The presentation focuses on multiscale wave propagation problems. Elastic wave propagates either inside a body or on surface. Shock wave generates micro-cracks, and results in the macro-cracks growth. In order to investigate such problems, an upgraded version of ParFUM (Parallel Framework for Unstructured Meshes) is used on top of Charm++ so that we have a simple and convenient environment for parallel programs which operate on unstructured meshes, while achieving excellent scalability even for complex applications. Dynamic crack microbranching processes are investigated by means of a large-scale computational fracture mechanics approach using the finite element method with special interface elements and a topological data structure representation. The fracture events will be represented by interface elements with tractions across the interface that follows a nonlinear cohesive model driven by work conjugate displacement jumps. Special operators to reduce mesh dependency are addressed, such as nodal perturbation and edge-swap. The ultimate goal is to achieve physical fragmentation simulations involving evolving mesh information when truly extrinsic cohesive elements are inserted adaptively.

NAMD is a portable parallel application for biomolecular simulations. NAMD pioneered the use of hybrid spatial and force decomposition, a technique used now by most scalable programs for biomolecular simulations, including Blue Matter and Desmond developed by IBM and D. E. Shaw respectively. NAMD is developed using Charm++ and benefits from its adaptive communication-computation overlap and dynamic load balancing. This talk focuses on new scalability challenges in biomolecular simulations: using much larger machines and simulating molecular systems with millions of atoms. We describe new techniques we have developed to overcome these challenges. Since our approach involves automatic adaptive runtime optimizations, one interesting issue involves harmful interaction between multiple adaptive strategies, and how to deal with them. Unlike most other molecular dynamics programs, NAMD runs on a wide variety of platforms ranging from commodity clusters to supercomputers. It also scales to large machines: we present results for up to 65,536 processors on IBM's Blue Gene/L and 8,192 processors on Cray XT3/XT4 in addition to results on NCSA's Abe, SDSC's DataStar and TACC's LoneStar cluster, to demonstrate efficient portability. Since our IPDPS'06 paper two years ago, two new highly scalable programs named Desmond and Blue Matter have emerged, which we compare with NAMD in this talk.

Simulations of galaxies forming in their cosmological context poses a number of challenges to performance on large parallel machines. The first is the very non-local nature of gravitational forces. Galaxies are influenced by the gravitational forces originating tens of megaparsecs away, requiring significant communication in the force solver. Second is the enormous spatial dynamic ranges involved, from megaparsecs to sub-parsec scales, requiring dynamic hierarchical data structures. Third is the vast time scales involve, from less than 1 million years to the age of the Universe, posing significant challenges for load balancing. This talk will present how these challenges have been addressed in the design of ChaNGa, the Charm N-body GrAvity solver.

Accelerators such as Graphical Processing Units (GPUs) and specialized cores, such as the Synergistic Processing Elements (SPEs) on the Cell processor, are being used with greater frequency in the realm of parallel computing to speedup computationally heavy portions of code. These systems are comprised of multiple types of processing elements, each with unique characteristics, strengths, weaknesses, and programming paradigms. Developing applications can be challenging since many architectural details must be taken into account. In this talk we will summarize the ongoing efforts to allow the Charm++ Runtime System to utilize accelerators while trying to abstract away as many architectural details as possible. Specifically, we will cover work related to the Cell processor and GPUs.

ParFUM is a framework for writing unstructured mesh codes in Charm++. In this talk I will describe the software architecture of ParFUM and discuss the services it provides to applications, including partitioning, load balancing, and adaptivity. I will also give a brief overview of successful ParFUM applications and discuss their performance and scalbility.

PetaFLOPS-class computers are currently being developed and even larger computers are being planned. Our BigSim project is aimed at developing tools that allow one to develop, debug and tune/scale/predict the performance of applications before such machines are available so that the applications can be ready when the machine first becomes operational. It also allows easier "offline" experimentation of parallel performance tuning strategies --- without using the full parallel computer. To the machine architects, BigSim provides a method for modeling the impact of architectural choices (including the communication network) on actual, full-scale applications. In this talk, we will present our simulation framework which consists of an emulator and a simulator; we will focus on the recent progress in integrating instruction level simulation with our framework, and out-of-core emulation support.

In this talk, we will cover three different programming paradigms that leverage the Charm++ programming system. The first one is Adaptive MPI (AMPI), an implementation of the popular MPI standard. AMPI is based on Charm++, and implements the traditional MPI tasks with user-level migratable threads. Thus, AMPI provides advanced features such as dynamic load balancing and automatic overlap between computation and communication to traditional MPI codes. Porting legacy MPI codes to AMPI typically involves no change to the sources. We will review AMPI's basic features, and discuss its current status. We will also talk about Charisma, a parallel object orchestration language that explicitly specifies global flow of control and data. We will proceed to discuss MSA, which encourages the disciplined use of shared address spaces. We will also talk about how these languages fit into our research agenda of creating various incomplete languages that are targeted towards specific programming paradigms and which are interoperable over a common ARTS, thereby promoting productivity.

In this talk I will discuss some of the more recent features added to CharmDebug, the Charm++ debugger. Emphasis will be given to introspection and memory debugging.
A new Python module allows the user to upload fragments of Python code into the application being analyzed. This uploaded code can access the application's data and perform introspection, for example by checking whether some data is within a validity range.
The combination of CharmDebug with a memory module inside Charm++ allows the debugger to collect various information about the status of the memory, such as Allocation Trees and Allocation Graphs. In addition, some operations can be performed on the allocated memory, such as leak detection and cross-object corruption.

Over the last years, scripting languages have been gaining importance in many applications. One area in which these languages have not been much explored is parallel programming. Parallel programming has always been strongly associated with scientific usage, but has recently gained new fields of action. With this change, the development of new programming paradigms of parallel programming becomes necessary in order to make development easier and applications more dynamic. Scripting languages may come into play here, bringing dynamism, flexibility and simplicity to applications. In this paper we describe the integration of the Lua scripting language into the Charm++ framework. With the resulting binding, any distributed object (chare) can be written either in Lua or in C++.

This talk describes a new scalable stream mining algorithm called NoiseMiner that analyzes parallel application traces to detect computational noise, operating system interference, software interference, or other irregularities in a parallel application's performance. The algorithm detects these occurrences of noise during real application runs, whereas standard techniques for detecting noise use carefully crafted test programs to detect the problems. This paper concludes by showing the output of NoiseMiner for a real-world case in which 6 ms delays, caused by a bug in an MPI implementation, significantly limited the performance of a molecular dynamics code on a new supercomputer.

This talk is split into two sections covering recent work to optimize Charm++ for new and popular architectures. The first section, delivered by Chao Mei, will focus on shared memory multicore machines. The second section, delivered by Eric Bohm, will introduce the new CkDirect API for RDMA, and discuss preliminary results of its use on Infiniband and Blue Gene/P.