Thread

  • What is it?
    In our glossary entry about cores, we explained that multiple cores can help processors handle a large number of tasks simultaneously. Threads serve a similar purpose, except they are the smallest sequence of instructions that can be executed independently within each core. In other words, while cores are a hardware component, threads are a virtual concept that exist inside the cores. Through a process known as multithreading (sometimes called hyperthreading), each core can run more than one thread—that is, it can begin working on a second set of instructions while waiting for the first set of instructions to be completed. In this way, the processor is used to its fullest potential.

  • Why do you need it?
    Just as most modern processors employ a multi-core architecture, most modern cores can engage in multithreading. While a processor is usually designed to distribute tasks evenly among all its cores before it burdens a single core with more than one task, the inclusion of multiple threads in each core has the effect of multiplying the processor's computing capabilities. Innovative computing techniques, such as parallel computing and high performance computing (HPC), can benefit from the higher number of cores and threads contained in modern processors.

  • How is GIGABYTE helpful?
    The advanced processors installed inside GIGABYTE servers are suitable for multithreading. For example, a single 64-core AMD EPYC™ CPU contains 128 threads, and GIGABYTE's H262 Series of High Density Servers can house up to eight such CPUs. If you do the math, this means a fully populated 42U (42 rack units) server rack can provide more than 20,000 threads. If you require servers that are capable of handling a large number of tasks simultaneously, you would do well to choose processors with the highest core and thread counts, and pair them with GIGABYTE's many server solutions, whether it's H-Series High Density ServersG-Series GPU Servers, or something else out of GIGABYTE's comprehensive product line.