moe_silu_mul

Function moe_silu_mul 

Source
pub fn moe_silu_mul(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    gate: DevicePtr,
    up: DevicePtr,
    output: DevicePtr,
    total_elements: u32,
    stream: u64,
) -> Result<()>
Expand description

Element-wise SiLU activation + multiply: output[i] = silu(gate[i]) * up[i].

Grid: (ceil(total_elements/256), 1, 1) Block: (256, 1, 1)