#define NUM_CHARS_LINE 20 //number of characters per lcd line

char lcd_buff[NUM_CHAR_LINE+1]; //+1 to be sure!!

void timer_tick_1s(void) //we get called every second
{
static char shift = 0;
lcd_output(shift);
shift++;
if (shift > NUM_CHARS_LINE) shift = 0;
}



void lcd_output(char offset)
{
char off;
char loop;

lcd_cursor(0); //set cursor to top left
off=offset; //set starting position in the line buffer
for(loop=0;loop<NUM_CHARS_LINE;loop++) //loop for 20 characters (1 line)
{
if (off >=NUM_CHARS_LINE) off = 0; //do the wrap around
lcd_out(lcd_buff[off]); //write char to the lcd
off++;
}
}

Whilst not super optimised, hopefully it illustrates the point. I'm sure if you want it in assembler, you can convert the code easily!