From Exascale Supercomputing to FAIR data ...
According to Moore's law, the number of transistors in an integrated circuit doubles about every two years, giving us faster computers in the same pace. Latest when miniaturisation reaches atomic levels, Moore's law will be coming to an end (likely in 5 to 10 years). While we see in 2020 s...
| Main Author: | |
|---|---|
| Format: | Text |
| Language: | English |
| Published: |
https://b2share.eudat.eu
2020
|
| Subjects: | |
| Online Access: | https://dx.doi.org/10.23728/b2share.a6a4682fe1f74b32b8b67948f7ce6965 https://b2share.eudat.eu/records/a6a4682fe1f74b32b8b67948f7ce6965 |
| Summary: | According to Moore's law, the number of transistors in an integrated circuit doubles about every two years, giving us faster computers in the same pace. Latest when miniaturisation reaches atomic levels, Moore's law will be coming to an end (likely in 5 to 10 years). While we see in 2020 still an increase in number of transistors, general purpose CPU cores do not really get faster anymore. This did lead to the rise of specialised "accelerators" (such as GPUs). With current High-Performance Computing clusters being in the PetaFLOPS range, Exascale (i.e. 10^18 FLOPS=floating point instructions per second) is the next (and maybe final) sonic barrier to break. The first part of this talk will report from the DEEP-EST Exascale research project that develops a modular supercomputing architecture to provide different hardware modules each employing different accelerator technologies tailored to specific problem domains, such a simulation or machine learning. The second part of this talk will cover the European Open ... : Presentation slides for presentation held as part of the seminar series organized by the University of Iceland Engineering Research Institute, 17.4.2020. Video recording of presentation available as stream via: https://hi.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=5f48b270-b9d2-4b8b-8ed6-aba000e6be49 ... |
|---|