Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Wait state
open-in-new
<p>A wait state is a delay experienced by a computer <a href="/facts/Central_processing_unit/Jxp1bqMV">processor</a> when accessing external <a href="/facts/Computer_storage/wPHKFpdx">memory</a> or another device that is slow to respond. </p><p>Computer microprocessors generally run much faster than the computer's other subsystems, which hold the data the CPU reads and writes. Even memory, the fastest of these, cannot supply data as fast as the CPU could process it. In an example from 2011, typical PC processors like the <a href="/facts/Intel/SMF0gJJX">Intel</a> <a href="/facts/Core_2/9gJP9o8p">Core 2</a> and the <a href="/facts/AMD/CB1G7QTb">AMD</a> <a href="/facts/Athlon_64_X2/aB8h01HY">Athlon 64 X2</a> run with a clock of several <a href="/facts/GHz/3mzLUghh">GHz</a>, which means that one <a href="/facts/Clock_cycle/9UfTzGHx">clock cycle</a> is less than 1 <a href="/facts/Nanosecond/gARLTp4c">nanosecond</a> (typically about 0.3 ns to 0.5 ns on modern desktop CPUs), while main memory has a <a href="/facts/Memory_timings/rCLwdVPz">latency</a> of about 15–30 ns. Some second-level <a href="/facts/CPU_cache/x8BYFcdv">CPU caches</a> run slower than the processor core. </p><p>When the processor needs to access external memory, it starts placing the address of the requested information on the <a href="/facts/Address_bus/R45TKwS7">address bus</a>. It then must wait for the answer, that may come back tens if not hundreds of cycles later. Each of the cycles spent waiting is called a wait state. </p><p>Wait states are a pure waste of a processor's performance. Modern designs try to eliminate or hide them using a variety of techniques: <a href="/facts/CPU_cache/x8BYFcdv">CPU caches</a>, <a href="/facts/Instruction_pipeline/3zy6CcX1">instruction pipelines</a>, <a href="/facts/Instruction_prefetch/cuCGl0Vm">instruction prefetch</a>, <a href="/facts/Branch_prediction/pGPhFMj5">branch prediction</a>, <a href="/facts/Simultaneous_multithreading/bk8ec9yn">simultaneous multithreading</a> and others. No single technique is 100% successful, but together can significantly reduce the problem. </p>