Dear All,
We are looking for a top candidate to join our research group as a PhD
student.
We have one opening in the area of Smalltalk-based Electronic Design
Automation. Please could you broadcast to any interested student.
The position is to be filled as soon as possible. The project starts
on 1st october 2011.
Kind Regards,
LoIc Lagadec
ARDyT - Fault tolerant dynamically reconfigurable architecture
Position type: PhD Student
Functional area: Brest, France
Research theme: Smalltalk, Reconfigurable architecture, innovative
toolkit design
Project: ARDyT
Contact:
[hidden email]
Title: Instrospection plan compliant EDA toolset
Reconfigurable computing (eg. FPGAs) combines the flexibility of
software with the high performances of hardware.
This comes from the on-demand metamorphosis of the hardware - at
runtime - by "loading" a new circuit on the reconfigurable fabric, while
this technology also supports extreme optimizations, by allowing
drastic changes to the datapath itself, in addition to tailoring the
control flow as traditionaly met in microprocessors.
While improving performances has been the key issue for years, very
few efforts have been reported in debug automation.
Debuging requires introspection capabilities, controlability and
fast-changes support. HLS (High Level Synthesis) tools allow to
compile high-level (C like) languages to netlist (electronic level),
hence guarantee fast changes but controlability and instrospection are
still very limited.
In the scope of the ARDyT project (supported by the French National
Research Agency), some evolutions will be carried out over a classical
FPGA such as embedding extra resources that will offer some
introspection capabilities.
The PhD thesis will contribute to refactor/re-design an existing
smalltalk based toolset in order to take into account these extra
resources hence new paradigms such as to combine the computation plan
and the instrospection plan of the proposed architecture. In addition,
debug controllers will be made available on-demand within the netlist
of the application.
This offers new exciting perspectives, in a scope of domain space
exploration (DSE). A model oriented approach will be considered to
ensure the durability of the resulting software and of its design
method.
This PhD will be co-supervised by the Lab-STICC MOCS and IRISA.
_______________________________________________
Esug-list mailing list
[hidden email]
http://lists.esug.org/mailman/listinfo/esug-list_lists.esug.org
Locked
