Measuring Execution Time of a Single Command

I am trying to run an elementary benchmark of a MySQL database query using the Time::HighRes and DBD::mysql modules.  The idea is to query four tables of different sizes and measure the time to query each with Perl.  However, I suspect that my results are invalid because the first table always takes the longest to query, no matter how big it is, and the general trend is that subsequent queries take less and less time.  I suspect that Perl is trying to optimize execution and does commands out of order, which would be problematic for the method I'm using to measure elapsed time.  My question is this: How can I ensure that I am really measuring the time of the MySQL select statement?  

Below is the code snippet that does the 4 queries.  $ARGV[$i] is the size of the table being queried.  I am using Time::HighRes to take the measurements.  
-----------------------------------------------------------------------------------------------------  
$i = 1;
while($i <= 4) {
                # Find the middle record in the table
                $query = $ARGV[$i]/2;
                $sth = $dbh->prepare("SELECT * FROM `benchmark".$i."` WHERE `int`=".$query.";");
                # Set Timestamp
                my $start_time = [gettimeofday];
                exec($sth->execute);
                # Measure & Record Query Execution Time
                $query_time = tv_interval($start_time);
                $sth->finish;
                print "Query time was: ".$query_time." seconds\n";
                $i++;
        }
-----------------------------------------------------------------------------------------------------  
My apologies if I've left anything unclear.  I'm not even 100% sure that Perl is the factor responsible for the confusing results.  

Thanks,
Chris