We have reached the point where we can propose a pre-release. It is limited in functionnalities, but it is a good illustration of the principles envisionned in FT-LA. This pre-release includes all the necessary kernels to devellop an ABFT matrix-multiply. We also provide a test application taking advantage of those features to provide the Summa matrix-multiply algorithm.
This release is also compatible with the FTB 0.5 API from the CIFTS project. The demo application illustrates how it can benefit from intelligent scheduling descisions trough FTB cooperation.
