llama.cpp
e22cd0aa - metal : extend mul_mv_ext to BF16, Q2_K, Q3_K (#20250)

Commit
14 hours ago
metal : extend mul_mv_ext to BF16, Q2_K, Q3_K (#20250) Enable mul_mv_ext small-batch kernels (BS 2-8) for BF16, Q2_K, and Q3_K quantization types. These types previously fell through to the slower single-row mul_mv path. BF16 uses the float4 dequantize path (like F16). Q2_K and Q3_K use the float4x4 K-quant path (like Q4_K/Q5_K/Q6_K). Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Author
Parents
Loading