q4k_mmq_gemm

spark_model::layers::ops

Function q4k_mmq_gemm 

Source 
pub fn q4k_mmq_gemm(
    gpu: &dyn GpuBackend,
    kernel_nc: KernelHandle,
    kernel_wc: KernelHandle,
    a_q8: DevicePtr,
    w_q4k: DevicePtr,
    out_bf16: DevicePtr,
    m: u32,
    n: u32,
    k: u32,
    stream: u64,
) -> Result<()>
Expand description

Q4_K MMQ GEMM: C[m,n] (bf16) = A_q8[m,k] x W_q4k[n,k]. Fused bf16 store.