tiny.simd.sumsOf8

1pub fn sumsOf8(comptime D: type, value: D.Vector) D.repartition(wideLane(wideLane(wideLane(D.Lane)))).Vector {2    const W = wideLane(wideLane(wideLane(D.Lane)));3    const R = D.repartition(W);4    if (D.lane_count < 8) @compileError("sumsOf8 requires at least eight lanes");5    var result: R.Vector = undefined;6    inline for (0..R.lane_count) |index| {7        const base = index * 8;8        result[index] = widen(W, value[base]) + widen(W, value[base + 1]) +9            widen(W, value[base + 2]) + widen(W, value[base + 3]) +10            widen(W, value[base + 4]) + widen(W, value[base + 5]) +11            widen(W, value[base + 6]) + widen(W, value[base + 7]);12    }13    return result;14}

1pub const sumsOf8 = reduce.sumsOf8;