Hyper-Threading technology is an innovative design from Intel that enables multi-threaded software applications to process threads in parallel within each processor resulting in increased utilization of processor execution resources. To make it short, it is to place two logical processors into a single CPU. As a result, an average improvement of 40% in CPU resource utilization yields higher processing throughput.
The history of Hyper-Threading, Anand Lal Shimpi from Anadtech was the first to uncover rumors about Jackson Technology's existence during the Spring 2001 IDF show, and he speculated that the technology might also be included in P4 chips, but was disabled at the fab. In Fall 2001 IDF, it was verified though Intel did not announce a time line for desktop P4-capable Hyper-Threaded processors. .
The processors have done an incredible job of trying to squeeze as much similarity as possible out of a single thread of execution. They have implemented features like super-pipelining, register renaming, out-of-order execution, and other techniques to process as many instructions from a serial code stream as possible, both simultaneously and at various stages in the processor pipeline. .
Processor architects say that it's becoming increasingly difficult to derive additional performance-boosting instruction-level parallelism (ILP) from x86 code streams using even more advanced micro architectural techniques than being used today. The incremental performance increases are arguably not worth the additional transistors, size, and power utilization costs required. The architects realize that different techniques must be used to improve performance. As process technologies advance, they will keep adding more onboard cache - even though more costly cache yields diminishing returns. Often, at any process technology level, it's very hard to squeeze out more ILP returns from x86 code streams. .
Hyper-Threading brings the idea of thread-level-parallelism (TLP) to single processors.
