Habana Gaudi processor

E660956

The Habana Gaudi processor is a specialized AI training accelerator designed by Habana Labs (an Intel company) to deliver high-performance, scalable deep learning computation in data centers.

Try in SPARQL Jump to: Surface forms Statements Referenced by

All labels observed (1)

Label Occurrences
Habana Gaudi processor canonical 1

Statements (47)

Predicate Object
instanceOf AI accelerator
deep learning training processor
announcedBy Habana Labs NERFINISHED
architecture custom AI training architecture
competesWith NVIDIA A100 NERFINISHED
NVIDIA H100 NERFINISHED
connectivity RoCE-based Ethernet networking
high-speed on-chip fabric
deploymentFormFactor PCIe card
mezzanine card
server system
designer Habana Labs NERFINISHED
ecosystem Intel AI portfolio
feature HBM-based memory subsystem
matrix math engines
on-chip SRAM
programmable Tensor Processing Cores
manufacturer Habana Labs NERFINISHED
optimizedFor TCO efficiency
large-scale training clusters
ownedBy Intel NERFINISHED
parentCompany Intel NERFINISHED
powerOptimizedFor performance-per-watt
purpose AI training acceleration
deep learning training
softwareStack SynapseAI NERFINISHED
successor Habana Gaudi2
supports BF16 NERFINISHED
FP16
INT8
PyTorch NERFINISHED
TensorFlow NERFINISHED
data parallelism
distributed data loaders
large batch training
mixed-precision training
model parallelism
multi-node training
scale-out training
standard Ethernet networking
targetMarket cloud service providers
enterprise data centers
targetWorkloads NLP models
computer vision models
recommendation systems
useCase cloud AI training
data center AI workloads

Referenced by (1)

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

Goya inference processor relatedProduct Habana Gaudi processor