# GGUF Alignment and Quantization Block Matrix > Source-anchored research for GGUF Alignment and Quantization Block Matrix. Run status: COMPLETED. Parent: [Local Inference Acceleration: Memory Architectures, Metal Kernels, and Latency Optimization](https://llms-explorer.com/tree/local-inference-acceleration-and-kernels/) · 1 facets · 2 facts · page: https://llms-explorer.com/tree/gguf-alignment-and-quantization-block-matrix/ ## Facts and statements - GGUF binary layout enforces 32-byte alignment for all tensor weight arrays, ensuring zero-copy vector instructions (NEON and AVX) execute without unaligned memory access penalties. (high confidence) — [source](https://github.com/ggerganov/llama.cpp/blob/master/docs/gguf.md) - K-quants (Q4_K_M) partition weights into super-blocks of 256 weights containing sub-blocks with independent scales and offsets, preserving perplexity comparable to unquantized FP16 at a fraction of memory bandwidth. (high confidence) — [source](https://github.com/ggerganov/llama.cpp/blob/master/docs/gguf.md)