NVIDIA Volta architecture

URI: https://gptkb.org/entity/NVIDIA_Volta_architecture
GPTKB entity
Statements (39)
Predicate Object
gptkbp:instanceOf gptkb:graphics_card
gptkbp:announced gptkb:GTC_2017
gptkbp:architecture parallel computing
gptkbp:compatibleWith consumer GeForce GPUs
gptkbp:developedBy gptkb:NVIDIA
gptkbp:feature dynamic parallelism
increased memory bandwidth
improved energy efficiency
NVLink 2.0 support
Tensor Cores for deep learning
hardware acceleration for AI workloads
gptkbp:firstGPUModel gptkb:NVIDIA_Tesla_V100
gptkbp:firstReleased 2017
gptkbp:hasConcept gptkb:GV100
https://www.w3.org/2000/01/rdf-schema#label NVIDIA Volta architecture
gptkbp:marketedAs AI research
data centers
high-performance computing
gptkbp:maxCUDAcores 5120
gptkbp:maxTensorCores 640
gptkbp:maxTransistors 21.1 billion
gptkbp:memoryBusWidth 900 GB/s
gptkbp:notableProduct gptkb:NVIDIA_Tesla_V100
gptkb:NVIDIA_Quadro_GV100
gptkb:NVIDIA_Titan_V
gptkbp:predecessor gptkb:NVIDIA_Pascal_architecture
gptkbp:processNode 12nm FinFET
gptkbp:RAM 32 GB HBM2
gptkbp:successor gptkb:NVIDIA_Turing_architecture
gptkbp:supports gptkb:NVLink
gptkb:HBM2_memory
gptkb:CUDA_Compute_Capability_7.0
gptkb:Tensor_Cores
gptkbp:usedIn gptkb:NVIDIA_Tesla_V100
gptkb:NVIDIA_Quadro_GV100
gptkb:NVIDIA_Titan_V
gptkbp:bfsParent gptkb:NVLink
gptkb:NVLink_2.0
gptkbp:bfsLayer 6