finally solved...
sort -R was way too slow for such file and the second solution didn't work at all :/

but i found the rl command which does the same thing faster...
so the complete code:

Code:

FILE=$1
COUNT=$2
rl -c $COUNT $FILE | awk -v FS='>' '{
        $0=tolower($3);
        gsub(/[^a-z]/, " ");
        split($0, A, " ");
        for(key in A)   count[A[key]]++;
}
END {
        for(key in count) {
        words += count[key];
        uniq++;
    }
        printf("Total words: %d\tUnique words:%d\n", words, uniq);
}'

thank you for your help!

Do you know how to pass the variables inside awk (words,uniq) outside its scope? so i can use them in the rest of the script.