They should have still tried 2T timing even if the performance would have been less.. They also only tried one latency config and benched it ~10 times... imo kinda pointless because they were all scaled about the same. Summary: more!