Understanding CPU characteristics - CompTIA A+ 220-801: 1.6

Processor Speed

• Real speed: CPU clock
• Historical qualification of speed
• Megahartz (MHz) or gigahertz (GHz)

• Actual speed
• Combination of different factors
• Clock speed, CPU architecture, bus speed, bus width, L1 cache size, L2 cache size, operating system capabilities

• CPU manufactures moving away from marketing performance in clock cycles
• No broadly accepted measure of performance
• Use a benchmark that works for you

Processor Cores

Quad-core Sandy Bridge processor

• Dual-core / Quad- core / Multi-core
• Multiple cores
• Each core probably has its own cache
• The entire chip may have a shared cache

 CPU cache

•  Super fast memory
• There;s not much of it
• Holds data, instructions and/ or results
• Level 1 cache
• First check

• Level 2 cache
• Secondary data

• Level 3 cache
• Still on the chip, these days

Hyperthreading

• Hyper-Threading Technology (HTT)
• One CPU, acts like two
• Doesn't actually work as fast as two
• 15% to 30% performance improvement
• Operating system must be written fo HTT
• Windows XP and later

Virtualization support 

• Run other operating systems within a single hardware platform
• Multiple operating systems share physical hardware components
• Virtualization in software was limited
• Performance and hardware management challanges

• Virtualization added to the processor
• Hardware is faster and easier to manage
• Intel Virtualization Technology (VT)
• AMD Virtualization (AMD-V)

32 bit vs 64 bit

• Data "width"
• Datapath, integer size, memory addresses
• Registers, address bus, data bus
• Larger bus sizes move more data
• 64 bit data bus can move twice as much as a 32 bit data bus
• Even if the processor is 32 bits

• Bus size may not necessarily match the CPU architecture
• Software must be optimized to take advantage of 64 bit architectures
• Operating systems, hardware drivers, applications

Integrated GPU

• Graphics processing unit
• Part of a video adapter, motherboard, or CPU
• Graphics rendering requires a lot of work


• Dedicated hardware speeds the process

• On-die GPUs are increasingly common
• But not as powerful as separate adapters