Function moe_expert_gate_up_act_nvfp4_grouped 

pub fn moe_expert_gate_up_act_nvfp4_grouped(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    geometry: Fp8GroupedGeometry,
    input: DevicePtr,
    gate: Nvfp4ExpertTables,
    up: Nvfp4ExpertTables,
    act: DevicePtr,
    expert_offsets: DevicePtr,
    sorted_token_ids: DevicePtr,
    active_experts: DevicePtr,
    active_count: DevicePtr,
    sh_gate: &QuantizedWeight,
    sh_up: &QuantizedWeight,
    sh_act: DevicePtr,
    n: u32,
    k: u32,
    cap: u32,
    num_tokens: u32,
    stream: u64,
) -> Result<()>

2026-09-27: Grouped NVFP4 gate+up and SiLU over k-wide input rows. cap is fp8_grouped_active_cap; the first ceil(num_tokens / geometry.rows_per_pass) block rows are the shared expert. Writes the FP32 product silu(bf16(gate)) * bf16(up), [positions, n] for the routed experts into act and [num_tokens, n] for the shared expert into sh_act (2026-10-02: the tensor-core kernel writes it as BF16 hi + lo in the same bytes, which only its down kernel reads).