Hardware
CPU
Also known as: Central Processing Unit, Processor
The primary processor in a computer — executes the OS, runs applications, and handles general-purpose computation. Core count determines how many tasks run simultaneously.
The CPU executes instructions. It's the general-purpose brain of the system — fetching instructions from memory, decoding them, executing them, writing results. Everything from running the OS to processing database queries to executing application logic runs through the CPU.
Cores and threads
Modern CPUs have multiple cores — independent processing units on a single chip. A server CPU might have 8, 16, 32, or 64 cores. Hyper-threading (Intel) or SMT (AMD) allows each physical core to run two threads simultaneously, roughly doubling the apparent core count for multithreaded workloads.
For virtualization, core count directly determines how many vCPUs you can allocate across VMs. A dual-socket server with two 32-core CPUs can support workloads requiring up to 128 vCPUs (with 2:1 overcommit).
CPU generations and architectures
CPU generations matter for server purchasing. Intel Xeon and AMD EPYC are the dominant server CPU lines. Performance, power efficiency, memory bandwidth, and PCIe lane count vary significantly between generations. For most infrastructure purchases, comparing specific benchmark workloads (compute-intensive vs memory bandwidth vs I/O-heavy) is more useful than comparing core counts alone.
CPU vs GPU
CPUs are designed for sequential processing — a small number of powerful cores handling complex, varied instructions. GPUs have thousands of simpler cores optimized for parallel processing of the same instruction across large datasets. AI workloads have made the distinction relevant for a much broader audience. See GPUA processor with thousands of small cores optimized for parallel computation. Originally designed for graphics, GPUs now drive AI inference and training, and any workload requiring massive parallel processing. for detail on where each applies.