Passing vector with std::ref to threads does not seem to update the actual vector rather threads only updating local vector copy

I am facing a strange issue when passing a vector std::ref to pool of threads.

I cannot put the exact code but I have something like : Approach A

void myfunc(int start, int end, std::vector<double>& result)
{
   for(int i=start; i<end; ++i)
    {
        result[i] = ...
    }
}


vector<double> myvec(N);
const size_t num_threads = std::thread::hardware_concurrency();
vector<std::thread> threads;
threads.reserve(num_threads);

for(int idx=0; idx<num_threads; ++idx)
   threads.push_back(std::thread(myfunc, (idx*nloop/num_threads), (idx+1)*nloop/num_threads), std::ref(myvec)));

for(std::thread& t : threads)
  { 
    if(t.joinable())
     t.join();
  }

some_read_opration(myvec)...

But if I explicitly create threads and divide tasks by hard coding ranges, it gives me expected values.

This works: Approach B

std::thread t1(myfunc, 0, nloop/2, std::ref(myvec));
std::thread t2(myfunc, nloop/2, nloop, std::ref(myvec));

t1.join();
t2.join();

some_read_opration(myvec)...

I want parallelize my loop in more generic way, not sure where I am going wrong here. In the A approach : The issue I am facing is that, it seems like each thread updates result[] element locally but its not being reflected at myvec in the function which creates threads, even if I am passing std::ref(myvec). Any help would be great.