uhetmaniuk

@lucbv If you prefer an enum, then I can change the code accordingly. Let me know. Note that this option is (currently) only propagated up to `KokkosBatched::SerialGemmInternal`. In particular, it...

@e10harvey No problem. I can use the tags inside `struct Trans` (by adding `struct ConjNoTranspose`). It seems to me that the structures `SerialTrmmInternal*` duplicate the code for covering the conjugate...

@e10harvey I keep getting this error when building on weaver with CUDA. ``` ptxas fatal : Unresolved extern function '_ZN13KokkosBatched7details8identityINS_6VectorINS_4SIMDIdEELi8EEEEET_RKS6_' make[2]: *** [perf_test/blas/blas3/CMakeFiles/KokkosBlas3_perf_test.dir/KokkosBlas3_perf_test.cpp.o] Error 255 make[1]: *** [perf_test/blas/blas3/CMakeFiles/KokkosBlas3_perf_test.dir/all] Error 2...

Thanks @e10harvey I ended up using a template based on type. It compiles on weaver (with and without CUDA).