Home

Lambda Wiki

Lambda — AI inference accelerator targeting tapeout on TSMC 16nm through the TSMC University FinFET Program.

Teams

• Inference Software — Model execution stack, runtime, compiler, kernels, quantization, scheduling, and performance optimization.
• Architecture — System architecture, microarchitecture, accelerator datapaths, memory hierarchy, interfaces, and design specifications.
• RTL — SystemVerilog implementation, module hierarchy, coding guidelines, integration, and implementation status.
• Verification — Functional verification, simulation infrastructure, testbenches, assertions, coverage, and regression testing.
• Physical Design — ASIC implementation flow including synthesis, floorplanning, power planning, place-and-route, timing closure, and signoff.