The fastest way to get huge array of random numbers

Question

I'm making a game. There are lootboxes. Player can open at the same time thousands of lootboxes, and each lootbox have several items.
I have an extension method that returns an array of random elements:

 static public IEnumerable<T> Random<T>(this IEnumerable<T> collection, int countOfRandomElements)
    {
        var collectionSize = collection.Count();
        for (int id = 0; id < countOfRandomElements; id++)
            yield return collection.ElementAt(random.Next(collectionSize));            
    }

And as you might have guessed when player opens to many lootboxes at once (3k+) heavy lags begin. Collection size is changing from 3 items to 100 approximately. So I decided to try writing a method that will randomly assign number to each object from the collection. Equity of this number is easy to find. But I can't correctly calculate standard deviation because it's changes for different size of collection.

Now I'm looking for any clue in probability theory, but my research has not yielded any results so far. Could anyone help me please?

---

Update. Thanks to comments I make small change. Now it works much more faster. I didn't realize that LINQ .ElementAt method is to slow until replace it with .ToArray() . 3 millions of objects with help of this method are generating about 1.5 sec now instead 1 minute as it was with .ElementAt for collection of 30 elements. Simple solution - fascinating execution time impromvent.

    static public IEnumerable<T> Random<T>(this IEnumerable<T> collection, int countOfRandomElements)
    {
        var array = collection.ToArray();
        var collectionSize = array.Length;
        for (int id = 0; id < countOfRandomElements; id++)
            yield return array[random.Next(collectionSize)];
    }