Records

records/

• records/2026-08-26-01-handoff-llm-integration.md — Record: LLM-integration handoff — 2026-08-26 — read
• records/2026-08-26-02-dsh-provider-findings.md — Record: DSH provider findings — 2026-08-26 — read
• records/2026-08-26-02-findings-and-reorder.md — Record: Findings and task reorder — 2026-08-26 — read
• records/2026-08-26-03-cache-discard.md — Record: Cache discard — 2026-08-26 — read
• records/2026-08-26-04-tuning-applied.md — Record: Tuning applied — 2026-08-26 — read
• records/2026-08-26-05-version-review.md — Record: llama.cpp version review — 2026-08-26 — read
• records/2026-08-26-06-dsh-and-cache-reuse-applied.md — Record: DSH settings and cache-reuse applied — 2026-08-26 — read
• records/2026-08-26-07-copilot-removed.md — Record: Copilot extension removed — 2026-08-26 — read
• records/2026-08-26-08-minerva-deployed-and-athena-dedup.md — Record: Minerva deployed; athena service deduped — 2026-08-26 — read
• records/2026-08-26-09-kv-quant-analysis.md — Record: KV-cache quantization analysis — 2026-08-26 — read
• records/2026-08-26-10-kv-quant-applied.md — Record: KV-cache quantization applied — 2026-08-26 — read
• records/2026-09-02-01-mi25-vulkan-device-lost-tuning.md — Record: MI25 Vulkan device-lost — GPU-preset stability tuning — 2026-09-02 — read
• records/2026-09-02-02-llamacpp-rebuild-qwen38.md — Record: llama.cpp rebuild on beast (build 10790) + Qwen3.8-Flash-Next (qwen38) CPU preset — 2026-09-02 — read
• records/2026-09-03-01-athena-gpu-fixes-qwen27-llamacpp-current.md — Record: athena GPU fixes + qwen27 preset; llama.cpp current on both hosts — 2026-09-03 — read
• records/2026-09-03-02-athena-qwen27-load-fixes.md — Record: athena qwen3_5/qwen27 load fixes — source sync to 10802, repo mix-up, VRAM contention — 2026-09-03 — read
• records/2026-09-03-03-beast-tuning-verification.md — Record: beast per-model tuning verification + op-offload discovery — 2026-09-03 — read
• records/2026-09-03-04-cpu-gpu-router-split-and-cache-trim.md — Record: beast cache trim + CPU/GPU router split (VRAM guard) — 2026-09-03 — read
• records/2026-09-03-05-gpu-context-parallel-fix.md — Record: GPU-router context fix — parallel 1 + gemma-26b large context restored — 2026-09-03 — read
• records/2026-09-10-01-llamacpp-mmap-preset-key-removed.md — Record: llama.cpp 0.4.0 removed the `mmap` preset key — 2026-09-10 — read
• records/2026-09-21-01-mi25-inference-benchmarks.md — Record: MI25 inference benchmarks on beast — 2026-09-21 — read
• records/2026-09-25-01-llama-vscode-config-applied.md — Record: llama.vscode three-tier configuration applied — 2026-09-25 — read
• records/2026-09-25-02-mi25-rebenchmark-11193.md — Record: MI25 re-benchmark on llama.cpp 11193 — 2026-09-25 — read
• records/2026-09-26-01-llama-vscode-router-fim-incompatibility.md — Record: llama.vscode FIM is incompatible with router mode — 2026-09-26 — read
• records/2026-10-15-01-therock-rocm-gfx900-mi25.md — TheRock ROCm on beast.lan — gfx900 (AMD Instinct MI25) — read
• records/2026-10-22-01-rules-adoption.md — Record: Adopt the shared rules — read
• records/README.md — README for records — read