Hopper

E790551

Hopper is an NVIDIA GPU architecture designed for high-performance computing and AI workloads, featuring advanced capabilities such as enhanced Tensor Core acceleration.

Try in SPARQL Jump to: Statements Referenced by

Statements (47)

Predicate Object
instanceOf GPU microarchitecture
announcedBy NVIDIA Corporation NERFINISHED
architectureGeneration NVIDIA Hopper architecture NERFINISHED
codename Hopper NERFINISHED
designedFor data center GPUs
developer NVIDIA NERFINISHED
feature DPX instructions
FP8 precision support
HBM3 memory support
NVLink support
PCIe Gen5 support
Transformer Engine NERFINISHED
confidential computing capabilities
enhanced Tensor Cores
enhanced sparsity support
fourth-generation Tensor Cores
improved energy efficiency
improved mixed-precision performance
large-scale parallelism
multi-instance GPU (MIG)
marketSegment AI supercomputing
cloud computing
enterprise computing
memoryType High Bandwidth Memory NERFINISHED
namedAfter Grace Hopper NERFINISHED
optimizationFor data analytics
large language models
recommendation systems
scientific simulation
predecessor Ampere NERFINISHED
successor Blackwell NERFINISHED
supportsComputationType BF16 GENERATED
FP16 GENERATED
FP32 GENERATED
FP64 GENERATED
FP8 GENERATED
INT8 GENERATED
TF32 GENERATED
supportsParallelProgrammingModel CUDA NERFINISHED
CUDA Graphs NERFINISHED
NVIDIA NCCL NERFINISHED
targetApplicationDomain artificial intelligence
deep learning
high-performance computing
machine learning
usedInProduct NVIDIA H100 Tensor Core GPU NERFINISHED
NVIDIA H200 Tensor Core GPU NERFINISHED

Referenced by (1)

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