using regular words alone, it should be difficult to deal with, because the content matched is equivalent to being consumed and will not be matched twice.
if it is js, you can write

var str="SegmentFault";
var regex=/[\u4e00-\u9fa5]{2}/g;
var matchStr=null;
var result=[];
while((matchStr=regex.exec(str))!=null){
    result.push(matchStr[0]);
    regex.lastIndex--;
}

Recursive

var string = "SegmentFault";
var reg = /[\w]{2}/;
function seg(str) {
  if (str.length) {
    console.log(str.match(reg)[0]);
    str = str.substring(1, str.lenfth);
    seg(str);
  }
}
seg(string);

different languages have different processing methods. In JavaScript, global matching regularities have a lastIndex attribute that adjusts the starting position of the next match.

const str = "SegmentFault"
const matcher = /[\u4e00-\u9fa5]{2}/g
const result = []
while (true) {
  const m = matcher.exec(str)
  if (!m) { break }
  result.push(m[0])
  matcher.lastIndex = matcher.lastIndex - m[0].length + 1
}
console.log(result)