It kind of makes sense that they do 2x1 dithering for red and blue. That way they keep the number of colors under 2^16 and can still use the 5+6+5 representation, using 16 bits per pixel. I guess the video RAM still uses 65536 colors, but some of them map to the same pattern.
The thing is, that "color combinations" are not good enough. In top of the 11% number of colots less, colors are not uniform, but simulated with patterns of 2x2 pixels, which is probably very noticeable.
There are inexpensive A/D chips that can be easily programmed from a PC. They can store the data they meassure in a small local RAM and then transfer it to the PC.
I know that some guys use these A/Ds for things like "listening" to the stars using a radio with a parabolic anthena. Probably some E.T. searchers can help you here.
I personally like the idea of writting small or medium sized programs that solve pieces of the problem, much in the Unix spirit. Huge monolithic programs are difficult to read and mantain. If you make several programs with well defined behaviour, you are giving a non-programmer (say, a sys admin) the possibility to make new things using those tools.
In my company we all know C++ and Perl, and we use a combination of both for a lot of things. I prefer not to use shell-scripts, because Perl is perfectly fine for that.
I agree with most of your comment, but I really doubt that "once the infracstructure settles down" anything is going to improve. Computers are engineering as much as they are science. I think you cannot forget the history of mathematics when you try to predict the future of computers.
We have been making maths for about 2600 years now, and we don't have less open questions as we had before. The number of open problems is increasing, because every question you "kill" creates four more (anyone knows the hydra myth?).
As I see it, the situation with computers is going the same way. In the early days of computing, nobody had to face standard compliance problems, race conditions in multi-threading programs, security issues in open networks or virtually any other problem that makes software suck today. And as we evolve into more complex projects, we will have to face more difficult problems.
It is more difficult to be a good mathematician today than it was being one two hundred years ago, and it is easier to be a good computer engineer/scientist today than it will be in fourty years.
Slashdot Mirror
It kind of makes sense that they do 2x1 dithering for red and blue. That way they keep the number of colors under 2^16 and can still use the 5+6+5 representation, using 16 bits per pixel. I guess the video RAM still uses 65536 colors, but some of them map to the same pattern.
The thing is, that "color combinations" are not good enough. In top of the 11% number of colots less, colors are not uniform, but simulated with patterns of 2x2 pixels, which is probably very noticeable.
Can someone, please, moderate this post higher? Right now it's Score:0, while it's the exact right answer to the two cases proposed.
"fondly"?s "?
"yore"?
"prey"?
"folly"?
"imperviou
My goodness! I guess Google will not classify this page as English!
Take a look here. I specially like the last paragraph about "reimplementing" the bug.
If they did the first reading when the machine was running and the second one after stopping, the increase in voltage is just the expected behaviour.
There is an internal resistance in every power supply, and Ohm's Law makes the voltage reduce if you are consuming some current.
If they really had to work 17 years to make this trick, that's pretty sad.
There are inexpensive A/D chips that can be easily programmed from a PC. They can store the data they meassure in a small local RAM and then transfer it to the PC.
I know that some guys use these A/Ds for things like "listening" to the stars using a radio with a parabolic anthena. Probably some E.T. searchers can help you here.
I personally like the idea of writting small or medium sized programs that solve pieces of the problem, much in the Unix spirit. Huge monolithic programs are difficult to read and mantain. If you make several programs with well defined behaviour, you are giving a non-programmer (say, a sys admin) the possibility to make new things using those tools.
In my company we all know C++ and Perl, and we use a combination of both for a lot of things. I prefer not to use shell-scripts, because Perl is perfectly fine for that.
I agree with most of your comment, but I really doubt that "once the infracstructure settles down" anything is going to improve. Computers are engineering as much as they are science. I think you cannot forget the history of mathematics when you try to predict the future of computers.
We have been making maths for about 2600 years now, and we don't have less open questions as we had before. The number of open problems is increasing, because every question you "kill" creates four more (anyone knows the hydra myth?).
As I see it, the situation with computers is going the same way. In the early days of computing, nobody had to face standard compliance problems, race conditions in multi-threading programs, security issues in open networks or virtually any other problem that makes software suck today. And as we evolve into more complex projects, we will have to face more difficult problems.
It is more difficult to be a good mathematician today than it was being one two hundred years ago, and it is easier to be a good computer engineer/scientist today than it will be in fourty years.