Unified Memory

GPTKB entity

Statements (24)
Predicate Object
gptkbp:instanceOf computer memory architecture
gptkbp:allows automatic data migration
single memory address space
gptkbp:alsoKnownAs shared memory (in Apple context)
gptkbp:benefit reduced need for explicit memory copies
simplified programming model
gptkbp:contrastsWith discrete memory architecture
gptkbp:enables shared memory between CPU and GPU
dynamic memory allocation across devices
gptkbp:feature gptkb:Apple_M1
gptkb:Apple_M2
gptkb:Apple_M3
CUDA 6.0 and later
NVIDIA Pascal architecture and later
https://www.w3.org/2000/01/rdf-schema#label Unified Memory
gptkbp:used_in gptkb:machine_learning
gptkb:Apple_Silicon
gptkb:NVIDIA_CUDA
gptkb:AMD_ROCm
high-performance computing
graphics processing
gptkbp:bfsParent gptkb:CUDA_6.0
gptkb:NVIDIA_CUDA_Toolkit
gptkbp:bfsLayer 7