Performance of subsequent query with the same table

[Sophia]

Hi,
I am trying to update approx. 3800 rows in a table in two steps:
1. set all values to 0
2. set some of them in a loop to a calculated value
The first step is fast (< 1 s). However the second step takes around 3 s. Maybe there is some problem in the code, because if I run the code without the first step or run the first query on another table, the second step is very fast, too (< 0.1 s). It is slow only when executed after the first step on the same table.
Here is some code:

Class.forName("net.ucanaccess.jdbc.UcanaccessDriver");
conn = DriverManager.getConnection("jdbc:ucanaccess://" + path + ";singleconnection=true" ,"", ""); 

// First update (reset all values)
long start = System.currentTimeMillis();         
PreparedStatement ps2 = conn.prepareStatement("UPDATE mytab SET val = 0.0;"); 
ps2.executeUpdate();
ps2.close();
System.out.println("Update 1: " + (System.currentTimeMillis()-start)/1000.0 + " s");

// Second update
start = System.currentTimeMillis();
conn.setAutoCommit(false);
int count = 0;
PreparedStatement ps = conn.prepareStatement("UPDATE mytab SET val = ? WHERE id = ?;");

for(int j= 1; j <= 3600; j++){
      double value;
      // calculate some value ....
      value = 1.3;
      // update table under certain conditions
      if(true){
              ps.setDouble(1, value);
              ps.setInt(2, j);
              ps.addBatch();
              count++;
       }
       if(count > 200){
               ps.executeBatch();
               count = 0;
        }
}
ps.executeBatch();
conn.commit();
System.out.println("Update 2: " + (System.currentTimeMillis()-start)/1000.0 + " s");

How to avoid this? Maybe some mistake?
Thank you for any advice!

[Gord Thompson]

This issue is being addressed on Stack Overflow.