The problem I face is reading from a magnetic card reader, that sends me bit-to-bit info, and as i've posted, i can read one bit while the data_ready line goes low (50 uSecs).

Due to the mag. enconding scheme, i have to use some rules:
1) every 2 bit you read, xor them into a single bit. So after you get 10 bit, you have 5 "real" bits

2) every 5 bits (already xor'ed in step 1) check parity (5 bits-> one nibble and one parity bit). The nibble is a BCD number.

3) After you read the card, that most of the cases is about 40 numbers:
(40 * (5 bits)) * 2 bits) = 400 bits from the reader, you have tho check is it is a good card and use a triac or raise a LED to show 'wrong card'.

For that i was considering a standard 8051 with 11.0592 for clock, but as is almost sure i'll have to put more fancy stuff like LCD, 4x4 keyboard, RS485, etc; i guess if Dallas DS80C320 is a better option right from the start.
For coding, i'm times better in C, than asm, as the only asm i know is x86. To get the better from both i as considering the main routine in C and the ISR in asm.

My concern about time was that in standart 8051, 1 opcode = 1 uSec in the better case. This plus context switch to the interrupt routine can make it hard to get the bit in the 50 uSecs, and can't delay too much in it or i'll miss the next bit 250 uSecs later

Thanks for the prompt answer.
Bruno Jessen