FMA instruction set

Wikibooks has a book on the topic of: <i>X86 Assembly/AVX, AVX2, FMA3, FMA4</i>

<p>The FMA instruction set is an extension to the 128- and 256-bit <a href="/facts/Streaming_SIMD_Extensions/mMa6EFyB">Streaming SIMD Extensions</a> instructions in the <a href="/facts/X86/Kypq5fgu">x86</a> <a href="/facts/Microprocessor/Fx9CaQht">microprocessor</a> <a href="/facts/Instruction_set/8p8vwUeE">instruction set</a> to perform <a href="/facts/Fused_multiply%25E2%2580%2593add/L26C9A1Y">fused multiply–add</a> (FMA) operations. There are two variants:
</p>
<ul><li>FMA4 is supported in <a href="/facts/AMD/CB1G7QTb">AMD</a> processors starting with the <a href="/facts/Bulldozer_(microarchitecture)/fUPzoGrd">Bulldozer</a> architecture. FMA4 was performed in hardware before FMA3 was. Support for FMA4 has been removed since <a href="/facts/Zen_1/oChXIp3y">Zen 1</a>.</li>
<li>FMA3 is supported in AMD processors starting with the <a href="/facts/Piledriver_(microarchitecture)/Fhxly06H">Piledriver</a> architecture and <a href="/facts/Intel_Corporation/SMF0gJJX">Intel</a> starting with <a href="/facts/Haswell_(microarchitecture)/C19Wlzhm">Haswell processors</a> and <a href="/facts/Broadwell_(microarchitecture)/32jqNWRS">Broadwell processors</a> since 2014.</li></ul>

FMA instruction set open-in-new

FMA instruction set