Tensor cores

GPTKB entity

Statements (46)
Predicate Object
gptkbp:instanceOf hardware component
processing unit
gptkbp:architecture gptkb:Volta
gptkb:Hopper
gptkb:Ampere
gptkb:Turing
gptkbp:developedBy gptkb:NVIDIA
gptkbp:enables energy efficiency in AI workloads
faster inference of neural networks
faster training of neural networks
high throughput computing
gptkbp:features gptkb:NVIDIA_GPU_architectures
gptkbp:foundIn gptkb:NVIDIA_A100
gptkb:NVIDIA_Tesla_V100
gptkb:NVIDIA_GPUs
gptkb:NVIDIA_H100
gptkb:NVIDIA_RTX_series
https://www.w3.org/2000/01/rdf-schema#label Tensor cores
gptkbp:introducedIn gptkb:NVIDIA_Volta_architecture
2017
gptkbp:marketedAs AI acceleration
gptkbp:optimizedFor matrix multiplication
convolution operations
gptkbp:purpose accelerate matrix operations
gptkbp:relatedTo RT cores
CUDA cores
gptkbp:supports BF16
FP16
INT8
FP32 accumulation
mixed-precision computation
gptkbp:usedBy gptkb:TensorFlow
gptkb:cuDNN
gptkb:PyTorch
data scientists
AI researchers
cloud computing providers
gptkbp:usedIn gptkb:artificial_intelligence
deep learning
scientific computing
gptkbp:bfsParent gptkb:NVIDIA_RTX
gptkb:NVIDIA_GPUs
gptkb:NVIDIA_RTX_30_series
gptkb:NVIDIA_GeForce_RTX_30_series
gptkb:RTX_series
gptkbp:bfsLayer 6