IBM has been driving the competition mad with its RAMP-C transaction processing benchmark, which it uses as evidence that its own machines are so much better than anyone else’s – yet refuses to release the code so that impartial benchmarkers can try to replicate the results – but ICL has called IBM’s bluff and seems to have the perfect answer: it has come up with benchmark figures for its new Series 39 models that it says are measurements based on ICL’s interpretation of the IBM RAMP-C benchmark (CI No 982); IBM can say that ICL has got it all wrong, but ICL can be sure its message gets to its prospects long before the IBM refutation gets there.