{ "schema_version": 1, "builder_plan": { "scenario": "A self-hosted assistant is too expensive because its model uses most of the GPU memory.", "change": "Try the paper's supported lower-precision weights on a copy of your serving setup.", "baseline": "The same model at its original precision, on the same GPU.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Peak memory, task pass rate, latency, and total serving cost at the same context and load.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2609.14864", "title": "GGUF-Metadata Prediction of Single-Sequence llama.cpp Throughput Across Three Systems", "summary": "Replaces empirical per-host throughput benchmarking or naive total-parameter cost estimates with roofline-shaped predictors using quantization-specific scale factors fitted from GGUF metadata. Costs are low in compute and memory, but the method requires reference measurements to fit host-specific efficiency coefficients and adds a modeling layer to the deployment pipeline. Fitted efficiencies do not transfer universally across systems (RTX 5080 test MAPE near 36%), prefill prediction is far less reliable than decode prediction, and quantization ladders reorder across runtime stacks, so choices made from transferred coefficients can be wrong.", "rationale": "The reader can use metadata-based throughput estimates for quantization and model selection on a single GPU or Apple Silicon, but the demonstrated host-specificity of the fitted coefficients means local validation against their own hardware is required before trusting the predictions in production capacity decisions.", "technique": "GGUF-metadata roofline throughput prediction", "family": "quantizacao", "recommendation": "testar", "published": "2026-09-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.14864", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "Replaces empirical per-host throughput benchmarking or naive total-parameter cost estimates with roofline-shaped predictors using quantization-specific scale factors fitted from GGUF metadata. Costs are low in compute and memory, but the method requires reference measurements to fit host-specific efficiency coefficients and adds a modeling layer to the deployment pipeline. Fitted efficiencies do not transfer universally across systems (RTX 5080 test MAPE near 36%), prefill prediction is far less reliable than decode prediction, and quantization ladders reorder across runtime stacks, so choices made from transferred coefficients can be wrong.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.14864", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Active-parameter decode predictor reaches 13.1% MAPE versus 49.4% when charging total parameters on one held-out host set (14.4% vs 55.3% and 36.1% vs 51.9% on the other two); prefill baseline errors reach 108.2% on the RTX 5080.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.14864", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [ { "anchor": "S2.EGx1", "label": "(1)", "section": "2 Predictor", "role": "proposed_method", "latex": "D(c) =S\\frac{P_{a}}{P}+K(c), \\widehat{T}_{d}(c) =\\eta_{h,q}\\frac{B_{h}}{D(c)},", "mathml": "D⁡(c)=S​PaP+K⁡(c),T^d​(c)=ηh,q​BhD⁡(c),", "context": "Let measured Td​(c) and Tp​(c) denote decode and prefill throughput. Let S be GGUF file size, P and Pa total and activated parameter counts, Bh measured bandwidth on host h, and c existing-prefix depth. We model bytes per generated token and decode tokens per second as", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" }, { "anchor": "S2.E3", "label": "(3)", "section": "2 Predictor", "role": "baseline", "latex": "\\widehat{T}_{p}=\\phi_{h,q}\\frac{F_{h}}{2P_{a}},", "mathml": "T^p=ϕh,q​Fh2​Pa,", "context": "We also test the deliberately simple zero-prefix prefill model", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" } ], "equations_status": "selected", "core_kind": "formula", "equations_fetched_at": "2026-10-02", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }