Skip to content

Processor Architecture and Design — Glossary

TermMeaning
ABI (Application Binary Interface)Binary-level contract: calling convention, name mangling, layout, library format
Access timeTime to read or write a memory location
Active-low signalA signal that is asserted when low (0)
ALU (Arithmetic Logic Unit)Hardware block that performs arithmetic and logic operations
AMAT (Average Memory Access Time)Hit time + miss rate × miss penalty
API (Application Programming Interface)Source-level contract: functions, signatures, headers
Architecture (system)Structure and organization of a computer's components
AssemblerTranslates assembly mnemonics into machine code
AtomicityAn operation that completes entirely or not at all
BacktrackingFlushing instructions that were speculatively executed
Big-endianByte order with the most-significant byte first
Binary-coded decimal (BCD)Decimal digits encoded in 4-bit binary
BitBinary digit, 0 or 1
Booth's algorithmMultiplication algorithm for signed binary numbers
Branch predictionHardware guessing the outcome of a branch to reduce control hazards
BusShared set of wires carrying data, address, or control signals
Bus arbitrationDeciding which device may use a shared bus
Byte-addressableEach byte has its own address
CacheSmall fast memory keeping copies of recently/frequently used data
Cache lineA unit of data transferred between cache and main memory
Cache coherenceKeeping multiple cached copies of the same data consistent
Cache hitThe requested data is found in cache
Cache missThe requested data is not in cache; must be fetched
CaptureSaving processor state when an interrupt occurs
Carry flagProcessor flag set when an arithmetic result carries out of the most-significant bit
Chip selectSignal enabling a particular chip/device on a bus
Clock cycleOne period of the clock signal
Clock frequencyNumber of cycles per second (Hz)
Combinational logicLogic whose output depends only on current inputs
Condition codeProcessor flag (Z, C, N, V) reflecting arithmetic results
Context switchSaving and restoring process state to change the running task
Control hazardPipeline stall caused by not knowing the next instruction address
Control unitHardware that generates control signals sequencing CPU operations
Convention (calling)Rules for passing arguments and returning values between functions
CPU (Central Processing Unit)The processor that fetches, decodes, and executes instructions
CPI (Cycles Per Instruction)Average clock cycles a processor needs per instruction
Critical pathLongest delay path that sets the clock period
Cross-platformWorking across different hardware/OS combinations
Data busBus carrying actual data, usually bidirectional
Data hazardPipeline stall caused by a value not yet being available
DecodeDetermining what operation an instruction specifies
Dedicated hardware illusionThe virtual machine view where each program thinks it owns the hardware
Delay slotAn instruction executed after a branch but before it takes effect
DMA (Direct Memory Access)Hardware transfer of data between device and memory without CPU help
Effective addressThe final memory address computed for an operand
ExecutePerforming the operation specified by a decoded instruction
ExceptionA synchronous interrupt usually caused by the current instruction
External interruptA hardware interrupt generated outside the CPU (e.g. by a device)
Fan-outNumber of inputs one output can drive
FetchReading the next instruction from memory
Flip-flopA 1-bit memory element storing its input on a clock edge
Floating-point unit (FPU)ALU specialized for floating-point arithmetic
Global descriptor table (GDT)x86 data structure describing segments, gates, and privilege levels
Hardwired controlControl unit built from fixed logic rather than microcode
InstructionA single machine-code operation
Instruction cacheCache storing recently fetched instructions
Instruction Register (IR)Register holding the currently decoded instruction
Instruction Set Architecture (ISA)The programmer-visible contract of an architecture
InterruptAn asynchronous signal that pauses normal execution to handle an event
Interrupt acknowledgeHandshake signal/response when the CPU services an interrupt
Interrupt latencyTime from the interrupt request to the start of the ISR
Interrupt service routine (ISR)Function executed when an interrupt is taken
Interrupt vectorIndex or address pointing to the correct ISR
I/O (Input/Output)Interface between a computer and external devices
JobA program in the context of the operating system
Little-endianByte order with the least-significant byte first
