thead_sdgemm_lmul4

Took the original rv64iv assembly kernel for sdgemm operation and adapted it to use register grouping over 4 registers setting LMUL=4

thead_conversion

Conversion of the original rv64iv kernels in order to be compiled with GCC 14 with -march=rv64gc_xtheadvector