← Back to C-Kernel-Engine Docs Doxygen Source Documentation
fused Directory Reference

Files

file  attention_mlp_fused.c [code]
 Mega-Fused Attention + MLP Block.
 
file  fused_rmsnorm_linear.c [code]
 Fused RMSNorm + Linear (GEMV) kernel.
 
file  gemv_fused_quant_bias.c [code]
 Fused GEMV kernels with online quantization and bias.
 
file  mega_fused_attention_avx.c [code]
 Mega-Fused Attention for AVX (256-bit) and AVX-512 (512-bit)
 
file  mega_fused_attention_decode_q5_0.c [code]
 Mega-fused attention decode with Q5_0 weights.
 
file  mega_fused_attention_decode_q5_0.h [code]
 Mega-fused attention decode with Q5_0 weights - Header.
 
file  mega_fused_attention_prefill.c [code]
 Mega-fused prefill attention kernel.
 
file  mega_fused_attention_prefill_q8_0.c [code]
 Mega-fused prefill attention kernel with Q8_0 out-proj.
 
file  mega_fused_outproj_mlp_prefill.c [code]
 Mega-fused post-attention block for prefill.
 
file  prefill_fused_gemm.c [code]
 Fused kernels for prefill phase with proper 2D tiling.
 
file  rmsnorm_q8_k_fused.c [code]
 Fused RMSNorm + Q8_K Quantization kernel.
 
file  rmsnorm_qkv.c [code]
 Fused RMSNorm + QKV Projection.