An OpenGL driver is a full OpenGL implementation. A lot of the optimizations that NVIDIA does in the high-levels of their drivers could easily be used by a competitor. Since crappy drivers is the main thing holding ATI back, it would be very stupid of NVIDIA to help them out in that catagory.
Visual C++ is a pile of stinking shit. A C++ compiler cannot call itself a C++ compiler if it only has half-ass support for a nearly 5 year old standard! VC++ 7 is getting better, and the 7.1 beta is supposed to be quite good, but GCC was there a good while before, and GCC's C++ support is *extremely* robust. Visual Studio has a good code editor (though, the auto-indent tries to force you to layout your code in weird ways, like no indent after public: declaration) and a very good class browser, but other than that, it's not great at all. GDB is perfectly comparable to Visual C++'s debugger, and the remote debugging feature has much less overhead on the target (good for developing embedded code) and an open protocol. Visual C++'s make system uses binary files (ugh) and can be quite delicate and easy to fool (at least in VC++ 6.x). Visual C++ is just peachy for writing Windows/MFC/RAD code, but for anything else, the numerous UNIX tools are far better.
PS> Don't even get me started on Visual Source Safe...
That's a bit unfair of a characterization. It's mainly that Windows developer support is all there in one place. If you're doing high-level stuff, it might not be as deep as UNIX developer support can be (though, UNIX documentation isn't that great, at least in the OSS world) but it's quick to get to and well organized. With UNIX stuff, you have to go to a bunch of seperate places to find documentation.
I hate Microsoft much as the next guy, but the headline is *way* overwrought. If you actually read the linked article, it's just an honest pro/con comparison. They mention certain advantages of UNIX (text configuration, small size) and certain advantages of Windows (better internationalization, more developer support, better throughput). Entirely realistic and a perfectly fine rationale document. There are some bits I disagree with (eg. Visual Studio being better than the UNIX development tools) but overall, this is just a document written by an engineer weighing the various issues involved in switching from UNIX to Windows.
You might try to get a refund when you don't like a product, but most people don't. People put up with substandard quality all the time. For example, the cable company in our area has some seriously atrocious service. There are alternatives, thanks to deregulation, but do you think that consumers go to all the trouble needed to subscribe to one of the small alternative providers? No. They stick with the status quo.
I never said people hate McDonald's. I said that people would hate having a McDonald's commercial in their game. Very different beast. As for your "whiney bullshit" stuff, your idea is just a byproduct of an overly simplistic thought process.
1) People, when given the appropriate environment, do have fairly high standards for service quality. 2) People are intimidated by large corporations, to the point where that intimidation overrides their desire for quality. 3) There is a minority out there not indimidated by corporations, who are fully willing to boycott products that they don't like.
In the context of the above three statements, the "whinney bullshit" looks very different. Instead of complaining because the majority wants something they do not, the minority is complaining about something nobody wants, but something that the majority is too intimidated/lazy/etc to reject outright. Given the lassitude of the majority, there is nothing else the minority can do but to speak out against the product, hoping to find another way to convince the company to change its practices. Further, your last sentence becomes very different. It's not just the scared/lazy people who deserve it that get stuck with bad products, it's the vocal minority as well.
The sad fact is that the majority probably does hate the McDonald reference, but puts up with it anyway. Do you realize the huge, enormous number of Windows users that detest all the crap sitting in their system tray popping up at them at random times? Do you think they keep quiet because they like those interruptions, or because they're so programmed into believing that computer companies are monolithic entities that you cannot appeal to? Notice the difference in people's behavior when dealing with a small, local shop and a large corporation. With the small shop, the feel like they can change things. If the service isn't up to standards, they'll complain quite loudly. With a large corporation, they get this feeling that they're irrelevent, and no amount of customer feedback will do any good. Thus, they'll bend over and take everything the company wants to give them.
PS> A personal "fuck you" goes out to RealPlayer for starting all this system tray non-sense. I remember back when I got Windows 95, there was not a single program that auto-started and clogged up my system tray with useless junk. Then, RealPlayer (4.x or 5.x I think) came and started it, and now we've gotten to the point where AIM will autostart with several glaring Windows and adds on your desktop.
Not everything is a right, and not everything has to be spelled out in the constitution. The first thing you don't realize is that it's not fully under your control. The majority of users out there don't care (because they're stupid and don't know any better, not because they don't mind the advertising) so you not buying the game isn't going to accomplish anything. Given that, it is entirely within your rights as a consumer of a product to complain about certain aspects of that product in order to get them changed. If developers can fill their products with garbage, I can certainly bitch at them for it! There is your third thing!
