In-Memory Computing & Emerging Memory
Compute-in-memory architectures and emerging nonvolatile devices that reduce data movement or use memory-device physics directly in computation.
Evaluate these technologies as mixed device/architecture systems: accuracy, endurance, variability, converters, peripheral circuits, mapping and system-level data movement all matter.
Architecture is stable; technology maturity is not assumed. Public claims are anchored to current authoritative sources and are separated from Aurora concept work.
Technical coverage
Six focused routes keep device physics, interfaces, packaging, controls and verification separable.
Analog In-Memory Compute
Crossbar matrix-vector operations, mixed-signal precision and peripheral overhead.
Open technical route →
Phase-Change Memory
PCM device physics, analog states, drift, programming and endurance.
Open technical route →
ReRAM / Memristive Devices
Resistive switching, crossbars, variability and selector requirements.
Open technical route →
MRAM & Spintronics
Magnetic state, nonvolatility, switching mechanisms and embedded use.
Open technical route →
Near-Memory / Digital CIM
SRAM/DRAM-adjacent compute, digital compute-in-memory and dataflow.
Open technical route →
Heterogeneous AI Acceleration
Combining analog/digital accelerators with CPUs/GPUs and software mapping.
Open technical route →In-Memory Computing & Emerging Memory system view
Evaluate these technologies as mixed device/architecture systems: accuracy, endurance, variability, converters, peripheral circuits, mapping and system-level data movement all matter.