HPC developers usually try to solve the load imbalance problem in their own applications. In fig. 1 we can see a typical life cycle when developing an HPC application. This process is time-consuming for the programmer/expert and spends computational resources. Moreover, when talking about highly tuned applications, this cycle must be done for each different architecture the application will run and each input set used.

The objective of DLB is to avoid this cycle and offer a tool that can be used without a previous performance analysis of the application (see fig. 2). Therefore, we will reduce significantly the time spent by the programmers modifying the application, debugging and analyzing the sources of performance loss and also minimize the amount of computational resources necessary to run the performance tests.

DLB is a library devoted to speedup hybrid parallel applications (see fig. 3). And at the same time, DLB improves the efficient use of the computational resources inside a computing node.

To improve the performance of applications and maximize the use of resources inside a computational node DLB will attack the load imbalance at different levels. Instead of fighting the load imbalance DLB adapts the execution at runtime using the malleability of the inner level of parallelism (i.e. changing the number of threads in OpenMP).

This dynamism allows DLB to react to different sources of imbalance: Algorithm, data, hardware architecture and resource availability among others. The DLB approach to redistribute the computational resources, at runtime, depending on the instantaneous demand, can improve the performance in different situations:

• Hybrid applications with an imbalance problem at the outer level of parallelism.
• Hybrid applications with an imbalance problem at the inner level of parallelism.
• Hybrid applications with serialized parts of the code.
• Multiple applications with different parallelism patterns.

DLB will use the malleability of the inner level of parallelism to change the number of threads of the different processes running on the same node. There are different load balancing algorithms implemented within DLB. They all rely on this main idea but they target different types of applications or situations.

In fig. 4 we can see an example of a DLB load balancing algorithm. In this case, the application is running two MPI processes on a computing node, with two OpenMP threads each one. When MPI process 1 reaches a blocking MPI call it will lend its assigned CPUs (number 1 and 2) to the second MPI process running in the same node. This will allow MPI process 2 to finish its computation faster.