Logic gatePhysical device implementing a Boolean function
Logical addressAn address used by a program before translation (virtual address)
Machine languageBinary code directly executed by the CPU
Memory mappingAssigning memory addresses to devices or files
Memory management unit (MMU)Hardware translating virtual to physical addresses
MicrocodeLow-level instructions implementing higher-level machine instructions
MicroinstructionA control step in a microprogrammed control unit
Miss penaltyTime to fetch a block from the next level on a cache miss
Mode bitsProcessor state bits selecting user/kernel mode
MMU (Memory Management Unit)See Memory management unit
MultiprogrammingRunning multiple programs in memory to keep the CPU busy
Multiplexer (MUX)A selector choosing one of several inputs onto one output
Native codeMachine code compiled for the local processor
Non-maskable interrupt (NMI)An interrupt that cannot be disabled by software
OperandA value on which an operation acts
Operating system (OS)System software managing hardware and programs
PageA fixed-size block of virtual/physical memory (e.g., 4 KiB)
Page faultAn exception when a referenced page is not in physical memory
Page frameA fixed-size block of physical memory
PagingMemory management scheme mapping fixed-size pages to frames
PCI / PCIeBus standards for connecting peripheral devices
Performance counterHardware register counting events like cycles or misses
PipelineOverlapping execution of instruction stages
Pipeline stallPausing the pipeline because an operand or resource is unavailable
Pipeline throughputRate at which the pipeline completes instructions
PrefetchingLoading data/instructions before they are needed
Privilege levelThe CPU's permission level (e.g., ring 0 kernel, ring 3 user)
ProcessorSynonym for CPU
Program counter (PC)Register pointing to the next instruction to fetch
Protection ringHardware isolation level (0–3 on x86) controlling access to resources
Pseudo-ROMRAM that behaves like ROM during boot but can be updated
Real addressA physical memory address
RegisterSmall fast storage inside the CPU
Register fileA collection of registers selectable by number
Relative jumpA branch whose target is offset from the current PC
Reset vectorAddress the CPU jumps to on power-on or reset
RISC (Reduced Instruction Set Computer)ISA style with few simple, fixed-size instructions
ROM (Read-Only Memory)Non-volatile memory that is hard to modify
Rotational latencyTime waiting for the correct disk sector to rotate under the head
SegfaultShort for segmentation fault; an invalid memory access
Segmentation faultProtection fault from accessing disallowed memory
SemaphoreA synchronization primitive with an integer value and queue
Set-associative cacheA cache where each set holds n lines; compromise between direct and full
Shadow registerA backup register set used on mode changes
Side effectAn operation that changes state beyond its return value
SignalA wire that asserts to notify the processor of an event
Soft interruptAn interrupt flagged in software for later, safer handling
Software interruptA trap or system call raised intentionally by a program
Source-synchronousA clocking technique transferring data with a clock from the source
SRAM (Static RAM)RAM that retains a bit without refresh, using a flip-flop
StorableA value that can be saved and restored
Structural hazardA conflict where hardware resources required by one stage are unavailable
System busThe collection of buses connecting CPU, memory, and I/O
System callA controlled request by a program for OS service
TLB (Translation Lookaside Buffer)Hardware cache of recent virtual-to-physical address translations
TickA clock pulse
Timestamp counterRegister counting CPU cycles since boot
Timing analysisEstimating how long operations take
TrapA synchronous interrupt, e.g. a system call or exception
Translation lookaside bufferSee TLB
Two-level cacheTypical CPU cache: private L1 (+L2) per core and shared L3
User modeThe restricted CPU mode where applications run
V (overflow flag)Processor flag set when a signed arithmetic result overflows
Vector interruptAn interrupt that supplies its own vector automatically
Virtual addressAn address used by a program before translation to a physical address
Virtual memoryMemory management giving each process an address space larger than RAM
Wait stateExtra cycles inserted when a device is slower than the bus
Write bufferA queue allowing writes to complete without stalling the CPU
Write-throughA cache policy updating the lower level on every write
Write-backA cache policy writing the lower level only when a line is evicted

Built from Markdown with VitePress.