CUDA.jl 5.6 and 5.7: Allocator cache, and asynchronous CUBLAS wrappers

Published by

on •1 min read

CUDA.jl v5.6 adds support for the new GPUArrays.jl caching allocator interface, which should improve performance of repetitive, memory-heavy applications. CUDA.jl v5.7 brings a greatly improved CuRef type, which enables fully asynchronous CUBLAS calls.