search

jam31118 / cu-tridiag

A set of tridiagonal solver routines using CUDA
0 stars 0 forks source link

use gtsv2() rather than just gtsv() #1

Closed jam31118 closed 5 years ago

jam31118 commented 5 years ago

This is for controlling concurrency on the user side since gtsv2() allows user to do memory allocation and deallocation.

jam31118 commented 5 years ago

Implemented at 4047162cc85e839f5560d3378404e20cf57ba963, closing this issue.