Visualizing complex dynamics in many-core accelerator architectures

Abstract

Abstract-While many-core accelerator architectures, such as today's Graphics Processing Units (GPUs), offer orders of magnitude more raw computing power than contemporary CPUs, their massive parallelism often produces complex dynamic behaviors even with the simplest applications. Using a fixed set of hardware or simulator performance counters to quantify behavior over a large interval of time such as an entire application execution run or program phase may not capture this behavior. Software and/or hardware designers may consequently miss out on opportunities to optimize for better performance. Similarly, significant effort may be expended to find metrics that explain anomalous behavior in architecture design studies. Moreover, the increasing complexity of applications developed for today's GPU has created additional difficulties for software developers when attempting to identify bottlenecks of an application for optimization. This paper presents a novel GPU performance visualization tool, AerialVision, to address these two problems. It interfaces with the GPGPU-Sim simulator to capture and visualize the dynamic behavior of a GPU architecture throughout an application run. Similar to existing performance analysis tools for CPUs, it can annotate individual lines of source code with performance statistics to simplify the bottleneck identification process. To provide further insight, AerialVision introduces a novel methodology to relate pathological dynamic architectural behaviors resulting in performance loss with the part of the source code that is responsible. By rapidly providing insight into complex dynamic behavior, AerialVision enables research on improving many-core accelerator architectures and will help ensure applications written for these architectures reach their full performance potential.

Analyzing Machine Learning Workloads Using a Detailed GPU Simulator

“…However, since these papers use profilers, unlike our work they can only provide higherlevel analysis about the behaviors of the applications. In comparison, GPGPU-Sim provides detailed information on memory usage, power, efficiency, can easily be extended to provide additional statistics, and can output useful plots of relevant statistics using AerialVision [32].…”

Nyami: a synthesizable GPU architectural model for general-purpose and graphics-specific workloads

“…Several papers have explored using graphical representations of pipeline state to examine behavior, including [5,30]. Figure 3 illustrates this functionality with four strands.…”

Performance Visualization for Large-Scale Computing Systems: A Literature Review

“…Microscopic profile in [4], PC-Histogram in [24] Timeline views. Use of timeline views in performance visualization systems can be classified into the following groups: • Describing run-time behaviors and communication paths Fig.…”

“…• Facilitating source code level analysis Fig. 3(a) presents a visualization of source code execution over time from AerialVision [24]. Program code is laid out in ascending source line number along the vertical axis.…”

“…In our survey, we found the use of such abstractions in performance visualization systems to be very limited. The only type being used is macro-micro composition [4,24], in which both global picture and local details are provided, and detailed information accumulates into larger coherent structures. Focus + context abstractions are effective in combining overview and detail information in a single dynamic display.…”