Hyper-Threading: Why 2 Cores Can Act Like 4 Threads

Why Cores Can't Run at Full Capacity Continuously

A CPU core often waits — for data from memory, for a previous instruction to finish, or for I/O devices. During these waits the core sits idle.

Analogy: a chef waiting for water to boil could instead chop vegetables, raising overall efficiency.

Hyper-Threading , introduced by Intel in 2002, lets one physical core handle two threads simultaneously; when one thread waits, the other uses the execution units.

How Hyper-Threading Works

Each physical core contains multiple resources:

Execution units (ALU, FPU): perform calculations

Registers : temporary data storage

Cache : high-speed data access

Some resources are underused when a thread stalls (e.g., execution units idle during memory fetch). Hyper-Threading virtualizes one physical core into two logical cores (threads) . The two threads share the same execution units but have independent register copies and portions of cache.

When thread A waits for data, thread B immediately takes over the execution units — like the chef chopping vegetables while water boils.

Performance Gain: Typically 15%-30%, Not Double

Because both threads share one physical core's execution units, they still contend when both need compute simultaneously. Analogy: one chef cooking two dishes is faster than one dish, but not as fast as two chefs.

Gain varies by workload:

Multitasking (many programs open): noticeable improvement

Video encoding, 3D rendering (highly parallelizable): larger gains

Single-threaded applications : almost no gain, sometimes slight regression

Gaming : some titles benefit, but less than adding a real core

Intel Hyper-Threading vs AMD SMT

Hyper-Threading is Intel's branding; AMD calls the same principle SMT (Simultaneous Multithreading) . Both mainstream lineups support it:

Intel Core i5/i7/i9 mostly include Hyper-Threading

AMD Ryzen series all support SMT

Costs of Hyper-Threading

Power increase ~5%-10% : maintaining extra thread state consumes more energy

More heat : higher power draw raises cooling demand

Security risk : the 2018 "Port Contention" vulnerability (CVE-2018-3620) exploited Hyper-Threading, allowing two threads on the same core to spy on each other; high-security environments may disable it

Checking Hyper-Threading in the OS

Windows :

Right-click "This PC" → Properties → view processor info

Or Task Manager → Performance → CPU → see "Logical processors" count

Linux : lscpu Check "Thread(s) per core"; if it shows 2, Hyper-Threading/SMT is enabled.

macOS :

Apple M-series chips use a different hybrid architecture (performance/efficiency cores) and do not employ traditional Hyper-Threading

Exam Points (Software Qualification Exam)

Basic concept of Hyper-Threading

SMT (Simultaneous Multithreading) principle

Difference between logical processors and physical cores

Fundamentals of parallel computing: thread-level parallelism

Summary

Hyper-Threading lets one physical core run two threads concurrently, using otherwise idle execution resources. It does not create a new core; it makes the existing core busier. "4 cores 8 threads" means four real compute cores can handle eight tasks simultaneously, yielding roughly 20%-30% performance uplift.