Intel’s Xeon brand of CPUs has had a long history already, which all started off with the original Pentium Pro. Whereas the Pentium Pro had integrated L2-cache in various cache size configurations it actually wasn’t mounted on the CPU-die but rather packaged with the CPU on the ceramic substrate, which was okay for a server solution, but rather expensive for widespread consumer level application. Thus when Intel decided to go from a socket approach to a slot approach with their Pentium II CPUs, it basically did so to cut cost and make 512Kb of off-die cache an affordable solution.
With the introduction of the Pentium II and Pentium III Xeon CPUs Intel implemented things similarly. Firstly they moved to a slot approach, Slot-2, much like their move to Slot-1 in the consumer market. The Slot-2 approach was basically meant to cut down on cost as well as provide enough room for larger cache size configurations, which now had enough printed circuit board area to be implemented. Furthermore it provided a smooth transition to higher clockspeed versions of the Xeon, as the Slot-2 provided a safe and pin-compatible upgrade path.
However in the past two years Intel has been diligently working on introducing their new socket based CPUs versions for all of their CPUs, Celeron, Pentium III and Pentium 4, and slowly phasing out the whole Slot-1 architecture. So it may come as no surprise that the Xeon too would inevitably move from a slot to a socket and with the new Intel Xeon DP that’s exactly what has happened.
Fig 1. Intel Pentium III Xeon CPU in the Slot-2 form factor, roughly twice the size of the Slot-1 Pentium II/III CPUs.