Are you sure it was really running at 3.4 when you were timimg QMDs? I ask because 225 ppd is almost exactly what it should put out at 2.26 GHz, 2.26 x 95ppd/GHz = 214.7ppd. The 95ppd/GHz figure comes from a lot of folks not just me. Silver built a 2.26 GHz cellyd, oc'd to 3.4 GHz with some tight ram timings and got 435ppd, 128ppd/GHz. On QMDs, celeron Ds are memory bandwidth bound until FSB goes over ~750 effective. At this point memory bandwidth exceeds ~4000mb/sec and production improves dramatically. A 50% oc results in 100% increase in production. If you've got everything working right, these are the results you'll see.