An operation computing a × b + c in one instruction, rounding only once at the end without rounding the intermediate product. It is faster and more accurate than separate multiply and add and is heavily used in matrix computation.
Field: Computer Arithmetic
Available instructions differ by CPU; a * b + c, for example, becomes one instruction if the CPU has a fused multiply-add (FMA) and two (multiply and add) otherwise.
Which describes instruction selection, one stage of code ge…Arithmetic shift bfloat16 Booth's algorithm Carry-lookahead adder Carry-save adder Catastrophic cancellation Exponent field Fixed point Floating-point absorption FP8