Startup EnCharge aims to solve the analog compute problem with metal-layer capacitors that help boost energy efficiency and performance.
|
|
Scooped by
Richard Platt
onto Internet of Things - Technology focus June 19, 2025 9:37 PM
|
Your new post is loading...
Analog computing is not a new idea, but the emergence of math-heavy AI workloads in recent years has prompted several startups to build new architectures based on some of the same concepts. In general, the basic operations of multiplication and addition are achieved within a memory array. A memory cell stores a weight, acting as a variable resistor with resistance in some way proportional to the weight value. Data is encoded onto a voltage, which, when supplied to the memory cell, effectively multiplies the data value by the weight value. Output wires are joined together such that currents combine as a simple form of addition. This is a very low-energy way to multiply and add, the two math operations required for matrix multiplication, which form the bulk of AI workloads. Having computation take place in the memory—where the weights are already stored—also means less data movement is needed, which is more energy efficient. Other companies’ analog computing schemes have had various levels of success over the years. Mythic uses an array of Flash memory cells as a matrix multiply accelerator, for example, but this requires complex calibration algorithms for process and temperature variations that can reduce precision. Other types of memory can be used; Tetramem uses RRAM in its memory array. D-Matrix uses modified SRAM for analog multiply, combined with digital addition in its scheme to get around problems with precision and accuracy in all-analog designs.