Bach­e­lor’s The­sis / Stu­dent Re­search Pro­ject

Ab­stract

Ab­stract mod­el­ing of HW/SW sys­tems is a rel­a­tively new re­search topic. This tech­nique aims to cap­ture only the es­sen­tial pa­ra­me­ters of soft­ware and hard­ware that in­flu­ence their tim­ing be­hav­ior.

This stu­dent pro­ject’s goal is to im­ple­ment a Python-based gen­er­a­tor that cre­ates Ma­chine Learn­ing Ac­cel­er­a­tor de­scrip­tions based on the Abstract Com­puter Archi­tec­ture Des­crip­tion Lan­guage (ACADL). Fur­ther­more, those ar­chi­tec­tures will be eval­u­ated using dif­fer­ent meth­ods for run­time es­ti­ma­tion.

An Ex­am­ple of a sim­ple ma­chine learn­ing ac­cel­er­a­tor mod­elled with ACADL is pre­sented here:

Re­quire­ments

• Python
• Suc­cess­fully at­teded the lec­ture “Grund­la­gen der Rech­ner­ar­chitek­tur” and/or “Par­al­lele Rech­ner­ar­chitek­turen” (op­tional)
• Linux (op­tional)

Con­tact

Lübeck, Kon­stan­tin

Jung, Alexan­der

Bring­mann, Oliver