Intel Deep Learning Boost (VNNI)
GPTKB entity
Statements (27)
Predicate | Object |
---|---|
gptkbp:instanceOf |
CPU instruction set extension
|
gptkbp:abbreviation |
gptkb:Vector_Neural_Network_Instructions
|
gptkbp:architecture |
x86
|
gptkbp:availableOn |
2019
|
gptkbp:category |
gptkb:artificial_intelligence
gptkb:machine_learning hardware acceleration |
gptkbp:developedBy |
gptkb:Intel
|
gptkbp:feature |
fused multiply-add instructions
int8 and int16 support optimized for convolutional neural networks |
https://www.w3.org/2000/01/rdf-schema#label |
Intel Deep Learning Boost (VNNI)
|
gptkbp:improves |
inference performance
training performance |
gptkbp:introducedIn |
gptkb:Intel_Alder_Lake
gptkb:Intel_Sapphire_Rapids gptkb:Intel_Ice_Lake gptkb:Intel_Tiger_Lake gptkb:Intel_Cascade_Lake gptkb:Intel_Cooper_Lake |
gptkbp:purpose |
accelerate deep learning workloads
|
gptkbp:relatedTo |
gptkb:AMX
gptkb:AVX-512 |
gptkbp:usedIn |
client CPUs
data center CPUs |
gptkbp:bfsParent |
gptkb:Xeon_Scalable
|
gptkbp:bfsLayer |
5
|