A flip-flop can store 1 bit of information: either it’s at H or it’s at L. Accordingly, SRAM can be implemented by arraying together a large number of flip-flops and adding the necessary select capability. Because SRAM works much faster than DRAM and flash memory, it is commonly used to build CPU caches and registers.
In practice, CPU memory and registers do not use logic gates—such as those built with RS flip-flops—because the circuitry gets too big. This memory is typically implemented using four or six FETs per bit.