What Is Hyperthreading?

Hyperthreading is Intel’s name for simultaneous multithreading (SMT), a technique that lets one physical processor core run two instruction streams, or threads, at the same time. The operating system sees each core as two logical processors, so a 6-core processor with hyperthreading shows 12 in Task Manager or in Linux.

Intel introduced Hyper-Threading Technology in 2002, first on Xeon and then on Pentium 4 processors. AMD added SMT with its Zen architecture in 2017 and uses the generic name. The idea is the same in both: a core spends much of its time waiting, for data from memory or for a branch to resolve, and a second thread can use the execution resources that would otherwise sit idle.

A core with hyperthreading duplicates the parts that hold a thread’s state, such as the registers and the instruction pointer, but shares the expensive parts: the execution engines, the caches and the connection to memory. The core’s scheduler picks instructions from both threads every cycle and issues whichever are ready.

Because the two threads share one set of execution resources, the gain depends on how often a single thread leaves them idle. Code that stalls on memory gains the most. Code that already keeps the core busy, such as tight maths loops, gains little and can even slow down slightly as the threads compete for cache. Across typical multithreaded workloads, the extra throughput is usually around 15 to 30%, well short of a second core.

Support has changed with recent generations, so the thread count on a datasheet is worth checking rather than assuming twice the core count.

The hybrid split between performance and efficient cores is covered in our guide to P-cores, E-cores and Intel hybrid architecture, and the Series 2 changes in Intel Core Ultra explained.

Hyperthreading is a throughput feature. It helps when there are more runnable threads than cores, and it does nothing for a single thread on its own.

In an industrial computer the question is less about benchmark scores and more about consistent behavior. A gateway or vision system running several containers usually benefits from hyperthreading. A machine controller with strict cycle times may run better with SMT disabled and specific cores reserved for the control task, an approach called core isolation. The choice is made per system, and it is worth testing the real application both ways before fixing the setting for a fleet.

Licensing can also matter. Some software and hypervisors are licensed per core and others per thread, so the thread count affects cost as well as performance. Socketed Intel processors with P-cores, such as those in our Intel Bartlett Lake range, keep hyperthreading on their P-cores, while AMD Ryzen Embedded processors provide SMT across most of the family.

To change the setting, look in the BIOS or UEFI for an option named Hyper-Threading, Intel HT Technology or SMT Control. In Windows, Task Manager shows cores and logical processors separately. In Linux, the lscpu command reports threads per core, and the file /sys/devices/system/cpu/smt/control lets you switch SMT off at runtime.

We set SMT, core isolation and BIOS options as part of every build, so each industrial computer ships configured for the way your application uses its cores.