- The current implementation of syrk_small computes the entire C matrix rather than computing triangular part. This implementation is not efficient. AMD-Internal: [CPUPL-1571] Change-Id: I9a153207471a55e52634429062d18ba1a225fed9