benefit to convert float values to bitwise integer before comparing?

Question

When we compare two objects of the same class in c# which is generated from a proto file, we can use Equals() since grpc protobuf overwrites Equals() for us to compare each fields.

However, for float and double fields, there's ProtobufEqualityComparers like this:

  /// <summary>
  /// Provides a central place to implement equality comparisons, primarily for bitwise float/double equality.
  /// </summary>
  public static class ProtobufEqualityComparers
  {
    /// <summary>
    /// Returns an equality comparer for <typeparamref name="T" /> suitable for Protobuf equality comparisons.
    /// This is usually just the default equality comparer for the type, but floating point numbers are compared
    /// bitwise.
    /// </summary>
    /// <typeparam name="T">The type of equality comparer to return.</typeparam>
    /// <returns>The equality comparer.</returns>
    public static EqualityComparer<T> GetEqualityComparer<T>()
    {
      if (typeof (T) == typeof (double))
        return (EqualityComparer<T>) ProtobufEqualityComparers.BitwiseDoubleEqualityComparer;
      if (typeof (T) == typeof (float))
        return (EqualityComparer<T>) ProtobufEqualityComparers.BitwiseSingleEqualityComparer;
      if (typeof (T) == typeof (double?))
        return (EqualityComparer<T>) ProtobufEqualityComparers.BitwiseNullableDoubleEqualityComparer;
      return !(typeof (T) == typeof (float?)) ? EqualityComparer<T>.Default : (EqualityComparer<T>) ProtobufEqualityComparers.BitwiseNullableSingleEqualityComparer;
    }

Specifically, there's a BitwiseDoubleEqualityComparer used for double values and BitwiseSingleEqualityComparer for float values.

Why do we need them?

I believe BitwiseDoubleEqualityComparer and BitwiseSingleEqualityComparer cannot solve precision issues when comparing floating numbers.