Hyper-threading and cores

In short: A processor has several cores that compute independently. Hyper-threading (SMT at AMD) lets each core handle two tasks (threads) at the same time.

In more detail: A core has many execution units that often wait (for example for data from memory). With two threads, the core uses the waiting time for the other. Depending on the program, this brings about 10 to 30 % more throughput, not double.

In Depth

• P and E cores: Newer processors mix performance cores (fast) and efficiency cores (frugal).
• More cores help: with rendering, compiling, video editing, virtualisation, servers.
• Single-core speed: counts in games and many everyday programs.
• Amdahl’s law: The non-parallelisable part of a program limits the gain from more cores.
• Threads in programming: thread (concept), concurrency.
• Cache: fast buffer memory inside the processor. A larger cache helps many games.

See also: CPU, Concurrency