Actually, Whoppers are perhaps the worst food item on the face of the planet. Something like 40-50 grams of fat per serving! That's more than half your daily allowence!
The Kyro II isn't a more efficient architecture, it uses a totally different rendering method than most cards. Comparing it's efficiency to a normal card's is like comparing PovRay to Renderman. For the same rendering task, they do different things, and their results are different, so efficiency isn't directly comparable. It's not like comparing a G4 to a P4, where you basically execute the same logical operations and see which one executes each operation in the least number of cycles.
That said, there is a reason the Kyro approach isn't used as often. It has some serious issues with transparency and alpha-blending. Besides, NVIDIA and ATI both include certain Kyro-esque features, like occlusion culling (removing objects that will be hidden by objects drawn later) that provide most of the benifets of the Kyro in high-overdraw situations within the context of a standard 3D pipeline.
As for Intel, I think what they're doing with the P4 is great. All I care about is final performance and how much it costs. High-clockspeed vs IPC is a *design decision* driven by the realization that wide-slow architectures just don't cut it for the kind of code most people run.
Every time I've seen a problem with the NVIDIA drivers, it's been from somebody using dual-head. I think the TwinView code has some glitches in it, and apparently (from what I gather from others I've talked to) a resource leak under certain circumstances (acquiring/releasing the screen). TwinView works fine for me over here, but I only use it for watching DVDs on my TV, so I don't stress it was much.
Ebuild versioning is supposed to reflect changes that will affect an existing installation. If a modification to an ebuild is made just to correct a tiny problem that 0.1% of users encounter because of a particular software configuration, it's a bad idea to change the revision number of the ebuild, since it will force a recompile of that package the next time the user runs 'emerge world'. Sure you could add a patch level to each ebuild, so you had ebuild with a software version, ebuild revision, and patch-level, then allow emerge to skip patch-levels of the same revision when considering a package for update, but that would add another level of complication to an already (necessarily) complex versioning scheme, and wouldn't gain you anything significant.
What I want to know are the algorithms! For example, human beings have nearly perfect (excepting edge cases like optical illusions) object recognition. Even if you don't know what the hell something is, we can tell it is a seperate object, independent of other objects in the scene. Also, the occipital lobe does some extremely funky processing in breaking down what is essentially a pixel grid (the receptors in the eyes) into lines curves and whatnot. That, IMO, is far more interesting than the raw processing capabilities of the brain.
I don't know, I think Carmack said something about the floating point format giving you more dynamic range then you get with integer. Ie, the largest value for a 32-bit float is a whole lot larger than the largest value for a 32-bit integer.
Um, exactly how much experience do you have with Gentoo? I've been using it as my only OS for several months now, and ports has served me not only in installing ports and whatnot, but allowing me to edit my ebuilds to test alpha-pre-rc software like I have a tendency to. Specifically, what don't you like about Portage? Do you realize there are extension tools (in gentoolkit) that add package management on top of emerge?
I've given up trying to figure out AMD's naming schemes. I meant the Hammer-based Athlon processors. Besides, I don't do gaming. If a $1000 machine can provide the same performance as dual Opterons for the stuff I do, gimme!
It depends. The article mentions that one of the main uses of the increased bandwidth is to enable high performance anti-aliasing. When you do multisample antialiasing, you jitter the image slightly to blur polygon edges. On the interior of polygons, jittering will often fail to change the color of the pixel at all, which will allow 4 pixels (for an ideal 4x AA case) to be compressed to one. It is mainly at the edges of polygons, where jittering mixes the polygon and background colors that the resultant pixels differ and thus can't be compressed. Even without antialiasing, I'd bet the compression buys enough to make up for the 3 GB/sec difference in memory bandwidth.
The card will support Linux (and FreeBSD now!) just like every other NVIDIA card in the GeForce line. And it will support Linux with performance equal to it's performance in Windows. So it just means the fastest 3D hardware with Linux support just got faster. This is good news for Linux users, because it one-ups ATI's soluation, which *doesn't* have Linux support.
Slashdot Mirror
An OpenGL driver is a full OpenGL implementation. A lot of the optimizations that NVIDIA does in the high-levels of their drivers could easily be used by a competitor. Since crappy drivers is the main thing holding ATI back, it would be very stupid of NVIDIA to help them out in that catagory.
