pub fn moe_weighted_sum_blend_prefill(
gpu: &dyn GpuBackend,
kernel: KernelHandle,
output: DevicePtr,
expert_out: DevicePtr,
expert_weights: DevicePtr,
shared_out: DevicePtr,
input: DevicePtr,
gate_weight: DevicePtr,
hidden: u32,
top_k: u32,
k: u32,
num_tokens: u32,
stream: u64,
) -> Result<()>Expand description
Fused weighted sum + sigmoid blend for N-token prefill.
Grid: (ceil(hidden/256), num_tokens, 1) Block: (256, 1, 1)