Frank 57264f7faa record performance fixes: FA_ALL_QUANTS build flag (dankserver), RDNA triton FA env (dankclaw) ...

- frank-default-large: GGML_CUDA_FA_ALL_QUANTS=ON is mandatory in the
  FastMTP build; without it q8_0 KV + FA falls back to a slow path and
  prefill collapses ~12x (95 -> 1312 t/s at 8K, measured)
- frank-default-small: FLASH_ATTENTION_TRITON_AMD_ENABLE=TRUE + pure-python
  flash_attn (triton_amd) + stub; without it the vision encoder warmup
  OOMs via torch SDPA

2026-08-17 16:30:22 -07:00

..

Pipeline Parallelism

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

Dankclaw Improvements.md

PP kit + Strix research: Qwen3.8-27B RPC split (Dankclaw head / Dankserver worker)

2026-08-16 05:40:27 -07:00

Dankclaw System Specs.md

docs: measured PP link math (H=5120 verified, 20.5 MB/ubatch wire), NIC inventory

2026-08-17 05:06:41 -07:00

Dankserver System Specs.md

docs: measured PP link math (H=5120 verified, 20.5 MB/ubatch wire), NIC inventory

2026-08-17 05:06:41 -07:00

dankserver-vram-growth.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

deployment.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

DFlash DSpark Speculative Decoding.md

docs: boot policy (only PP units enabled), deploy.sh dankserver hazard, stale-claim cleanup

2026-08-17 03:30:00 -07:00

FastFlowLM NPU Inference.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

frank-default-large-benchmark.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

frank-default-small-benchmark.md

docs: record dual-f16 KV refit on frank-default-small (2026-08-14)

2026-08-14 12:59:10 -07:00

frank-redo-2026-08-17.md

record performance fixes: FA_ALL_QUANTS build flag (dankserver), RDNA triton FA env (dankclaw)

2026-08-17 16:30:22 -07:00

Hybrid CPU GPU Processing Dankserver.md

docs: polish pass — fix stale names, duplicates, branding, and README

2026-08-10 21:12:23 -07:00

index.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

ling-3.0-flash-trial.md

docs: boot policy (only PP units enabled), deploy.sh dankserver hazard, stale-claim cleanup

2026-08-17 03:30:00 -07:00

ling-3.0-flash-vllm-plan.md

docs: boot policy (only PP units enabled), deploy.sh dankserver hazard, stale-claim cleanup

2026-08-17 03:30:00 -07:00

Quantization by Hardware Architecture.md

docs: reflect Ling-3 vLLM (frank-default-large) on Dankclaw

2026-08-10 23:37:27 -07:00

qwen3.8-spec-matrix.md

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

qwen3.8-strix-optimizations.md

PP kit + Strix research: Qwen3.8-27B RPC split (Dankclaw head / Dankserver worker)

2026-08-16 05:40:27 -07:00

RAM CPU Offloading in VLLM.md

docs: polish pass — fix stale names, duplicates, branding, and README

2026-08-10 21:12:23 -07:00

Running Qwen3.6 27B FP8 on Dankclaw.md

docs: reflect Ling-3 vLLM (frank-default-large) on Dankclaw

2026-08-10 23:37:27 -07:00

Running Qwen3.6 35B A3B FP8 on Dankserver.md

docs: boot policy (only PP units enabled), deploy.sh dankserver hazard, stale-claim cleanup

2026-08-17 03:30:00 -07:00

Shared Memory Architecture Dankclaw.md

docs: polish pass — fix stale names, duplicates, branding, and README

2026-08-10 21:12:23 -07:00

VLLM Index.md

docs: polish pass — fix stale names, duplicates, branding, and README

2026-08-10 21:12:23 -07:00

VLLM Overview.md

docs: polish pass — fix stale names, duplicates, branding, and README

2026-08-10 21:12:23 -07:00

vllm_research_findings.md

feat: initial vLLM Qwen3.6-35B-A3B-FP8 serve script with DFlash drafting on dankserver

2026-08-07 18:28:38 -07:00