it was all about fill-rate. In fact, it was precisely because it wasn't all about fill-rate that 3DFX died (hardware T&L on the geforce being the final nail in its coffin). I think you'll find that programmable pipelines have been pretty successful too.
More programmability is not just a gimmick, it's where real-time graphics is heading.
By view couldn't be any more different. ATi is losing the battle, and by a long way. Here's why.
Over the last 2 generations of cards, nvidia has made huge leaps in terms of features, particularly in terms of shaders. Pixel shaders can now be very long. They support conditional branching, so if statements and loops are possible without unrolling.
Now the geforce FX series, while great in terms of features, had well documented problems with 32-bit performance. However, these problems have been completely resolved in the 6 series. The 6 series of cards are superior to ATI's offerings in every sense, except possibly power consumption (and FYI, the GT doesn't require 2 slots).
OTOH, ATi has completely failed to innovate over the last 3 years. Every revision since the 9700 has been effectively just a speed increase. Their latest cards give basically nothing new in terms of features over the 9700 pro. In terms of capability, their latest cards are inferior to nvidia's FX cards.
As an owner of a 9700 and a hobbyist developer, I'm very familiar with the limitations. The shader length is highly restricted, conditional branching can't be done, so loops have to be unrolled. For this reason, even the latest ATI cards can't fully support the OpenGL Shading Language. What can be done on an FX or a Geforce 6 in one pass could take 10 or more passes on an X800. Many important features for shadow mapping are hopelessly missing, such as rendering to a depth texture, and hardware linear filtering.
So it looks to me like ATi are struggling to keep up in terms of performance, and they've put so much resources into just keeping the peformance acceptable that they've completely failed to innovate. And while gamers might not have noticed this before, they are starting to with Doom 3, and as developers push shader tech to its limits, they will really start to see the limitations of their cards. Hopefully they can fix the situation with their next generation of cards, but my next card will certainly be a nvidia.
I'm guessing whoever modded this down doesn't know what they're talking about.
But for what it's worth, the parent raises some interesting points, although I disagree that a bytecode/pseudo asm route is a better method.
In reality, the ASM-like bytecode used probably bears little resemblance to the machine code instruction sets actually used by the various cards. Forcing vendors to use such a low level instruction set allows very little room for optimisation, whereas allowing vendors to write their own compilers gives much better scope to improve the code at the machine code according to their hardware's instruction sets and limitations.
There may be a few teething problems at first (witness the arguments about nvidia's GLSL compliance on OpenGL.org), but I'm convinced its a better method overall.
TVs don't have a resolution in x. They're analogue horizontally. Although in effect, there is a minimum limit to the spatial acuity in x, this will vary greatly depending on the set.
Heh. You can laugh all you want about the price, but wait until some 19 megapixel camera appears that requires about a gig or so per photo,
19 megapixels is only 57 MB (assuming 8 bits per channel). You'd need ~323 megapixels per photo to fill a gig. Large numbers of photos seems like more likely use than copiously large photos somehow.
Or save yourself the effort. The article takes two pages to say that Avalon has proper transparency support for sub-elements and uses double buffering to prevent flicker/animation artifacts.
It's because they were there long before anyone else, and they've got most of the teething problems dealt with. Having tried Sainsbury's online service recently, they have a long way to go to catch up.
Online grocery shopping is already well-established in the UK. I've actually been using Tesco online for groceries for the past 3 years or so, and in my experience, yes, they do pick out good quality stuff. My only complaint is that they do occasionally screw up and deliver the wrong thing, but they've always given a full refund when that has happened. They're very keen for people to use their service, so they put a lot of effort into these issues.
Tesco charge 5 for deliveries (~$8), which works out pretty well IMO if you're doing a big shop. My only gripe is they don't put the stuff away for you (You need a maid/wife for that, and I hear they're expensive)
The state of AI currently is such that even simple understanding of english isn't possible. Just the term "Artificial Intelligence" is highly misleading. This program will likely do nothing more complex than a search for certain key-words and phrases. It won't have any understanding of the answers, and it will not be grading work fairly.
The people who okayed this system either have no understanding of the limitations of AI, or they don't care and want to save money. Either way, law suits will be coming.
What's next? SHOCKER: McBride blows his nose!! I know there's a weekly SCO quota on slashdot, but can we keep it to the vaguely significant stories please?
The only thing that surprised me was that the author used jpeg for graph images. So not only are the colours in the legend not clear, and the text bearly readable, the images are actually much bigger than they would be with PNG or gif.
Beginning OpenGL Game Programming is a pretty good book for learning OpenGL. It's slanted towards windows development, but most of the book is platform independent (you just won't be able to compile the examples without modifying them). It covers a lot of extensions too, and deals with core OpenGL up to version 1.5.
