Skip to main content

emit_wmma_cpasync_program

logicaffeine_compile::codegen_cuda::gemm_tile

Function emit_wmma_cpasync_program 

Source 
pub fn emit_wmma_cpasync_program(
    kind: MmaKind,
    tile: &WmmaTile,
    stages: usize,
    m: usize,
    k: usize,
    n: usize,
) -> String
Expand description

Checksum-matching cp.async GEMM program for a tensor-core kind (bf16/f16) — Σ C ≡ the scalar oracle on int8-range operands (exact in bf16/f16, wide f32 accumulate). Launches the double-buffered gemm_wmma_cpasync.