Article count:1813 Read by:3428572

Account Entry

[Fresh News] GDC 2024: Work graphs, mesh shaders, FidelityFX, developer tools, CPU optimization, etc.

Latest update time:2024-03-14
    Reads:

We're heading to San Francisco again!

The annual Game Developers Conference (GDC) is only a week away! This event has always been of great significance to us. Every year we announce our latest progress in the field of game graphics here, share valuable advice and information, and hope to help everyone on their game development journey.

GDC 2024 is no exception, we’ve prepared a wealth of exciting talks that we can’t wait to share with you!

Before we get into the specific schedule, we would like to express our sincere thanks to our special guests who are speaking with us this year. A huge thank you to Shawn Hargreaves (Microsoft® Direct3D®), Hampus Siversson (Massive Entertainment), Marco Bouterse (Nixxes Software), and Miguel Petersen (Striking Distance Studios) for their time and effort in joining us to make our presentation even better.


Advanced Graphics Summit

(Advanced Graphics Summit)


First, we are honored to be invited again to speak at the Advanced Graphics Summit. This year, we invited Matthäus Chajdas (AMD expert) and Shawn Hargreaves (Microsoft Direct3D development manager) to introduce the revolutionary Microsoft Work Graphs API. You can't miss it!

theme

GPU Work Graphs: Welcome to the future of GPU programming

time

Monday, March 18, 1:20 pm (local time)

Place

Room 303, South Hall

Presenter

Matthaus Chajdas (AMD expert)

Shawn Hargreaves (Microsoft Direct3D Development Manager)

Work graphs are a new API that allows the GPU to distribute workload autonomously without going through the host. Unlike previous APIs such as ExecuteIndirect, work graphs enable fully parallel producer/consumer relationships on the GPU. It supports high-performance GPU job creation, efficient recursive algorithms, expansion and compression, and many other features.

Join Matthäus Chajdas and Shawn Hargreave to learn about this new API and how it will change the future of GPU programming! Together they will discuss which workloads can benefit quickly from work graphs and where they can expect long-term growth. This presentation will give viewers an in-depth look at the changes to the API and HLSL, and we'll use live examples to show viewers how to apply work graphs and integrate them into their own games and applications.

Matthäus will also participate in the Advanced Graphics Summit: All You Can Ask panel discussion with other industry experts at the same location at 5:30 p.m. Everyone is welcome to ask questions and speak freely!


Microsoft DirectX State of the Union


AMD expert Rob Martin will represent AMD at the Microsoft DirectX State of the Union this Thursday (March 21) and deliver a speech together with Microsoft (and other partners). At that time, he will introduce Work Graphs and DirectSR. We are very excited to be attending this event from Microsoft and look forward to meeting you!


Sponsored Speaking


As in previous years, we will host six sponsored talks featuring our engineers and special guests. You'll hear the latest developments and valuable information about AMD FidelityFX technology, the AMD Radeon™ developer tool suite, and AMD Ryzen™ CPU optimization. In addition to these familiar topics, we will also introduce mesh shaders and the new shader tool GPU Reshape.

Lectures are each one hour in length, including time for questions.

theme

Mesh shaders in AMD RDNA™ 3 architecture

time

Wednesday, March 20, 9 a.m. (local time)

Place

Room 3001, West Hall

Presenter

Lou Kramer (Technical Fellow, AMD)

Max Oberberger (Senior Software Engineer, AMD)

This talk will introduce the mesh shader pipeline and how it maps to the AMD RDNA™ 3 architecture. By comparing the mesh shader pipeline to traditional vertex shader and pixel shader pipelines as well as the compute shader pipeline, viewers can better understand their implementation and performance similarities and differences. In addition, the talk will also cover the performance of mesh shaders and how to efficiently utilize hardware when using mesh shaders.

Additionally, the talk will cover meshlet generation and meshlet compression, as well as how to optimize assets for mesh shaders. Viewers should be able to better judge whether mesh shaders are preferable to traditional vertex pipelines. This talk will also introduce how to use tools to analyze and debug mesh shaders, and share a practical example of mesh shaders.


theme

High-performance rendering with AMD FidelityFX™ Super Resolution 3 (FSR 3) in Snowdrop

time

Wednesday, March 20, 10:30 a.m. (local time)

Place

Room 3001, West Hall

Presenter

Hampus Siversson (Associate Engine Programmer at Massive Entertainment)

Colin Riley (AMD Senior Software Development Manager)

AMD FSR Super Resolution 3 (FSR 3) is AMD’s open source image upscaling and frame interpolation solution. Massive Entertainment has integrated AMD FSR 3 technology into the Snowdrop engine, which supports multiple games and platforms.

