https://aiinfra.blog/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/about/https://aiinfra.blog/blog/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-00-overview/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-01-brrrr-three-bottlenecks/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-02-transformer-parts-and-parameters/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-03-self-attention-and-autoregression/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-04-flops-and-kv-cache/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-05-memory-budget-and-roofline/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-06-week1-review-and-quiz/2026-09-04T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-07-colab-first-inference/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-08-cuda-async-and-timing/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-09-ttft-tpot-vs-theory/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-10-torch-profiler-perfetto/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-11-gaps-overhead-bound/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-12-week2-review/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-13-measure-bandwidth/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-14-measure-peak-flops/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-15-draw-your-roofline/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-16-batch-sweep/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-17-nominal-vs-measured/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-18-week3-review/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-19-runpod-vast-first-instance/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-20-persistence/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-21-bootstrap-script/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-22-three-layer-auto-shutdown/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-23-cost-dashboard/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-24-week4-review/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-25-gpu-anatomy/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-26-number-formats/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-27-cuda-execution-model/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-28-reading-spec-sheets/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-29-market-check/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/aiinfra-365/day-30-month1-review/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/blog/decode-is-memory-bound/2026-08-15T00:00:00.000Zhttps://aiinfra.blog/course/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/tags/2026-09-05T00:00:00.000Zhttps://aiinfra.blog/tags/a100/https://aiinfra.blog/tags/aiinfra-365/https://aiinfra.blog/tags/automation/https://aiinfra.blog/tags/bash/https://aiinfra.blog/tags/benchmark/https://aiinfra.blog/tags/colab/https://aiinfra.blog/tags/cost/https://aiinfra.blog/tags/cuda/https://aiinfra.blog/tags/decode/https://aiinfra.blog/tags/flops/https://aiinfra.blog/tags/kernel/https://aiinfra.blog/tags/kv-cache/https://aiinfra.blog/tags/nvidia-smi/https://aiinfra.blog/tags/perfetto/https://aiinfra.blog/tags/profiler/https://aiinfra.blog/tags/quiz/https://aiinfra.blog/tags/review/https://aiinfra.blog/tags/roadmap/https://aiinfra.blog/tags/roofline/https://aiinfra.blog/tags/runpod/https://aiinfra.blog/tags/t4/https://aiinfra.blog/tags/tensor-core/https://aiinfra.blog/tags/timing/https://aiinfra.blog/tags/transformer/https://aiinfra.blog/tags/vast/