Solution 1:

In prior versions before SQL Server 2005, DBCC PINTABLE was used to plug the table in memory but it had some inherent issues with data corruption. While the command DBCC PINTABLE still exists, it simply does nothing.

I think you are making it too complicated. SQL Server will read the data from disk only if it doesn't already exist in memory. Once it is in memory, it will stay there until it is NOT being used for sometime or it gives way for reading something else. If you have enough memory (this is completely relative) then your table may already be in memory. I have simplified this a bit here. It uses LRU-k algorithm to keep the pages in cache. 10 rows and 10 columns of smallint/int data will fit in one page i.e 8KB and you should stop pursuing your initial plan.

Ref: http://msdn.microsoft.com/en-us/library/ms191475.aspx

One the page is in cache, even if you update the data it won't be written to disk immediately and there is an asynchronous process (CHECKPOINT, Lazy writer) to flush this data to disk later on. You could update the data like 100 times and it is possible that SQL Server writes to disk only once. It depends on if you have messed with the recovery internal flag, amount of data changes and stuff like that. I guess at this time its too much info.