Function boxMullerWriteOut128Bytes

src/backend/cuda/kernel/random_engine.hpp:440–449

// Normalized writes without boundary checking

__device__ static void boxMullerWriteOut128Bytes(float *out, const uint &index,
                                                 const uint &r1, const uint &r2,
                                                 const uint &r3,
                                                 const uint &r4) {
    boxMullerTransform(&out[index], &out[index + blockDim.x],
                       getFloatNegative11(r1), getFloat01(r2));
    boxMullerTransform(&out[index + 2 * blockDim.x],
                       &out[index + 3 * blockDim.x], getFloatNegative11(r3),
                       getFloat01(r4));
}

__device__ static void boxMullerWriteOut128Bytes(cfloat *out, const uint &index,
                                                 const uint &r1, const uint &r2,