-
-
Notifications
You must be signed in to change notification settings - Fork 8
Open
Description
TODO:
- add a
Context
struct to all vectors/matrices/operators that can carry the details of the execution/allocation context ( e.g. the device and stream for cuda stuff, number of threads for cpu stuff) - add a set of vector structs that use cuda to allocate vectors on the gpu and implement the standard Vector traits (use cuBLAS)
- add a similar set of cuda matrix structs for dense matrices (use cuBLAS)
- add a cuda sparse matric struct (use cuda vectors and implement a CSC matrix)
- add a dense cuda linear solver (use cuSOLVER)
- add a sparse cuda linear solver (use cuSOLVER)
Metadata
Metadata
Assignees
Labels
No labels