One of the major problems of computer science since its inception, while today is not so into account in according to which programs, it is the memory management. I say this because, obviously, it is not the same thing your program look for information in a database with thousand records, even though linking tables, that will have to access a database with multiple tables containing millions of records each.
This last makes that today have to make very optimized queries so these take no time too, which in turn makes the user experience very bad. For trying to get a greater fluidity in this type of queries to databases of million and million of records from the MIT is has created Milk, a new language of programming that, according to the tests made, can reach the Quad of speed with algorithms common.
As Vladimir Kiriansky, PhD student in electrical engineering and computer science at MIT says:
It is as if whenever you want a spoonful of cereal, he opened the refrigerator and a carton of milk, you vertieses a tablespoon of milk, you cerrases the cardboard box and you pusieses it back in the fridge.
The principle of locality plays an important role in the management of the majority of current memory chips. This basically means that programs must assume that they will need other data stored in different plots of memory, with massive data, this always is. To try to solve this problem Milk allows developers to manage memory efficiently, especially in programs that employ few data but these are scattered.
In a program developed with Milk, when a kernel needs some data, instead of searching in main memory uses the address of the element stored locally. In this way, the system only consumes resources to search those data you need and which can be obtained efficiently. According to tests conducted by own MIT, programs written with Milk tend to be up to four times faster than the developed ones with other languages.
More information: computerworld