"Informative?" Yeesh. There should be a new mod rating:
"+1 Read Article"
Visual C++ is a pile of stinking shit. A C++ compiler cannot call itself a C++ compiler if it only has half-ass support for a nearly 5 year old standard! VC++ 7 is getting better, and the 7.1 beta is supposed to be quite good, but GCC was there a good while before, and GCC's C++ support is *extremely* robust. Visual Studio has a good code editor (though, the auto-indent tries to force you to layout your code in weird ways, like no indent after public: declaration) and a very good class browser, but other than that, it's not great at all. GDB is perfectly comparable to Visual C++'s debugger, and the remote debugging feature has much less overhead on the target (good for developing embedded code) and an open protocol. Visual C++'s make system uses binary files (ugh) and can be quite delicate and easy to fool (at least in VC++ 6.x). Visual C++ is just peachy for writing Windows/MFC/RAD code, but for anything else, the numerous UNIX tools are far better.
PS> Don't even get me started on Visual Source Safe...
That's a bit unfair of a characterization. It's mainly that Windows developer support is all there in one place. If you're doing high-level stuff, it might not be as deep as UNIX developer support can be (though, UNIX documentation isn't that great, at least in the OSS world) but it's quick to get to and well organized. With UNIX stuff, you have to go to a bunch of seperate places to find documentation.
Or you can buy the much cheaper (and identical :) Sager or Prostar machine. Don't use it on your lap though, if you ever plan to have kids.
I'm using it on a laptop right now and it's worked flawlessly so far. Maybe external flat-panel via DVI?
I hate Microsoft much as the next guy, but the headline is *way* overwrought. If you actually read the linked article, it's just an honest pro/con comparison. They mention certain advantages of UNIX (text configuration, small size) and certain advantages of Windows (better internationalization, more developer support, better throughput). Entirely realistic and a perfectly fine rationale document. There are some bits I disagree with (eg. Visual Studio being better than the UNIX development tools) but overall, this is just a document written by an engineer weighing the various issues involved in switching from UNIX to Windows.
You might try to get a refund when you don't like a product, but most people don't. People put up with substandard quality all the time. For example, the cable company in our area has some seriously atrocious service. There are alternatives, thanks to deregulation, but do you think that consumers go to all the trouble needed to subscribe to one of the small alternative providers? No. They stick with the status quo.
I never said people hate McDonald's. I said that people would hate having a McDonald's commercial in their game. Very different beast. As for your "whiney bullshit" stuff, your idea is just a byproduct of an overly simplistic thought process.
1) People, when given the appropriate environment, do have fairly high standards for service quality.
2) People are intimidated by large corporations, to the point where that intimidation overrides their desire for quality.
3) There is a minority out there not indimidated by corporations, who are fully willing to boycott products that they don't like.
In the context of the above three statements, the "whinney bullshit" looks very different. Instead of complaining because the majority wants something they do not, the minority is complaining about something nobody wants, but something that the majority is too intimidated/lazy/etc to reject outright. Given the lassitude of the majority, there is nothing else the minority can do but to speak out against the product, hoping to find another way to convince the company to change its practices. Further, your last sentence becomes very different. It's not just the scared/lazy people who deserve it that get stuck with bad products, it's the vocal minority as well.
Dude, this is perhaps the most insightful comment I've read all week! I wish I had mod points right now.
Not if the CPU won't let you make those memory accesses.
The sad fact is that the majority probably does hate the McDonald reference, but puts up with it anyway. Do you realize the huge, enormous number of Windows users that detest all the crap sitting in their system tray popping up at them at random times? Do you think they keep quiet because they like those interruptions, or because they're so programmed into believing that computer companies are monolithic entities that you cannot appeal to? Notice the difference in people's behavior when dealing with a small, local shop and a large corporation. With the small shop, the feel like they can change things. If the service isn't up to standards, they'll complain quite loudly. With a large corporation, they get this feeling that they're irrelevent, and no amount of customer feedback will do any good. Thus, they'll bend over and take everything the company wants to give them.
PS> A personal "fuck you" goes out to RealPlayer for starting all this system tray non-sense. I remember back when I got Windows 95, there was not a single program that auto-started and clogged up my system tray with useless junk. Then, RealPlayer (4.x or 5.x I think) came and started it, and now we've gotten to the point where AIM will autostart with several glaring Windows and adds on your desktop.
