This visualizer demonstrates a 3D heterogeneous photonics compute stack using six materials:
S (Silicon): Primary photonic waveguides and compute cores. Each core processes optical signals.Ti (Titanium): Through-silicon vias (TSVs) enabling vertical photon/electron transfer between layers.Cu (Copper): Thermal heat spreaders and bulk interconnects for power delivery.Zn (Zinc): Buffer/adhesion layers between heterogeneous materials.Ag (Silver): High-reflectivity mirrors for photon routing within layers.Au (Gold): Contact pads and I/O bonding for external interface.Progressive stacking scales compute density:
10x10x10 = 1,000 cores (3D Base Grid) 10x10x10x10 = 10,000 cores (4D Stack - 10 layers) 10x10x10x10x10 = 100,000 cores (5D Stack - 100 layers)
Each layer contains a 10x10 sub-grid. Layers are connected via Ti vias that route photons vertically with propagation delay based on layer distance.
Titanium vias introduce thermal bottlenecks. Thermal resistance per via scales with:
Rth = (L / (k * A)) * N_layers Where: L = Via length (50μm per layer) k = Ti thermal conductivity (21.9 W/m·K) A = Via cross-section (10μm²) N_layers = Number of stacked layers Typical values: 1 Layer: Rth ≈ 0.23 K/W per via 10 Layers: Rth ≈ 2.3 K/W per via 100 Layers: Rth ≈ 23 K/W per via
Higher Rth increases junction temperature and limits clock frequency. Pressure control can modulate via contact resistance.
Each layer computes a D value representing diffusion/dispersion:
D[layer] = Σ(active_cores) * pressure_factor / (1 + Rth * 0.01) Where pressure_factor = 1 + (pressure_GPa * 0.05)
OUT Σ aggregates all active photon pulses across layers. IN pulse injects 100 units. O1 readout samples current OUT state. 10kHz clock drives synchronous pulse propagation through vias.