Gaudi

E170432

Gaudi is Habana Labs’ AI training processor designed to accelerate deep learning workloads with high performance and efficiency.

All labels observed (1)

Label Occurrences
Gaudi canonical 2

How this entity was disambiguated

Statements (46)

Predicate Object
instanceOf AI training processor
deep learning accelerator
announcedAfter Habana Goya
announcedBy Habana Labs
architectureType custom AI accelerator architecture
belongsToProductFamily Habana Gaudi family
compatibleWith PyTorch
TensorFlow
contrastedWith Goya inference processor
countryOfOrigin Israel
deploymentFormFactor PCIe card
mezzanine card
designedFor AI training
data center deployment
deep learning workloads
developer Habana Labs
feature integrated high-speed networking
on-chip networking fabric
support for standard frameworks via software stack
focus cost efficiency
high performance
power efficiency
hardwareCategory accelerator card
intendedEnvironment cloud infrastructure
server systems
manufacturer Habana Labs
optimizationGoal scalability across multiple accelerators
throughput for large-scale training
parentOrganization Intel Corporation
surface form: Intel
positionInProductLine training-optimized processor
primaryUseCase neural network training
softwareStack SynapseAI
successor Gaudi2
Gaudi3
supports BF16 precision
FP16 precision
INT8 operations
distributed training
mixed-precision training
training of convolutional neural networks
training of transformer models
targetMarket cloud service providers
enterprise data centers
useCase computer vision training workloads
natural language processing training workloads
recommendation system training workloads

How these facts were elicited

Referenced by (2)

Full triples — surface form annotated when it differs from this entity's canonical label.

Habana Labs product Gaudi
Gaudi2 predecessor Gaudi