How to decrease the computer's time to run databases?

"“Tiny pointers,” which show the way to a piece of stored data, inspired the creation of a new kind of faster hash table." (Quanta, Undergraduate Upends a 40-Year-Old Data Science Conjecture)

"A young computer scientist and two colleagues show that searches within data structures called hash tables can be much faster than previously deemed possible." (Quanta, Undergraduate Upends a 40-Year-Old Data Science Conjecture)

"Sometime in the fall of 2021, Andrew Krapivin, an undergraduate at Rutgers University, encountered a paper that would change his life. At the time, Krapivin didn’t give it much thought. But two years later, when he finally set aside time to go through the paper (“just for fun,” as he put it), his efforts would lead to a rethinking of a widely used tool in computer science." (Quanta, Undergraduate Upends a 40-Year-Old Data Science Conjecture)

"The paper’s title, “Tiny Pointers (opens a new tab),” referred to arrowlike entities that can direct you to a piece of information, or element, in a computer’s memory. Krapivin soon came up with a potential way to further miniaturize the pointers so they consumed less memory. However, to achieve that, he needed a better way of organizing the data that the pointers would point to." (Quanta, Undergraduate Upends a 40-Year-Old Data Science Conjecture)

There was no use for that idea. 

However, the tiny pointers can help to create more effective databases. 

In the cases. That the system must handle even billions of database tables. There is the possibility that the query has two stages. 

The database tables are sorted under big pointers like boxes. 

There the tables that involve similar data are. The system uses pointers like maps. The main query is sent to the database cells or boxes. 

These tables are under certain topics. Those boxes are like cities. There are streets. That shows the system where it should find the right data. 

Then the subsystem searches the database from the street, blocks, and finally from the door number.

Or the system can search for the right thing. Using the name. But a tiny pointer turns that name to the door number. 

When somebody makes a query like "Where are the BMW gearboxes"?The system can search for the right table and event handler by using an algorithm. 

That connects the query words like "automobiles" to the box or door A21. There are databases. That handles cars and their spare parts. The system must have a connection for the words "cars" and "automobiles". 

But there is another way to sort the database. That is a straight line like a card file. The tables can be sorted like this. The tablet that responds to an event that requires the fastest reaction is the first. This kind of linear database can act as the system that should act as reflexes. 

When a computer user creates or uses a database. That requires a query. The query makes the program find the right table from the database structure. The problem is how the database program finds those tables. In professional database work, database creators keep every table as small as possible. 

The system must search all databases until it finds the right one. The system checks keywords from the query. 

Then it tries to find matching words from the database. This thing means that the system might compile the entire database structure to find the right tables. And in the worst case. That table is the last one on the list. This means that the system uses x-time for the database structure and the limit of the is the time that the system uses with each table. 

That time seems meaningless. Until we face an extremely complicated database structure. When there are billions of tables the time that the system uses databases compiling turns longer. 

If the query has no straight match in database keywords. That causes problems. Grammar errors can cause problems in the system match queries and tables. 

Especially if it requires very fast reactions. This kind of database is artificial intelligence.

Artificial intelligence means that in the large language model, LLM the user makes similar queries for the system. Then the system searches for the right table and data from the hard disk or the Internet. In the AI that controls physical robots and operates as an independent system, the query is replaced by the details of the things there the system must react to. When the system sees something it searches matches from the database. 

In that case, those databases must be sorted under the tiny marks. But the system must sort them like in boxes. On all boxes have etiquette. Like "response for traffic evens". That makes the system search the files very fast. 

The morphing neural system can make that process more effective. There certain boxes are under certain computers. The system routes the query to all of those computers. And they can search databases like card files. 

Then those tiny pointers are under one bigger pointer. That makes it easier to respond to things that require fast reactions. The morphing neural network can sort databases that involve certain types of reactions under certain big marks. 

The system can sort databases so that. The table that requires the fastest reaction place is the first in the line. 

That makes the system more effective. As you see small databases don't need special arrangements. 

 https://www.quantamagazine.org/undergraduate-upends-a-40-year-old-data-science-conjecture-20250210/?utm_source=flipboard&utm_content=topic%2Fsoftwaredevelopment