Skip to main content

emit_int8_gemm

logicaffeine_compile::codegen_cuda::matrix

Function emit_int8_gemm 

Source 
pub fn emit_int8_gemm(
    m: usize,
    k: usize,
    n: usize,
    weights: &[i32],
    acts: &[i32],
) -> Result<CudaGemmArtifact, CudaLowerError>
Expand description

Emit a DP4A int8 GEMM CUDA program for O(m×n) = A(m×k) · W(k×n), with weights (k*n row-major, W[r][c] = weights[r*n+c]) and acts (m*k row-major) baked in. Refuses (typed) if the RDU-shaped target does not advertise int8 weight-stationary GEMM — the exact certified gate the fabric uses.