Beyond the Silicon: Redefining GPU Innovation Through Software and Methodology

Title

Beyond the Silicon: Redefining GPU Innovation Through Software and Methodology

Speaker(s)

Antonio Salvemini

Date and time

2025-10-23T10:40:00Z

Session type

:speech_balloon: Talk (45 minutes)

Abstract

The next generation of GPU architectures is not just about faster hardware; it’s about the deep synergy between groundbreaking design and the software that makes it real. In this session, we will journey through the future of GPU innovation, starting with the transformative possibilities of ray tracing and path tracing. We will look beyond the silicon to the vast, open-source software stacks that power this work, exploring why building complex software hand-in-hand with hardware prototypes is no longer optional, it’s essential. Along the way, we will confront the challenges of continuous development, automated quality assurance, and scaling massive codebases without losing momentum. Finally, we will explore how development pipelines are being integrated into cloud-native environments, enabling more flexible, scalable, and resilient approaches to building the future of high-performance computing.

Speaker(s) bio

Antonio is a visionary leader in computer graphics engineering, blending deep technical expertise with robust marketing acumen to drive innovation across the digital media landscape. From his first lines of code at age 10, he has been driven by a passion for solving challenges and pioneering breakthrough technologies. Over decades, Antonio has mastered Virtual Reality, advanced rendering engine integrations, PBR materials, Computer Vision and plugin solutions for platforms like Autodesk 3ds Max, Rhino, Solidworks, Unreal Engine, and Unity. His career includes influential collaborations with industry titans such as mental images —the creators of the renowned mental ray rendering engine— and NVIDIA, where he was built up as a professional marketing manager while contributing significantly to high-performance graphics and real-time rendering solutions. Notably, he was instrumental in co-developing the first-ever interactive rendering engine over the internet.

As Director of Graphics Engineering at Bolt Graphics, Antonio leverages his combined technical expertise and deep understanding of user needs and market landscapes to drive impactful innovation. Currently, he is leading the development of a fast path tracing rendering engine, a robust distributed rendering system, and a suite of corollary technologies designed to redefine real-time rendering and visual effects.


:question: Questions about this session? Please reply to this topic, top questions may be featured during the live event!

The gaming community would love such upgradable designs so they don’t have to buy a new card just for more memory.
The expandable memory feature is great for CGI folks who have so many triangles that dont usually fit into GPU memory, so they use CPU for CGI, CPU also gives them more flexibility in operations by using AVX512 features.
e.g. pixar

I really wonder how capable to adapt to different niches outside of graphical applications? and this type of approach does memory leak issues are more obvious than traditional gpus?

Did the rapid rise and proliferation of AI play into the development of recent hardware?