{"id":"rack-dynamic-power","level":"infrastructure","name":"Dynamic load, reserve and power control","shortName":"Dynamic load, reserve and power control","role":"Accounts for changes in electrical demand over timescales from chip activity to facility operation.","description":"AI workloads alternate between arithmetic, communication, memory movement and idle intervals. Their electrical demand changes accordingly, and synchronized jobs can align changes across many devices. Power supplies, local decoupling, rack distribution and facility controls respond on different timescales. A daily energy total cannot reveal a fast voltage event, while a brief peak does not define annual energy. Power capping or workload scheduling can limit demand but may change completion time or service latency. The IEA discusses emerging power-system challenges from AI growth; a local engineering model must still use measured waveforms and the actual facility response.","specs":[{"label":"Timescale","value":"Fast transient, workload interval and billing period","context":"Measurements need sufficient temporal resolution for the question."},{"label":"Reserve","value":"Available response beyond committed demand","context":"Specify electrical capacity, energy duration and control latency separately."}],"interfaces":[{"name":"Synchronized workload activity","direction":"in","signal":"Time-varying current and power requests"},{"name":"Power-control response","direction":"out","signal":"Regulated supply, caps or shifted work"}],"tradeoff":"Power limits can improve electrical predictability and capacity sharing while reducing peak throughput or extending job duration.","whyAI":"Useful AI work depends on surviving demand changes without excessive throttling, errors or interruptions.","related":["power-integrity-budget","ups","scheduler"],"sources":[{"title":"Key Questions on Energy and AI: Executive Summary","url":"https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary"},{"title":"NVIDIA DGX GB Rack Scale Systems: Hardware","url":"https://docs.nvidia.com/dgx/dgxgb200-user-guide/hardware.html"}],"scale":"Chip voltage rails to facility heat rejection","kind":"architecture","branch":"Data-center physical systems","subbranch":"Electricity and heat transport","mechanism":["Record load at the timescale needed to observe the suspected event.","Separate local voltage dynamics from rack and facility power changes.","Estimate how many jobs can align their demand and what reserve can respond in time.","Apply caps or scheduling controls and measure their effect on useful results.","Integrate power over time for energy accounting without losing peak and recovery information."],"misconception":"An average megawatt figure cannot prove adequate transient response, and a short peak is not a continuous energy demand.","check":{"question":"What additional quantity is needed to turn power into energy?","options":["The duration or time history","Only the GPU model name","The number of Ethernet ports"],"answer":0,"explanation":"Energy integrates power over an interval, so both demand and time matter."},"science":[{"title":"Energy integral","equation":"E = ∫P(t)dt","explanation":"With P in MW and time in hours, E is MWh.","assumptions":"Metering boundary remains fixed; coarse samples can miss peaks even when the integral is approximately correct."}],"prerequisites":["power-supply","scheduler"],"learningObjective":"Match power measurements and controls to their relevant timescales.","workstream":"W09","visualFamily":"signal-power","engineeringExample":"Synthetic workload draws 1 MW for ten minutes and 0.5 MW for twenty minutes. It uses about 0.333 MWh over the half hour, while its peak remains 1 MW rather than the 0.667 MW average.","evidenceNotes":"No universal AI load waveform is assumed. Real simultaneous behavior and control response need facility and device telemetry.","labId":"library","scope":"No current calculator directly models this lesson’s main mechanism. Use the library for related bounds while retaining their stated omissions.","investigationId":"manufacturing","labScope":"No current calculator directly models this lesson’s main mechanism. Use the library for related bounds while retaining their stated omissions."}