I should probably mention that I did the tech editing for this book, so make of that what you will.
Well obviously these cards are not aimed at the average joe consumer. They're targetted at enthusiasts and developers, and as always, the mid-range cards will follow in due course.
Seriously, if you had any interest in 3D graphics or game development then the launch of next-gen chips from ATi and nvidia should be of interest to you. If it's not, then FFS, don't read the article, and spare us your ill-informed insights on the subject.
The reviews on THG show that the 6800 beats the significantly on most tests. On the few occasions where the ATi wins, it's not by much. I find it difficult to understand how the original poster came to conclusion that the ATi is faster.
But more significantly, the ATi card does not make any meaningful improvements over the original 9700 in terms of shader technology. Conditional branching in fragment shaders, for example, really is a required feature these days.
ATi have really missed the boat on this. NVidia's previous generation of hardware had better shader tech (although it was slow). ATi are now 2 generations behind. A lot of people expected more than an overclocked R300. You have to wonder what they've been doing these last few years.
TBH I'd forgotten about the former method. It's been a while since I coded anything in Java. As for the latter, I explicitly mentioned that you could do this with Java's exception handling.
My point wasn't specifically about Java, it was just that goto isn't evil by definition.
Well I suppose this is supposed to be funny, but actually, the idea that goto is always evil is just wrong. There are a few occasions when it's justified:
Try finding a better way to break out of an inner loop. For example:
for (int i=0;i<n;i++)
for (int j=0;j<m;j++)
if (blah[i][j]==1) goto doublebreak;
doublebreak:...
To exit from a function, while cleaning up properly afterwards:
if (error) goto error; ...
if (anotherError) goto error; ...
error:
close(files);
delete stuff;
return -1;
Well, ok, that one is pretty much covered by exceptions in Java, but in C it's useful and justified.
Now look what you've gone and done
on
The War Of The Word
·
· Score: 4, Funny
You slashdotted microsoft. Another tiny web site bites the dust. I hope you feel proud of yourselves.
Yup, while PNG can do 24-bit with no problem, it's lossless compression, which doesn't perform particularly well with photos, and it just can't compete with jpeg in terms of file size.
Slashdot Mirror
I realise you're desperate to draw a parallel with 3DFX somewhere, but things have moved on a bit since it was all about fill-rate. In fact, it was precisely because it wasn't all about fill-rate that 3DFX died (hardware T&L on the geforce being the final nail in its coffin). I think you'll find that programmable pipelines have been pretty successful too.
More programmability is not just a gimmick, it's where real-time graphics is heading.
By view couldn't be any more different. ATi is losing the battle, and by a long way. Here's why.
Over the last 2 generations of cards, nvidia has made huge leaps in terms of features, particularly in terms of shaders. Pixel shaders can now be very long. They support conditional branching, so if statements and loops are possible without unrolling.
Now the geforce FX series, while great in terms of features, had well documented problems with 32-bit performance. However, these problems have been completely resolved in the 6 series. The 6 series of cards are superior to ATI's offerings in every sense, except possibly power consumption (and FYI, the GT doesn't require 2 slots).
OTOH, ATi has completely failed to innovate over the last 3 years. Every revision since the 9700 has been effectively just a speed increase. Their latest cards give basically nothing new in terms of features over the 9700 pro. In terms of capability, their latest cards are inferior to nvidia's FX cards.
As an owner of a 9700 and a hobbyist developer, I'm very familiar with the limitations. The shader length is highly restricted, conditional branching can't be done, so loops have to be unrolled. For this reason, even the latest ATI cards can't fully support the OpenGL Shading Language. What can be done on an FX or a Geforce 6 in one pass could take 10 or more passes on an X800. Many important features for shadow mapping are hopelessly missing, such as rendering to a depth texture, and hardware linear filtering.
So it looks to me like ATi are struggling to keep up in terms of performance, and they've put so much resources into just keeping the peformance acceptable that they've completely failed to innovate. And while gamers might not have noticed this before, they are starting to with Doom 3, and as developers push shader tech to its limits, they will really start to see the limitations of their cards. Hopefully they can fix the situation with their next generation of cards, but my next card will certainly be a nvidia.
Er, good point.
I'm guessing whoever modded this down doesn't know what they're talking about.
But for what it's worth, the parent raises some interesting points, although I disagree that a bytecode/pseudo asm route is a better method.
In reality, the ASM-like bytecode used probably bears little resemblance to the machine code instruction sets actually used by the various cards. Forcing vendors to use such a low level instruction set allows very little room for optimisation, whereas allowing vendors to write their own compilers gives much better scope to improve the code at the machine code according to their hardware's instruction sets and limitations.
There may be a few teething problems at first (witness the arguments about nvidia's GLSL compliance on OpenGL.org), but I'm convinced its a better method overall.
