On Wed, 9 Nov 1994 11:24:14 +0000 (GMT), Simon Cooke said: > To tell the truth, as far as my tests indicate, the problem's with the speed > of the SAM...
Then obviously a DMA is needed. > To read a sector at the moment, you have to do the equivalent of: [setup deleted] > main.loop: > in a,(status) > bit 3,a > jr nz,read.byte > rla > ret nc > jr main.loop Do you _really_ have to do this much checking for each pair of bytes? If that's the case then there's something wrong with the hardware. You should just be able to tell the disk to read a sector and then do 512 INs in a row to read it. Even the ZX Microdrive did that (well, as far as I remember anyway). If an error occurs in the middle, the hardware should compensate by giving the program dummy values until al lthe bytes have been read, at which point it can check for an error indication. By the way, you can save T-states in that loop with: main.loop: in a,(status) rla ret nc and 16 jr z,main.loop (this loses 3 T-states if data is immediately available but saves 8 on each time round the loop). > read.byte: > in a,(data) > ld (hl),a > inc hl > in a,(data) > ld (hl),a > inc hl > jr main.loop What's wrong with "ini"? It takes 16 cycles to the 24 of "in a,(n);ld(hl),a; inc hl". Also, "jp" is two cycles faster than "jr". imc
