MIPS: Converting hexadecimal number to 32-bit binary

Question

I'm a beginner to MIPS. I'm trying to figure out converting user's inputted hexadecimal values into 32-bit binary number, and store it to an array. Here's what I think: For example, the number is (3ABCD123), then I read this value and split every character and put it in a string array. Then, for every value in the array, convert it to a 4-bit binary number and store it to another array. But I am stuck on splitting the user's inputted string into every character and store it in an array. Please give any suggestions or help, thanks!

Answer 1

MARS / SPIM have a string-input system call. A string is a char array; you don't have to manually "split" anything until you're looping over the input buffer with lbu. (Unfortunately the system-call doesn't return the length, leaving you to scan the buffer for a terminating 0 byte, or just write your loop to stop on the first non-hexdigit character instead of using a counter.)

Then you'll want to validate the input to check it's a hex digit, and calculate a 4-bit integer.

A hex digit is either a decimal digit '0'..'9' or 'a'..'f', maybe after folding upper-case to lower-case, so it will take some branching and subtraction to do either
digit = c - '0' or digit = lcase(c) - 'a' as appropriate.

You don't want to store separate 4-bit values in any array; shift + OR them into a 32-bit register to build up the final 32-bit integer. Only once you have that should you store the whole word into an array and read another string.

(MIPS is byte-addressable so you could store in chunks of 2 nibbles, but it's not sub-byte addressable so you can't sb 8 separate nibbles into 4 bytes and lw a complete 32-bit word.)

Shift/or into a register like total = (total<<4) | digit works even if the string of hex digits is fewer than 8 digits, assuming you zero the total to start with. The first (most-significant) ends up shifted to wherever its place value actually is, i.e. by the number of digits following it. Since you're shifting in a register, not messing around with memory, endianness doesn't enter into it. And numbers as strings of digits are always most-significant first, because that's the convention in most human writing systems that use Arabic numerals. (like 1234).