Actually, C# is a language developed by nvidia. It just happens to have exactly the same syntax as HLSL.
TVs don't have a resolution in x. They're analogue horizontally. Although in effect, there is a minimum limit to the spatial acuity in x, this will vary greatly depending on the set.
This argument sounds familiar somehow. The counter-argument went something like: if you don't like the terms, don't use it.
Heh. You can laugh all you want about the price, but wait until some 19 megapixel camera appears that requires about a gig or so per photo,
19 megapixels is only 57 MB (assuming 8 bits per channel). You'd need ~323 megapixels per photo to fill a gig. Large numbers of photos seems like more likely use than copiously large photos somehow.
Or save yourself the effort. The article takes two pages to say that Avalon has proper transparency support for sub-elements and uses double buffering to prevent flicker/animation artifacts.
It's because they were there long before anyone else, and they've got most of the teething problems dealt with. Having tried Sainsbury's online service recently, they have a long way to go to catch up.
Online grocery shopping is already well-established in the UK. I've actually been using Tesco online for groceries for the past 3 years or so, and in my experience, yes, they do pick out good quality stuff. My only complaint is that they do occasionally screw up and deliver the wrong thing, but they've always given a full refund when that has happened. They're very keen for people to use their service, so they put a lot of effort into these issues.
Tesco charge 5 for deliveries (~$8), which works out pretty well IMO if you're doing a big shop. My only gripe is they don't put the stuff away for you (You need a maid/wife for that, and I hear they're expensive)
The state of AI currently is such that even simple understanding of english isn't possible. Just the term "Artificial Intelligence" is highly misleading. This program will likely do nothing more complex than a search for certain key-words and phrases. It won't have any understanding of the answers, and it will not be grading work fairly.
The people who okayed this system either have no understanding of the limitations of AI, or they don't care and want to save money. Either way, law suits will be coming.
What's next? SHOCKER: McBride blows his nose!! I know there's a weekly SCO quota on slashdot, but can we keep it to the vaguely significant stories please?
Actually, I think you'll find Hell Freezing Over takes place slightly before DNF is released. But other than that, your post is 100% correct.
The only thing that surprised me was that the author used jpeg for graph images. So not only are the colours in the legend not clear, and the text bearly readable, the images are actually much bigger than they would be with PNG or gif.
>It is like saying that if I leave my back door unlocked at night, I am to blame if someone breaks in.
Technically it's not breaking in if the door is unlocked.
Beginning OpenGL Game Programming is a pretty good book for learning OpenGL. It's slanted towards windows development, but most of the book is platform independent (you just won't be able to compile the examples without modifying them). It covers a lot of extensions too, and deals with core OpenGL up to version 1.5.
I should probably mention that I did the tech editing for this book, so make of that what you will.
Well obviously these cards are not aimed at the average joe consumer. They're targetted at enthusiasts and developers, and as always, the mid-range cards will follow in due course.
Seriously, if you had any interest in 3D graphics or game development then the launch of next-gen chips from ATi and nvidia should be of interest to you. If it's not, then FFS, don't read the article, and spare us your ill-informed insights on the subject.
Last I checked, IE takes about a minute to start up, and it only shuts down when you shut down the computer.
The reviews on THG show that the 6800 beats the significantly on most tests. On the few occasions where the ATi wins, it's not by much. I find it difficult to understand how the original poster came to conclusion that the ATi is faster.
But more significantly, the ATi card does not make any meaningful improvements over the original 9700 in terms of shader technology. Conditional branching in fragment shaders, for example, really is a required feature these days.
ATi have really missed the boat on this. NVidia's previous generation of hardware had better shader tech (although it was slow). ATi are now 2 generations behind. A lot of people expected more than an overclocked R300. You have to wonder what they've been doing these last few years.
TBH I'd forgotten about the former method. It's been a while since I coded anything in Java. As for the latter, I explicitly mentioned that you could do this with Java's exception handling.
My point wasn't specifically about Java, it was just that goto isn't evil by definition.
Try finding a better way to break out of an inner loop. For example:
for (int i=0;i<n;i++) doublebreak:
To exit from a function, while cleaning up properly afterwards:
if (error) goto error;
if (anotherError) goto error;
error:
close(files);
delete stuff;
return -1;
Well, ok, that one is pretty much covered by exceptions in Java, but in C it's useful and justified.
You slashdotted microsoft. Another tiny web site bites the dust. I hope you feel proud of yourselves.
"what is Scribus really like?"
What about the more common question: "what is Scribus"? The uninformative summary doesn't help; neither does the slashdotted site.
Yup, while PNG can do 24-bit with no problem, it's lossless compression, which doesn't perform particularly well with photos, and it just can't compete with jpeg in terms of file size.