...
Active-Learning-based Surrogate Models for Empirical Performance Tuning
Performance models have profound impact on hardware-software co-design, architectural explorations, and performance tuning of scientific applications. Developing algebraic performance models is becoming an increasingly challenging task. In such situations, a statistical surrogate-based performance model, fitted to a small number of input-output points obtained from empirical evaluation on the target machine, provides a range of benefits. Accurate surrogates can emulate the output of the expensive empirical evaluation at new inputs and therefore can be used to test and/or aid search, compiler, and autotuning algorithms. We present an iterative parallel algorithm that builds surrogate performance models for scientific kernels and work-loads on single-core and multicore and multinode architectures. We tailor to our unique parallel environment an active learning heuristic popular in the literature on the sequential design of computer experiments in order to identify the code variants whose evaluations have the best potential to improve the surrogate. We use the proposed approach in a number of case studies to illustrate its effectiveness.
Applications and their challenges on Blue Waters
The leadership class Blue Waters system is providing petascale level computational and I/O capabilities to its partners. To date there are approximately 32 teams using Blue Waters to pursue their science and engineering on 22,640 Cray XE CPU compute nodes and 4,224 Cray XK GPU nodes with a 26 PB, 1 TB/s filesystem. The challenges encountered by the teams are as varied as the applications running on Blue Waters. This talk will provide an overview of the Blue Waters system, its recent upgrade in GPU computing capability and network dimension, and a discussion of the
applications and their challenges computing at scale on Blue Waters.
We will present our last result on the SMPI/SimGrid framework. SMPI now implements all the collective algorithms and selection logics of both OpenMPI and MPICH and even a few other collective algorithms from Star MPI. Together with a flexible network model and topology description mechanisme, this allowed us to obtain almost perfect prediction of NASPB and BigDFT on Ethernet/TCP based clusters. We are currently working on extending this work to other kind of networks as well as on mixing the emulation capability of SMPI with the trace replay mechanism. We are also working on improving the replay mechanism so that it handles seamlessly classical trace formats.
Fault tolerance has been presented as an emerging problem for decades, with researchers often claiming that the next generation of hardware will introduce new levels of failure rates that will destroy productivity and cause applications to become unusable. While it is true that as machines have scaled, resilience has become more and more of a concern, there are issues already affecting applications at current scales. Process failure remains a concern, though primarily for applications that can run at the largest scales or on very unstable hardware. For smaller applications however, there are other concerns, such as soft errors, performance loss, etc. This talk will cover some of the research being performed in the Programming Models and Runtime Systems group at Argonne National Laboratory to study these phenomena.