pub fn kquant_mmvq_experts_wn(
gpu: &dyn GpuBackend,
kernel: KernelHandle,
w_tables: DevicePtr,
y_q8: DevicePtr,
out_bf16: DevicePtr,
n: u32,
k: u32,
m: u32,
n_experts: u32,
y_stride_bytes: u32,
nwarps: u32,
stream: u64,
) -> Result<()>Expand description
2026-09-25: kquant_mmvq_experts_w with nwarps rows a block (the _w2 / _w8 entries
for 2 / 8; the plain _w entry is 4). One warp computes each row in the same way for any
nwarps. Fails unless nwarps is 2, 4 or 8.