Optimizing Cache Accesses with Tensor Memory Format Search for Transformers in TVM.

Xianghuan He Xitong Gao Juanjuan Zhao Chengxi Gao Kejiang Ye

Published in: CLOUD (2022)

Keyphrases

memory access
search algorithm
search space
main memory
search strategy
garbage collection
higher order
search methods
access patterns
memory hierarchy
multimedia
data access
memory requirements
limited memory
resource consumption
cache conscious