DArray: Implement efficient in-place matrix-matrix multiply

JuliaParallel / Dagger.jl

A framework for out-of-core and parallel execution

Other

610 stars 66 forks source link

Closed Rabab53 closed 3 months ago

Rabab53 commented 3 months ago

This PR implements:

Optimized mul! and * using datadeps
Optimized transpose/adjoint using datadeps
copyto! using datadeps
Optimized getindex with optional task-local caching
setindex! (to be optimized later)
A convenient test runner script (used like julia test/runtests.jl --test array/linalg/matmul --test array/copies)

jpsamaroo commented 3 months ago

Thanks @Rabab53!