Re: [MSX] MAP update, 2005-05-12
Laurens Holst wrote: Hi all, Today Iâve updated the MAP again: * Fixed a few cut-off numbers in the Z80 instruction set overview (http://map.tni.nl/resources/z80instr.php) (no, a JP C, with NC doesnât take 1 T-state :)). I have now also added the undocumented Z80 and the R800 instructions to the instruction set overview. Iâve also added R800 timing information. However, I still need details on the turboRâs wait state insertion policy. The current content is just a wild guess. If anybody can point me to or send me some resources which explains the timing in detail, that would be appreciated. You can find the MAP at: http://map.tni.nl/ ~Grauw -- Ushiko-san! Kimi wa doushite, Ushiko-san!! ___ MSX mailing list (msx@stack.nl) Info page: http://lists.stack.nl/mailman/listinfo/msx
Re: [MSX] MAP update, 2005-05-12
Patriek Lesparre wrote: * Replaced the Div16 16-bits division routine with a faster one by Flyguille. The speed gain seems quite humble, 3 T-states, but this one can be unrolled to gain quite some speed. Another 4 t-states can be saved by replacing XOR A ; 5 t-states LD L,A ; 5 t-states LD H,A ; 5 t-states with LD HL,0 ; 11 t-states The first sequence of instructions are only faster if both A and HL are supposed to be 0, which is not the case in this routine. Thanks! Though itâs only 3 states, as Iâm also counting the additional M1 wait state. If you donât mind, Iâll just keep the credit for you on this list, otherwise the documentâll become a maze of thanks this thanks that ^_^. Even though the tweak effectively doubles the gain in speed ;) ~Grauw -- Ushiko-san! Kimi wa doushite, Ushiko-san!! ___ MSX mailing list (msx@stack.nl) Info page: http://lists.stack.nl/mailman/listinfo/msx
Re: [MSX] MAP update, 2005-05-12
Grauw wrote: * Replaced the Div16 16-bits division routine with a faster one by Flyguille. The speed gain seems quite humble, 3 T-states, but this one can be unrolled to gain quite some speed. Another 4 t-states can be saved by replacing XOR A ; 5 t-states LD L,A ; 5 t-states LD H,A ; 5 t-states with LD HL,0 ; 11 t-states The first sequence of instructions are only faster if both A and HL are supposed to be 0, which is not the case in this routine. Greetz, Patriek ___ MSX mailing list (msx@stack.nl) Info page: http://lists.stack.nl/mailman/listinfo/msx
[MSX] MAP update, 2005-05-12
Hi all, Today Iâve updated the MAP again: * Fixed a few cut-off numbers in the Z80 instruction set overview (http://map.tni.nl/resources/z80instr.php) (no, a JP C, with NC doesnât take 1 T-state :)). * Replaced the Div16 16-bits division routine with a faster one by Flyguille. The speed gain seems quite humble, 3 T-states, but this one can be unrolled to gain quite some speed. * In addition to that, about two weeks ago I also optimized the Mult32 routine, thanks to Colin Douglas Howell. 96 T-states faster! You can find the MAP at: http://map.tni.nl/ ~Grauw -- Ushiko-san! Kimi wa doushite, Ushiko-san!! ___ MSX mailing list (msx@stack.nl) Info page: http://lists.stack.nl/mailman/listinfo/msx