Exploiting matrix symmetry to halve memory footprint offers an opportunity for accelerating memory-bound computations like Sparse Matrix-Vector Multiplication (SpMV). However, symmetric SpMV incurs ...