Sources: solar, wind, motors, grid. Key parameters: generator capacity, variability curves, conversion efficiency.
Models: Li-ion, solid-state, supercapacitors. Parameters: capacity, state-of-charge, recharge efficiency, lifecycle degradation.
Loads: mechanical, electrical, electronic (CPUs, GPUs, memory). Parameters: power states, transition costs, per-task energy cost.
Features: DVFS, power gating, shutdown, redundancy control.
Tracks Joules/sec per subsystem, ambient profiles, cooling methods, and system temperature.
VisualSim delivers accurate power estimates by modeling full system architecture and runtime behavior—rather than relying on static equations or spreadsheet activity factors.
Every component (processor, memory, interconnect, power amplifier, antenna, cable and network interface) shifts between states—active, idle, sleep, or wait etc—triggered by real events such as software task, interrupt, external data movement, DMA operation or network activity. The power for each state, transition between the states and the time before transition can be accessed from a Table or from an external Solver.
This produces a time-accurate power activity tied to the proposed workloads. Power becomes an emergent outcome of complete system behavior, not a separate calculation.
Power is computed from actual workloads and task activity, not from statistical assumptions.
Each component’s power is tied to accurate timing of transactions, memory accesses, and clock cycles.
The framework tracks voltage droops, jitter, and temperature rise caused by activity spikes and load changes.
The system simulates thermal feedback, battery degradation, generator response, and shock-induced stress, which analytical models ignore.
Power, performance, and functional behavior are evaluated together, ensuring consistency across domains.
VisualSim’s IP libraries are verified against lab data, timing diagrams, and silicon measurements to maintain accuracy across technology generations.
