Yambo enabled to run on 1024 cores

The new beta version of Yambo broke the 1000 CPU wall by running on the new FERMI super-machine in CINECA. This giant step forward in the development of Yambo was possible thanks to an innovative CPU distribution inside the code structure. In this new version of Yambo the user has several new parameters to perform a fine tuning on the distribution of tasks among the different levels of CPU. Indeed the code can use dynamical work allocation in order to assign to the same CPUs different tasks during a single run. This beta version of Yambo will be released shortly to the user community. After a first period of testing this innovative CPU distribution will become the standard code internal structure.
This work has been supported by CINECA in the framework of the PRACE 2IP activity in collaboration with the developer's team. Yambo users' community is rapidly growning, in particular in Italy.
This achievement permits now to consider Yambo as a mature code, at the pace of HPC challenges. In particular, this is an important milestone that makes Yambo very suitable for PRACE Tier-0 architectures, like FERMI.

Figure: The new schema of distribution of CPU on many parallelization levels.