Daisen: A framework for visualizing detailed GPU execution

Screenshot of the Daisen visualization tool.
Daisen provides a web-based interactive visualization tool that enables the examination of GPU execution traces. (A) The Overview Panel shows key performance metrics against time of all hardware components using small multiples, (B) the Task View demonstrates the hierarchical relationships between tasks, (C) the Component View displays all the tasks executed in a hardware component (after selecting one component of interest from the Overview), and (D) the legend bar shows all the tasks involved in the Component View and the Task View with color encoded.
Graphics Processing Units (GPUs) have been widely used to accelerate artificial intelligence, physics simulation, medical imaging, and information visualization applications. To improve GPU performance, GPU hardware designers need to identify performance issues by inspecting a huge amount of simulator-generated traces. Visualizing the execution traces can reduce the cognitive burden of users and facilitate making sense of behaviors of GPU hardware components. In this paper, we first formalize the process of GPU performance analysis and characterize the design requirements of visualizing execution traces based on a survey study and interviews with GPU hardware designers. We contribute data and task abstraction for GPU performance analysis. Based on our task analysis, we propose Daisen, a framework that supports data collection from GPU simulators and provides visualization of the simulator-generated GPU execution traces. Daisen features a data abstraction and trace format that can record simulator-generated GPU execution traces. Daisen also includes a web-based visualization tool that helps GPU hardware designers examine GPU execution traces, identify performance bottlenecks, and verify performance improvement. Our qualitative evaluation with GPU hardware designers demonstrates that the design of Daisen reflects the typical workflow of GPU hardware designers. Using Daisen, participants were able to effectively identify potential performance bottlenecks and opportunities for performance improvement. The open-sourced implementation of Daisen can be found at gitlab.com/akita/vis. Supplemental materials including a demo video, survey questions, evaluation study guide, and post-study evaluation survey are available at osf.io/j5ghq.
PDF | Preprint | DOI | Supplement | Code | Video Preview | Demo Video | Video Presentation | BibTeX
Yifan Sun
Ali Mosallaei
David Kaeli
Thumbnail image for publication titled: Daisen: A framework for visualizing detailed GPU execution
Daisen: A framework for visualizing detailed GPU execution

Yifan Sun, Yixuan Zhang, Ali Mosallaei, Michael D. Shah, Cody Dunne, and David Kaeli. Computer Graphics Forum—EuroVis/CGF. 2021. DOI: 10.1111/cgf.14303

PDF | Preprint | DOI | Supplement | Code | Video Preview | Demo Video | Video Presentation | BibTeX

Khoury Vis Lab — Northeastern University
* West Village H, Room 302, 440 Huntington Ave, Boston, MA 02115, USA
* 100 Fore Street, Portland, ME 04101, USA
* Carnegie Hall, 201, 5000 MacArthur Blvd, Oakland, CA 94613, USA