What is the technical reason for certain socket processors being faster than others? For example an AMD sempron 3300+/2.00ghz is slower than an AMD Athlon which is also 3300+/2.00ghz.
Well, you've got to compare equals here. The Sempron is AMD's value CPU. The Athlon brand is the high-end/enthusiast chip. The Sempron starts off with less on-die cache than the Athlon, which cuts out some performance. Unlike past CPU's, and Intel's current lineup, the Athlon64 (and Sempron derivatives) has the memory controller integrated into the CPU itself. Traditional designs have the memory controller grouped in with the chipset on the motherboard. This is where the Athlon64 gets it's performance from. The 64-bit capabilities, unlike commonly thought, are not the A64 line's secret. AMD initially launched the Athlon64 en masse in Socket 754 form, but later transitioned to Socket 939, to provide dual-channel memory capabilities the Socket 754 lacked. The other part is the HyperTransport frequency (HTT). This replaces the front-side bus (FSB) on the A64 architecture. It's kinda like the traditional FSB, but not exactly. The FSB doesn't exist, because it's pretty much eliminated with the memory controller being on the CPU itself. The HTT is the internal link between the CPU and memory controller. With Socket 754, the HTT link runs at 800MHz (it may also be referenced at 1600MHz); 939 runs at 1000MHz (1GHz), commonly referenced at 2000MHz (2GHz). All else is equal (clock speeds, cache), a Socket 939 CPU will perform better overall. With Intel, and also pre-Athlon64 architectures (including Socket 462 aka Socket A Semprons), the value chip was crippled in comparison to it's bigger brother (the Athlon or Pentium) in two ways: less cache, and a slower FSB. Another common tactic was slower clock speeds than the big boys.
