# Importance-matrix presence as a second axis in equal-size KLD comparisons > The covered fact is the llama.cpp scoreboard; the one thing absent is that vendor tables mix the two axes silently, shown below. Parent: [Mac local LLMs: Quantization evaluation](https://llms-explorer.com/tree/mac-local-llms-quantization-evaluation/) · 1 facets · 8 facts · page: https://llms-explorer.com/tree/importance-matrix-presence-as-a-second-axis-in-e/ ## Facts - The covered fact is the llama.cpp scoreboard; the one thing absent is that vendor tables mix the two axes silently, shown below. — source: `asserted` - Unsloth's Qwen3.5-35B-A3B table lists Quantizer, Quant Level, Disk Space, PPL, KLD 99.9% and Mean KLD, with no imatrix column, so rows from different vendors cannot be split into recipe effect and imatrix effect from the table alone. — [source](https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks) - The same page says imatrix "definitely" lowers KLD and PPL at 5-10% slower inference, then attributes the 5-10% slowdown to I-quants (iq3_xxs, iq2_s) in its speed table; the two statements disagree about the cause. — [source](https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks) - Unsloth (imatrix helps most at low bits, works on all widths) versus the llama.cpp README (no consistent gain from more imatrix tokens, and the benefit measured on Wikitext). Not averaged; the corpora differ. — [source](https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks) - A same-recipe, same-size pair (one GGUF type, with and without imatrix, imatrix on a held-out corpus versus in-domain) scored on Qwen3.5-class MoE: none found. — source: `asserted` - Already covered by size-matched-kld-reporting-for-local-quants.md (imatrix as a second axis at fixed size, with the q4_K_S and q2_K numbers) and unsloth-dynamic-gguf-methodology.md; nothing in them is contradicted. — source: `asserted` - Unsloth's Qwen3.5-35B-A3B full-benchmark table has no per-row imatrix status column. — [source](https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks) - The Unsloth page attributes a 5-10% inference slowdown to imatrix in a bullet and to I-quant types in its speed table. — [source](https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks)