fused multiply-accumulate (FMA)
GPTKB entity
Statements (18)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
floating-point operation
|
| gptkbp:abbreviation |
gptkb:FMA
|
| https://www.w3.org/2000/01/rdf-schema#label |
fused multiply-accumulate (FMA)
|
| gptkbp:improves |
performance of floating-point calculations
|
| gptkbp:instructionSet |
FMA3
FMA4 |
| gptkbp:introducedIn |
gptkb:IBM_POWER1
|
| gptkbp:operator |
(a × b) + c with a single rounding
|
| gptkbp:performs |
multiplication and addition in a single step
|
| gptkbp:reduces |
rounding errors
|
| gptkbp:standardizedBy |
gptkb:IEEE_754-2008
|
| gptkbp:supportedBy |
gptkb:ARM_processors
x86 processors |
| gptkbp:usedIn |
gptkb:machine_learning
scientific computing digital signal processing |
| gptkbp:bfsParent |
gptkb:VFPv4
|
| gptkbp:bfsLayer |
6
|