Function dense_gemv_batchm_split 

pub fn dense_gemv_batchm_split(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    input: DevicePtr,
    weight: &DenseWeight,
    output: DevicePtr,
    m: u32,
    y_blocks: u32,
    n: u32,
    k: u32,
    out_stride: u32,
    stream: u64,
) -> Result<()>

2026-09-27: dense_gemv_batchm over m rows split into y_blocks block rows of ceil(m / y_blocks) rows each (at most DENSE_GEMV_BATCHM_MAX_M): more blocks for a narrow weight such as the MoE router, and one launch past 16 rows. Each row’s result is the same as in any other split.