------------------------------------------ MAIL ------------------------------------------ Here is a letter from Employer/Alliance Hi All! In PLUTONIUM#O7 I (Employer/Alliance) I found this line: 12.75 clock cycles per byte!!! It can't be faster if someone comes up with it Let me know the fastest way, I'll be very happy. I hasten to inform you that the most fast (!?) way to transfer data to #Z8O is carried out in 1O.5 cycles per byte!!! By taking a calculator you can by pressing a simple combination keys, discover that if this is so, then there is a real opportunity throw away the entire (FULL) screen in one step jerking! I’ll make a reservation right away: I mean- There is a screen without colors (#18OO), but on there is enough music and still remains. I remember back in ZX-POWER there was proof on the unreality of displaying the screen behind the ope int and it looks like this: LD HL,NN: 1O PUSH HL: 11 I feel doubt arise... Well, now in more detail: In fact, this method is practically not has applications. The point is that you can store a sprite not separately, but in the procedure itself! those. you need to prepare the procedure in advance occupying, if I'm not mistaken, 12288 bytes. Then take the picture and 2 bytes each put in procedure. We must not forget that PUSH Decreases register Pna 2, and this means the picture will be printed back to front. As you noticed, the 12kb procedure is not causes optimism, but still it is fast three... although if you donate 2Chkb, then you can do fullscreen scrolling! By the way ! In the interior of this issue, I had happiness to watch such scrolling, here only it was cut in 3 places and reminds cash dragging windows in WINDOWS'9x! :-) If Comrade LYNX will say that when op int It's hard to read anything while scrolling It’s written there, then I’ll still believe it. And if he will say that he did not know about the existence2nd Speccy's screen, then I might think... And now a little about optimization: It is absolutely not necessary to type pro- whole 12kb procedure. XAS has !assm n and !cont directives which means: !assm #18OO/#O2 LD HL,NN PUSH HL !cont the specified program will be built in memory n times. But this is not optimization. We need to write a program that will build a procedure and substitute the bytes from pictures in it. Then your program will take no more than 2OO bytes in memory. This whole thing applies to programs as well. HIGHLANDER (PLUTONIUM#O7) Judging by the code of the viewer HIGHLANDER is not bad he's a good coder, so it won't be difficult for him to yes, implement such an algorithm, but what if what, I can provide you with the source (if you still don’t believe it) By the way, HIGHLANDER wrote that it is necessary to disable interruptions before working with stack, but for example I have never done this I did it and everything worked, for now. Just if you are absolutely sure that your the program goes into interruption, then nothingnothing terrible will happen. With this I will say goodbye, bye! p.s. Please do not be offended by me because... I slightly ran into Comrade. LYNX. (I just saw his scroll somewhere, and it seems it was in the Inforcom book) r.p.s. Speccy Rulez Cheva! r.r.r.s. If you want, I can write it down for you article about fast 3D graphics (with examples on asm!) Ed>we want! ------------------------------------------ End of mail... ------------------------------------------
Share your thoughts about the article