A cache memory apparatus includes an L1 cache memory, an L2 cache memory coupled to the L1 cache memory, an arithmetic logic unit (ALU) within the L2 cache memory, the combined ALU and L2 cache memory being configured to perform therewithin at least one of: an arithmetic operation, a logical bit mask operation; the cache memory apparatus being further configured to interact with at least one processor such that atomic memory operations bypass the L1 cache memory and go directly to the L2 cache memory.