Active Harmony is an auto-tuning package that supports a flexible plugin model to allow different search strategies and exchanging data with other Super tools such as Chill and TAU.
TAU Performance System is a portable profiling and tracing toolkit for performance analysis of parallel programs written in Fortran, C, C++, UPC, Java, Python.
CHiLL is a composable high-level loop transformation framework. It supports a wide variety of loop transformations on a whole loop nest or individual sub loop nests inside.
PAPI is a performance monitoring library that provides tool designers and application engineers with a common and coherent interface to the hardware performance counters (available on all modern CPUs) and other hardware components of interest (e.g., GPUs, network, and I/O systems).
PaRSEC (Parallel Runtime Scheduling and Execution Controller) is a framework that enables parallel computing through dataflow based task execution. It supports execution on distributed memory heterogeneous architectures and provides architecture-aware scheduling and management of micro-tasks, as well as automatic handling of the necessary communication between tasks.
An open-source extensible framework for the definition of domain-specific languages and generation of optimized (C, Fortran, CUDA, OpenCL) code for multiple architecture targets (e.g., CPUs, NVIDIA and AMD GPUs, Intel Phi), including support for empirical autotuning of the generated code.