HPG 2020 is extending an invitation to enter a submission into our new student project competition! Submissions will be judged based on novelty and impact and all submissions will be hosted on the HPG website. The top eight submissions will be awarded the opportunity to present a poster on their project, online during HPG, along with a free registration grant to attend HPG 2020. The top two projects judged on the submission and presentation will be recognized live during the conference and receive a small monetary award for 1st and 2nd respectively.
This year, the theme of the competition is “Creatively using novel new GPU hardware features” with a focus on the new ray-tracing and ML inference hardware available in current and near future GPUs. We invite students to submit work that creatively re-purposes specialized graphics hardware and APIs to create something they were not necessarily intended to. The goal of the competition is to encourage creative thinking and to leverage the novel perspective of the computer graphics student population (our newest members of the HPG community) to explore the opportunities provided by the next generation of graphics hardware.
GPUs are massively parallel processors, but specialized GPU hardware features (targeted towards specific applications) provide a fairly rich set of efficient geometry primitives. For example, the rasterizer unit which is designed to perform fast point in polygon tests for triangle rendering, has been re-purposed countless times to achieve alternatives goals. An example of a non-obvious use of ray-tracing hardware was presented at HPG 2019: “RTX Beyond Ray Tracing: Exploring the Use of Hardware Ray Tracing Cores for Tet-Mesh Point Location” which demonstrated using ray-tracing primitives to accelerate a non-ray-tracing application. Indeed ray-tracing hardware and APIs provide a robust geometric primitive that has potential applications well beyond image synthesis, and machine inference hardware has an extensive opportunity for relevant, not immediately obvious high-impact uses (for example: Monte-Carlo image denoising).
Projects are not required to be publication ready, however rigor and quality will be integral criteria for judging submissions. We encourage topics making use of, but not limited to RTX ray-tracing hardware, DXR/VKRay/FireRays APIs, tensor core architecture, mesh shaders, callable shaders, DLSS, variable rate shading, Vega 7nm new instructions and any other recent advances in hardware features or extensions of the traditional graphics pipeline.
To be clear, we are looking more for clever, fun, and maybe even just wacky things to do with ASICs and their specialized APIs rather than hacking an RTU to do physics. Please keep in mind the following motivations for choosing a project.
1. Because it is fun to try and has the potential to inform the creative process of the graphics community going forward.
2. Because it furthers understanding of how existing ASICs could be used, as a step toward determining what extensions and generalizations might look like for enabling a broader scope of applications.
Submissions must be in the form of a webpage. The page should introduce and summarize the project, provide details and descriptions of methods, provide results (images, charts, tables), and provide a link to a YouTube video displaying the work in action. Voice over on the video is welcome though not required. However, please include the text “submitted to HPG 2020 student competition” in the video description, so that we can identify the work as yours.
Submit the complete web page as a compressed zip to firstname.lastname@example.org
Please note upon submission you authorize HPG to host your webpage on the HPG 2020 website.
Project submission – Monday, June 1st
Notification of competition results – Monday, June 8th