High Performance Computing Glossary - T


One teraFLOPS (TFLOPS) are 1012 FLOPS.


ThinLinc is a software product which enables organisations to implement a server-based computing infrastructure. By centralising computing resources in the data center rather than distributing them amongst end-users, IT systems become more efficent, secure and accessible. End-users’ desktops and applications are published from a central location, providing a seamless and familiar environment regardless of which device the user is connecting from or where they are located. Availability is also drastically increased due to the redundancy inherent in a server-based solution.


A thread is a sequence of programmed instructions that follows a strict sequence, one step at a time, in other words, operating sequentially. Threads are generated and controlled by the relevant application software. In principle several threads can be combined to form a process, and they are executed either sequentially or alternately within that process, thereby avoiding any time advantage. Best performance is always the objective in the HPC field, and therefore a process usually comprises a single thread, which then runs on one specific core.

TOP500. The list

TOP500. The list is an international ranking system for supercomputers using the LINPACK benchmark (measure for system’s floating point computing power). The TOP500 table shows the 500 most powerful commercially available computer systems in the world.

Transient Calculation

The characteristic of a parameter to not follow a constant value over time is called non-stationarity or transience. In structural mechanics it is also named dynamic simulation. The opposite is a stationary calculation.

Transient calculations are of great importance in engineering, particularly for hydraulic engineering, flow calculations, heat transfer calculations or other numeric calculations, where objects that change their condition over time are examined. The additional differentiated consideration of a temporal component in a transient calculation requires a higher performance to finally solve all physical effects within a reasonable time.

Tests in the automotive sector showed that transient calculations on CPU 24/7 resources were 5-6 times faster than in-house workstations.