Job Questions > What's the toughest problem you've had to solve at > Sun Microsystems > Jerry's Answer

We were looking for ways to simulate really large, multi-processor servers with smaller systems. We noticed performance result discrepancies between two different models. I wrote instruction-by-instruction logs, wrote scripts to parse the logs and show important discrepancies. I made hypotheses about why the models differed, followed my hunches, learned new things about computer registers, and discoved the bug in a tool which had been written by another department some time earlier. We then made the corrections necessary to ensure the model's accuracy.