The program I made starts showing a small area around the origin: from (-2.5,-1.5) to (+1.5,+1.5)... this gives the first image. Clicking on any point of the touchscreen, it will start recalculating a new image, taking the touched point as a new center of the image and zooming in 2x. After 5 clicks you will see for instance what's shown on the second image. Not bad, right?
Then you can press any of the two shoulder buttons to zoom out, or click on the start button to reset the program to the initial setting.
Anyway the Nintendo DS has some limitations. Of course it doesn't feature a superscalar quadcore 2GHz+ processor. It only has a 67 MHz ARM946E, which also has no floating point unit at all, so each operation on floating point variables doesn't turn into a single (co)processor opcode, but into a series of integer operation. So, to keep the image generation time acceptable, I had to limit the number of maximum iterations of the aforementioned algorithm (actually to a very small value: 32). I also had to opt for single precision floating point variables, the fastest choice available.
Then you can press any of the two shoulder buttons to zoom out, or click on the start button to reset the program to the initial setting.
Anyway the Nintendo DS has some limitations. Of course it doesn't feature a superscalar quadcore 2GHz+ processor. It only has a 67 MHz ARM946E, which also has no floating point unit at all, so each operation on floating point variables doesn't turn into a single (co)processor opcode, but into a series of integer operation. So, to keep the image generation time acceptable, I had to limit the number of maximum iterations of the aforementioned algorithm (actually to a very small value: 32). I also had to opt for single precision floating point variables, the fastest choice available.
