Function moe_expert_silu_down_shared_fp8 

pub fn moe_expert_silu_down_shared_fp8(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    gate_out: DevicePtr,
    up_out: DevicePtr,
    down_weight_ptrs: DevicePtr,
    down_scale_ptrs: DevicePtr,
    output: DevicePtr,
    expert_indices: DevicePtr,
    sh_gate_in: DevicePtr,
    sh_up_in: DevicePtr,
    sh_down: &Fp8Weight,
    sh_down_out: DevicePtr,
    n: u32,
    k: u32,
    top_k: u32,
    stream: u64,
) -> Result<()>

Fused SiLU+down expert GEMV with shared expert for FP8 weights.

FP8 variant of moe_expert_silu_down_shared: uses 2 pointer tables (weight_ptrs + scale_ptrs) instead of NVFP4’s 3. Shared expert down weight passed as direct Fp8Weight pointer.

Grid: (ceil(N/8), top_k+1, 1) Block: (128, 1, 1)