search

jonysy / parenchyma

An extensible HPC framework for CUDA, OpenCL and native CPU.
75 stars 4 forks source link

Consider using a tensor lib for native memory (instead of flat slices) #20

Open jonysy opened 7 years ago

jonysy commented 7 years ago
jonysy commented 7 years ago

alexandermorozov:

ndarray solves first part by accepting constant values, slices, Vecs and iterators as initializers. But it doesn't allow to allocate uninitialized chunk of memory to be overwritten later; looks like there is no way to statically check if usage is correct, so checks have to be done at runtime and that'll have a small runtime cost.