Contact 
Potential of High Performance Computing
When writing a program, the first focus is correctness. Then follows the need for speed, thus the target becomes performance. High Performance Computing (HPC) aims at providing fastest computing solutions to both scientific and industrial problems. Many efforts have been made so far on the way to cutting-edge HPC processors and powerful supercomputers. The processing power of current and future supercomputers is impressive, indeed we are already in the exaflop era and the corresponding machines are emerging in a way that it might not take a long time to have them common. HPC is genuinely needed in various application domains with a noteworthy impact. Cutting-edge Artifical Intelligence is a specific case where HPC has led to another level of qualitative performance.
When writing a program, the first focus is correctness. Then follows the need for speed, thus the target becomes performance. High Performance Computing (HPC) aims at providing fastest computing solutions to both scientific and industrial problems. Many efforts have been made so far on the way to cutting-edge HPC processors and powerful supercomputers. The processing power of current and future supercomputers is impressive, indeed we are already in the exaflop era and the corresponding machines are emerging in a way that it might not take a long time to have them common. HPC is genuinely needed in various application domains with a noteworthy impact. Cutting-edge Artifical Intelligence is a specific case where HPC has led to another level of qualitative performance.
Expectations
HPC typically combines methods, machines and programming to deliver the fastest system to solve a given problem or run a specific computing task. With the increasing so-called FLOPS (Floating Point Operations per Seconds), commonly provided/advertised throught its nominal value (theoretical), the main expectation is to get a high fraction of the available potential power. Thus, important investments are made to acquire high-end HPC infrastructures with the aim of having the desired processing power (at least!).
HPC typically combines methods, machines and programming to deliver the fastest system to solve a given problem or run a specific computing task. With the increasing so-called FLOPS (Floating Point Operations per Seconds), commonly provided/advertised throught its nominal value (theoretical), the main expectation is to get a high fraction of the available potential power. Thus, important investments are made to acquire high-end HPC infrastructures with the aim of having the desired processing power (at least!).
Reality
The fact is that a high-end HPC unit/machine is complex: several nodes/cores, CPUs/GPUs/FPGAs, vector units, hierarchical memory, specific set of instructions, thus in order to get the desired level of power, genuine efforts have to be made in several programming directions with various skillful hardware considerations.
The fact is that a high-end HPC unit/machine is complex: several nodes/cores, CPUs/GPUs/FPGAs, vector units, hierarchical memory, specific set of instructions, thus in order to get the desired level of power, genuine efforts have to be made in several programming directions with various skillful hardware considerations.
HPC Programming
Thus, in order to take a significant advantage of hardware advances, we need powerful methods to solve problems together with highly skillful programming efforts and relevant frameworks.
Thus, in order to take a significant advantage of hardware advances, we need powerful methods to solve problems together with highly skillful programming efforts and relevant frameworks.
