A Tool Suite for Simulation Based Analysis of Memory Access Behavior

Abstract

Abstract. In this paper, two tools are presented: an execution driven cache simulator which relates event metrics to a dynamically built-up call-graph, and a graphical front end able to visualize the generated data in various ways. To get a general purpose, easy-to-use tool suite, the simulation approach allows us to take advantage of runtime instrumentation, i.e. no preparation of application code is needed, and enables for sophisticated preprocessing of the data already in the simulation phase. In an ongoing project, research on advanced cache analysis is based on these tools. Taking a multigrid solver as an example, we present the results obtained from the cache simulation together with real data measured by hardware performance counters.

Visualizing the Program Execution Control Flow of OpenMP Applications

“…3. The callgraph is largely similar to the callgraphs given by other tools, such as callgrind [11], with the exception that the nodes are not only functions but also OpenMP constructs and user-defined regions, and the (runtime) nesting of those constructs is shown in the callgraph view. The callgraph that ompP records is the union of the callgraph of each thread.…”

DynamO: a free ${\cal O}$(N) general event‐driven molecular dynamics simulator

“…Cache simulations have been performed using Callgrind, [63] and for a density of ρσ 3 = 0.5, approximately half of the cache misses result from accesses to the contents of the neighbor lists.The remaining cache misses are associated with accesses to particle and event data.…”

ROOT I/O Improvements

“…The analysis has been made using a variety of tools including igProf [2] , callgrind [3] [4] and AMD CodeAnalyst. For example, we improved by a factor 3 to 10, depending on the length and complexity of the class name, the performance of the TTree::SetBranchAddress and TTree::SetAddress routines.…”