Function gdn_prefill_persistent_smem_batched 

pub fn gdn_prefill_persistent_smem_batched(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    h_state_ptrs: DevicePtr,
    query: DevicePtr,
    key: DevicePtr,
    value: DevicePtr,
    gate: DevicePtr,
    beta: DevicePtr,
    output: DevicePtr,
    batch_size: u32,
    seq_len: u32,
    num_k_heads: u32,
    num_v_heads: u32,
    k_dim: u32,
    v_dim: u32,
    qk_stride: u32,
    v_stride: u32,
    gb_stride: u32,
    smem: u32,
    stream: u64,
) -> Result<()>

Batched WY32 persistent GDN prefill (uses gated_delta_rule_prefill_wy64_batched).