Round 6 Aftermath

After hastily collating results late yesterday, I have hashed together some commentary to let you all in on the story. I used the same scripts from round 1 to calculate the scores so I hope nothing went wrong. It all looks fairly correct from a glance at the database. I hope to add this information to the website soon.

Christian Schmidt takes the round with a narrower margin than most of the other round. I suspect the double d-clear did very well against W.F.B.'s stone. Macrae was the close second and some of the code seemed very familiar :-) John Metcalf is a late entry to the tournament and submitted a surprise scanner. Welcome.

The majority submitted p-spacers. Three warriors contained CLP although everyone seemed innoculated.

I didn't include the soup in the end, prefering Pihlaja's suggestion to modify pmars to add the SPL and NOP instructions. These were added to sim.c. I couldn't use the diffs that Joonas sent (through ignorance I am afraid) so I rolled my own. I hope I didn't break anything and that this didn't foil any cunning plans. The addition follows:

    INITIALINST.A_mode = INITIALINST.B_mode = (FIELD_T) IMMEDIATE;
    do {
      INITIALINST.A_value = (int) (8000.0 * rand() / (RAND_MAX + 1.0));
      INITIALINST.B_value = (int) (8000.0 * rand() / (RAND_MAX + 1.0));
      if (++type > 10) type = 1;
      if (type == 1) INITIALINST.opcode = (FIELD_T) SPL *8 + (FIELD_T) mB;
      else INITIALINST.opcode = (FIELD_T) NOP *8 + (FIELD_T) mF;

      memory[addrA] = INITIALINST;
    } while (++addrA < coreSize);

As in round 1, the 200 battles were run twice for each pair of warriors and these were averaged.

Link to the authors strategy lines and comments here