Recipe
qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1
qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Engine version
- lucebox-hub b6ffab4a9
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 1,024
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/Qwen/Qwen3.6-27B- Repository
- Qwen/Qwen3.6-27B
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Candidate: useful compatibility or speed evidence. The registry does not offer Run until promotion requirements are met.
Observed configuration
llama.cpp
Evidence only · candidate · reference
Candidate evidence — not a Run contract
Observed source tokens
Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.
build/test_dflashQwen3.6-27B-Q4_K_M.ggufz-lab/Qwen3.6-27B-DFlash--fast-rollback--ddtree--ddtree-budget=22--max-ctx=1024
| Flag | Value |
|---|---|
--ddtree-budget | 22 |
--max-ctx | 1024 |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 1,024 | — | 140.3 | 710 | observed | qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 1
- hardware id
- rtx-3090-ti-24gb
- id
- qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1
- model instance id
- qwen-qwen3-6-27b--q4-k-m
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- lucebox-hub b6ffab4a9
serving
- max concurrency
- 1
- max context tokens
- 1,024
- tensor parallel
- 1
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-30T09:10:02Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-30T09:10:02Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
metadata
- localmaxxing · backend
- cuda
- localmaxxing · hardware label
- RTX 3090 Ti
- localmaxxing · notes
- DFlash speculative decoding via Lucebox-Hub test_dflash binary (custom C++/CUDA on top of Luce-Org/llama.cpp@luce-dflash ggml fork). Block-diffusion draft from z-lab/Qwen3.6-27B-DFlash (BF16 safetensors, 3.46 GB, 2026-04-26 z-lab snapshot). DDTree tree-verify budget=22, chain_seed=1, fa_window=2048, KV f16. Bench: scripts/bench_he.py --n-gen 512 --ddtree-budget 22, mean of 10 HumanEval-style code-completion prompts (single-stream, batch=1). Per-prompt range 75.9-203.4 tok/s; peak sum_product 203.37 tok/s. Mean acceptance length 9.22, accept rate 57.7%. AR baseline same binary on same hardware: 35.79 tok/s (3.92x speedup). vLLM FP8 production single-stream on this rig: 44.36 tok/s (3.16x speedup). Critical CMake flags: -DCMAKE_CUDA_ARCHITECTURES=86 -DGGML_CUDA_GRAPHS=ON -DCMAKE_BUILD_TYPE=Release. Defaults compile sm_52 + graphs OFF, costs 3x.
- localmaxxing · observed command
- build/test_dflash Qwen3.6-27B-Q4_K_M.gguf z-lab/Qwen3.6-27B-DFlash --fast-rollback --ddtree --ddtree-budget=22 --max-ctx=1024
- localmaxxing · run id
- cmogqkco20007l704rhj8sq83
- localmaxxing · tokenized · arguments
- build/test_dflash, Qwen3.6-27B-Q4_K_M.gguf, z-lab/Qwen3.6-27B-DFlash, --fast-rollback, --ddtree, --ddtree-budget=22, --max-ctx=1024
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-recipe | www.localmaxxing.com/en/runs/cmogqkco20007l704rhj8sq83 ↗ |