With Hyperthreading enabled, you currently have 4 hard NUMA nodes with 10 physical cores and 20 logical cores each. On SQL Server 2012 and 2014 you will have 4 matching soft NUMA nodes with 20 logical cores each.
Upon upgrading to SQL Server 2016 or higher, auto soft NUMA will automatically kick in. You'll now have 2 soft NUMA nodes per hard NUMA node for a total of 8. The soft NUMA nodes will stripe across even and odd CPUs per hard NUMA node. For example, soft NUMA node 0 might contain schedulers 0, 2, 4, 6, 8, 10, 12, 14, 16, 18 and soft NUMA node 1 might contain schedulers 1, 3, 5, 7, 9, 11, 13, 15, 17, 19.
The blog post that you referenced is difficult to understand, but I believe you are meant to fall under the "Server with multiple NUMA nodes" category (you have 4) and the "Greater than 16 logical processors per NUMA node" category (you have 20). The recommendation as a result is "Keep MAXDOP at half the number of logical processors per NUMA node with a MAX value of 16", so Microsoft is recommending that you change MAXDOP to 10.
That doesn't match my experience with large servers. I would leave MAXDOP set to 8 unless you run into some kind of issue. IMO changing it from 8 to 10 in your scenario isn't meaningful and is just introducing a new change without a clear benefit. I blogged about how I disagree with the published guidance here.