daisytuner / sdfglib / 21394710662

push

github

Cublas gemm + Python integration (#477)

* cublas GEMM dispatcher draft

* blas2cublas pass draft

* small fixes

* correct cublas gemm

* optional error checks

* fix kernel without transfers

* remove duplicate kernel definition

* rename blas2cublass pass

* max and ceil grid fix

* Add guards for 0 dimensions

* malloc correct size

* add missing transfers

* update test

* cuda include path

* use cuda path

* always link cuda for now

57 of 310 new or added lines in 7 files covered. (18.39%)

21671 of 33125 relevant lines covered (65.42%)

394.07 hits per line