Not everything is a right, and not everything has to be spelled out in the constitution. The first thing you don't realize is that it's not fully under your control. The majority of users out there don't care (because they're stupid and don't know any better, not because they don't mind the advertising) so you not buying the game isn't going to accomplish anything. Given that, it is entirely within your rights as a consumer of a product to complain about certain aspects of that product in order to get them changed. If developers can fill their products with garbage, I can certainly bitch at them for it! There is your third thing!
Actually, Whoppers are perhaps the worst food item on the face of the planet. Something like 40-50 grams of fat per serving! That's more than half your daily allowence!
The VM stabilized in the middle of 2.4. Right now, we're in the middle of 2.5. That was a while ago, and the new VM is quite solid.
The Kyro II isn't a more efficient architecture, it uses a totally different rendering method than most cards. Comparing it's efficiency to a normal card's is like comparing PovRay to Renderman. For the same rendering task, they do different things, and their results are different, so efficiency isn't directly comparable. It's not like comparing a G4 to a P4, where you basically execute the same logical operations and see which one executes each operation in the least number of cycles.
That said, there is a reason the Kyro approach isn't used as often. It has some serious issues with transparency and alpha-blending. Besides, NVIDIA and ATI both include certain Kyro-esque features, like occlusion culling (removing objects that will be hidden by objects drawn later) that provide most of the benifets of the Kyro in high-overdraw situations within the context of a standard 3D pipeline.
As for Intel, I think what they're doing with the P4 is great. All I care about is final performance and how much it costs. High-clockspeed vs IPC is a *design decision* driven by the realization that wide-slow architectures just don't cut it for the kind of code most people run.
Every time I've seen a problem with the NVIDIA drivers, it's been from somebody using dual-head. I think the TwinView code has some glitches in it, and apparently (from what I gather from others I've talked to) a resource leak under certain circumstances (acquiring/releasing the screen). TwinView works fine for me over here, but I only use it for watching DVDs on my TV, so I don't stress it was much.
Ebuild versioning is supposed to reflect changes that will affect an existing installation. If a modification to an ebuild is made just to correct a tiny problem that 0.1% of users encounter because of a particular software configuration, it's a bad idea to change the revision number of the ebuild, since it will force a recompile of that package the next time the user runs 'emerge world'. Sure you could add a patch level to each ebuild, so you had ebuild with a software version, ebuild revision, and patch-level, then allow emerge to skip patch-levels of the same revision when considering a package for update, but that would add another level of complication to an already (necessarily) complex versioning scheme, and wouldn't gain you anything significant.
What I want to know are the algorithms! For example, human beings have nearly perfect (excepting edge cases like optical illusions) object recognition. Even if you don't know what the hell something is, we can tell it is a seperate object, independent of other objects in the scene. Also, the occipital lobe does some extremely funky processing in breaking down what is essentially a pixel grid (the receptors in the eyes) into lines curves and whatnot. That, IMO, is far more interesting than the raw processing capabilities of the brain.
I don't know, I think Carmack said something about the floating point format giving you more dynamic range then you get with integer. Ie, the largest value for a 32-bit float is a whole lot larger than the largest value for a 32-bit integer.
Um, exactly how much experience do you have with Gentoo? I've been using it as my only OS for several months now, and ports has served me not only in installing ports and whatnot, but allowing me to edit my ebuilds to test alpha-pre-rc software like I have a tendency to. Specifically, what don't you like about Portage? Do you realize there are extension tools (in gentoolkit) that add package management on top of emerge?
I've given up trying to figure out AMD's naming schemes. I meant the Hammer-based Athlon processors. Besides, I don't do gaming. If a $1000 machine can provide the same performance as dual Opterons for the stuff I do, gimme!
It depends. The article mentions that one of the main uses of the increased bandwidth is to enable high performance anti-aliasing. When you do multisample antialiasing, you jitter the image slightly to blur polygon edges. On the interior of polygons, jittering will often fail to change the color of the pixel at all, which will allow 4 pixels (for an ideal 4x AA case) to be compressed to one. It is mainly at the edges of polygons, where jittering mixes the polygon and background colors that the resultant pixels differ and thus can't be compressed. Even without antialiasing, I'd bet the compression buys enough to make up for the 3 GB/sec difference in memory bandwidth.
2/3
The card will support Linux (and FreeBSD now!) just like every other NVIDIA card in the GeForce line. And it will support Linux with performance equal to it's performance in Windows. So it just means the fastest 3D hardware with Linux support just got faster. This is good news for Linux users, because it one-ups ATI's soluation, which *doesn't* have Linux support.