Binning is the process of sorting manufactured chips by tested performance characteristics (speed, power, features) into different product grades, maximizing revenue from the natural distribution of silicon quality. Binning parameters: (1) Speed grade—maximum operating frequency (e.g., 3.0 GHz, 3.5 GHz, 4.0 GHz bins); (2) Power/leakage—idle and active power consumption; (3) Feature bin—number of working cores, cache size, functional units; (4) Temperature rating—commercial (0-70°C), industrial (-40-85°C), automotive (-40-125°C). How binning works: (1) Wafer sort—probe test identifies functional die and preliminary performance; (2) Package and assemble—good die packaged; (3) Final test—comprehensive speed, power, functionality testing; (4) Bin assignment—each chip assigned to specific product SKU based on test results. Product SKU examples: (1) Highest bin—premium product, highest clock, all cores working, lowest leakage; (2) Mid bin—standard product, moderate clock; (3) Lower bin—value product, some cores disabled, lower clock; (4) Salvage bin—reduced feature set, still functional. CPU example: an 8-core design where 2 cores are defective becomes a 6-core product (e.g., AMD Ryzen 5 from Ryzen 7 die). GPU example: NVIDIA disables streaming multiprocessors to create product stack (RTX 4090 → 4080 → 4070 from same die). Revenue optimization: instead of discarding chips that don't meet top-bin specs, sell as lower-tier products. Yield and binning interaction: as yield improves, more chips qualify for highest bins—binning strategy adjusts accordingly. Dark silicon: intentionally designed spare cores/units anticipating binning. Binning is essential for maximizing revenue from each wafer and creating diverse product portfolios from a single chip design.