In-memory computing is a new way of computing that aims to solve the memory latency problem. As the name suggests, in-memory computing enables the system memory to do some calculations the CPU would do otherwise, cutting down the amount of data that must be transferred between the CPU and DRAM.

Samsung and TSMC are actively working on memory capable of doing this, featuring MRAM memory cells. In-memory computing is still in the prototype phase, but progress is being made on the hardware side to make it a viable technology. With the help of conversion layers like PyPIM, software should be developed to support this computing method.