A tflop, or teraflop, system is a parallel supercomputing system that has the ability to compute one trillion floating point operations in a single second. This term applies in many ways to the basic systems that run database systems.
The tflop system is not the first parallel supercomputing system to be created, although it has become the most used. The Intel Paragon was in existence before it and is also a parallel processing system. Each of these systems is a reliable option. The tflop system, however, is not as difficult to program as the Intel Paragon system. The N-cube system preceded the Intel Paragon, but this is an outdated system and very rarely used.
According to the American Association for Artificial Intelligence (AAAI), the two systems created prior to the tflop worked well for their time. The AAAI says that, “the Paragon and the N-cube were two of the many milestones on the path towards the massively parallel, distributed memory systems of today.” Since the tflop is a parallel processing system, a large problem can be broken down into smaller, more manageable problems with this system. As such, it truly divides and conquers a difficult task. In fact, most scientists compare the process used in the system to using many computers to work on many smaller parts of a single project.
With the tflop system, the user can spread the work out rather than processing it all on one chip. This makes the work go much faster. Increased speed is possible because it is a distributed memory system. When memory is distributed, every aspect of the system has its own memory. This allows each point of the overall project to receive individualized attention. If a certain aspect of the project being run with the system is too big, the main component controlling that part can request extra memory.
The software running within the tflop system is called Cougar. It is designed to cut down on the work a program does, making more of the hardware available to the computer user. Tflop has revolutionized the computer industry in relation to databases because, with it in place, databases can work much faster.