https://akiyangdev.github.io/llm-inference-notes/https://akiyangdev.github.io/llm-inference-notes/articles/https://akiyangdev.github.io/llm-inference-notes/articles/ascend/deepseek-v4-w8a8-ascend-910c/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/ascend-dynamic-eplb-weight-migration/2026-10-24T11:57:30+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/deepseek-v4-dspark-parallel-layout-analysis/2026-10-24T13:53:26+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/deepseek-v4-speculative-distributed-consistency-source-analysis/2026-10-24T13:53:26+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/sglang-collectives-deepseek-v4/2026-10-24T14:23:26+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/sglang-rank-group-deepseek-v4/2026-10-24T14:23:26+08:00https://akiyangdev.github.io/llm-inference-notes/articles/distributed/why-large-models-need-multi-card-tp-dp-ep/2026-10-24T14:35:40+08:00https://akiyangdev.github.io/llm-inference-notes/articles/fundamentals/ai-infra-working-glossary/2026-10-24T13:21:06+08:00https://akiyangdev.github.io/llm-inference-notes/articles/fundamentals/prefill-vs-decode-performance/2026-10-25T09:26:31+08:00https://akiyangdev.github.io/llm-inference-notes/articles/fundamentals/sglang-decode-source-reading-primer/2026-10-24T13:42:31+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/128k-prefix-cache-benchmark-calibration/2026-10-24T10:43:21+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/128k-prefix-cache-remaining-costs/2026-10-24T11:42:32+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ai-driven-inference-autotuning-decision-loop/2026-10-21T19:50:15+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ai-driven-inference-autotuning-execution-and-evidence/2026-10-21T19:50:15+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/aisbench-sglang-first-request-to-report/2026-10-24T11:23:08+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/aisbench-whole-native-tpot-itl/2026-10-24T11:59:50+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ascend-910c-cp2-vs-cp4-prefill-throughput/2026-10-24T11:14:20+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ascend-910c-inference-tuning-effective-optimizations-and-costs/2026-10-24T10:11:05+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ascend-910c-inference-tuning-negative-results-and-stop-rules/2026-10-24T09:53:42+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ascend-910c-prefill-scenario-static-eplb/2026-10-24T11:57:30+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/ascend-950dt-single-node-pd-optimization/2026-10-17T16:41:23+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/decode-fast-throughput-limited-by-prefill/2026-10-25T09:26:31+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-indexcache-freq2-cross-layer-topk-reuse/2026-10-24T10:26:35+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-prefill-chunk-size-selection/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-pro-32k-prefix-cache-concurrency-tuning/2026-10-17T16:41:23+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-speculative-ascend-performance-critical-path-analysis/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-speculative-decoding-ascend-profiling/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-speculative-performance-analysis/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/deepseek-v4-w8a8-ascend-910c-performance/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/expert-count-export-hccl-gloo/2026-10-24T10:58:45+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/inference-tuning-workflow-postmortem/2026-10-21T20:08:36+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/npu-timeline-gaps-and-recoverable-speedup/2026-10-24T10:56:12+08:00https://akiyangdev.github.io/llm-inference-notes/articles/performance/sglang-stream-acceptance-cache-timing/2026-10-24T11:45:18+08:00https://akiyangdev.github.io/llm-inference-notes/articles/pr-reviews/sglang-pr-39120-tensor-view-storage-cache/2026-10-17T16:41:23+08:00https://akiyangdev.github.io/llm-inference-notes/articles/pr-reviews/sglang-pr-39871-attention-parallel-widths/2026-10-17T16:41:23+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/deepseek-v4-attention-source-analysis/2026-10-24T12:22:56+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/deepseek-v4-distributed-parallel-source-analysis/2026-10-24T12:50:24+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/deepseek-v4-moe-source-analysis/2026-10-24T12:22:56+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/deepseek-v4-pd-disaggregation-source-analysis/2026-10-24T14:35:40+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/gsm8k-accuracy-debugging-from-200-to-full-evaluation/2026-10-24T14:48:10+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/sglang-ascend-request-lifecycle/2026-10-25T09:26:31+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/sglang-modelrunner-source-analysis/2026-10-24T11:41:14+08:00https://akiyangdev.github.io/llm-inference-notes/articles/sglang/sglang-scheduler-source-analysis/2026-10-24T11:41:14+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-accept-decision-source-analysis/2026-10-24T14:48:10+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-dspark-adaptive-verify-scheduler/2026-10-24T14:58:59+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-eagle-draft-tree-source-analysis/2026-10-24T15:53:56+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-mtp-speculative-decoding-source-analysis/2026-10-24T16:03:42+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-speculative-decoding-source-analysis/2026-10-24T16:34:56+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-speculative-kv-commit-compact-rollback-source-analysis/2026-10-24T17:04:48+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/deepseek-v4-tree-attention-source-analysis/2026-10-24T17:36:48+08:00https://akiyangdev.github.io/llm-inference-notes/articles/speculative-decoding/sglang-speculative-decoding-scheduler-source-analysis/2026-10-24T11:22:18+08:00https://akiyangdev.github.io/llm-inference-notes/credits/https://akiyangdev.github.io/llm-inference-notes/guides/https://akiyangdev.github.io/llm-inference-notes/levels/advanced/https://akiyangdev.github.io/llm-inference-notes/levels/deep/https://akiyangdev.github.io/llm-inference-notes/levels/entry/https://akiyangdev.github.io/llm-inference-notes/levels/reference/https://akiyangdev.github.io/llm-inference-notes/series/https://akiyangdev.github.io/llm-inference-notes/skills/https://akiyangdev.github.io/llm-inference-notes/skills/ai-infra-technical-writer/https://akiyangdev.github.io/llm-inference-notes/skills/sglang-pr-reviewer/https://akiyangdev.github.io/llm-inference-notes/topics/ascend/https://akiyangdev.github.io/llm-inference-notes/topics/distributed/https://akiyangdev.github.io/llm-inference-notes/topics/fundamentals/https://akiyangdev.github.io/llm-inference-notes/topics/performance/https://akiyangdev.github.io/llm-inference-notes/topics/pr-reviews/https://akiyangdev.github.io/llm-inference-notes/topics/sglang/https://akiyangdev.github.io/llm-inference-notes/topics/speculative-decoding/