You'll learn how they integrated FSR 3 into the pipeline, what issues they faced, and how this integration helped FSR 3 improve itself during its development.


theme

AMD Ryzen™ Processor Software Optimization

time

Wednesday, March 20, 12:30 a.m. (local time)

Place

Room 3001, West Hall

Presenter

Ken Mitchell (AMD Expert)

Leverage AMD Game Engineering's rich experience accumulated from long-term cooperation with AAA game developers to break through CPU performance bottlenecks and achieve higher frame rates and more build efficiency!

Learn about exciting AMD Ryzen™ products, which use advanced processor and graphics technologies to power today's notebook, desktop and workstation PCs.

Let’s explore data flow, simultaneous multithreading, resource sharing, instruction set evolution, cache hierarchies, and coherence. Leverage powerful performance analysis tools and application profiling technologies like Windows Performance Analyzer, Concurrency Visualizer and AMD uProf.

Learn best practices and lessons learned. Upgrade system software and build settings. Seize valuable code optimization opportunities. Examples cover C/C++, assembly, and hardware performance monitoring counters.


theme

Global illumination with AMD FidelityFX™ Brixelizer and AMD FidelityFX SDK updates

time

Wednesday, March 20, 2:00 p.m. (local time)

Place

Room 3001, West Hall

Presenter

Dihara Wijetunga (Senior Software Engineer, AMD)

The talk will introduce how AMD FidelityFX™ Brixelizer works and explore how to use sparse distance fields in Brix GI to achieve diffuse and specular global illumination. Engineers will walk through the algorithm step-by-step and provide solutions to problems such as filling the irradiance cache and handling static and dynamic geometry.

The presentation will demonstrate the comparison of image quality with ground truth ray tracing global illumination, and provide some preliminary performance data to help the audience understand the performance of Brix GI. The talk will also showcase the Brix GI API and how it integrates into existing Brixelizer integrations.

This talk will also preview new samples released with the new version of FidelityFX SDK and describe changes to existing samples.


theme

Optimize games with the Radeon™ developer toolset and tweak shaders with GPU Reshape

time

Wednesday, March 20, 3:30 pm (local time)

Place

Room 3001, West Hall

Presenter

Amit Mulay (Senior Software Development Engineer, AMD)

Chris Hesik (Chief Technology Researcher, AMD)

Miguel Petersen (Senior Rendering Engineer, Striking Distance Studios (Spain)

The speech mainly has two parts. First, we'll introduce the latest features in the Radeon™ developer toolset, which includes a rich set of GPU profiling and analysis tools, including the recent addition of a GPU crash analysis tool. We'll show how developers can use these tools to optimize game performance and describe how games perform on AMD Radeon GPUs. We'll also introduce the newly redesigned Radeon Developer Panel and show viewers how to use it to quickly obtain and transfer data to other tools in the suite.

Afterwards, Miguel (Striking Distance Studios) will introduce GPU Reshape, a tool for real-time instruction-level verification of shader operations. We will discuss the challenges of validating tools between multiple intermediate representations (IRs) and how to achieve unified models through a custom IR with seamless bidirectional conversion capabilities. At the same time, we will also delve into current verification methodologies and possible future developments of the tools.


theme

Post-mortem analysis of GPU crashes with AMD Radeon™ GPU Detective (RGD)

time

Wednesday, March 20, 5:00 pm (local time)

Place

Room 3001, West Hall

Speaker Adam Sawicki (Chief Technology Researcher, AMD)

Amit Mulay (Senior Software Development Engineer, AMD)

Marco Bouterse (Lead Graphics Programmer at Nixxes Software)

AMD Radeon GPU Detective (RGD) is a new powerful development tool from AMD for post-mortem analysis of GPU crashes (TDR) on Windows. Through the talk, you'll learn when these types of failures occur, why they are difficult to debug, and how RGD can help quickly narrow the search for the source of a crash.

You'll learn how to capture and analyze AMD GPU crash dumps in games and how to analyze the information in the Crash Analysis report generated by the tool. You'll also hear about a case where the tool helped investigate and fix a specific crash in a game.


Looking forward to your arrival!


Hope to see you at GDC! But if you can’t attend this year, feel free to check out our upcoming videos and follow us on social media for the latest news!



Latest articles about

 
EEWorld WeChat Subscription

 
EEWorld WeChat Service Number

 
AutoDevelopers

About Us Customer Service Contact Information Datasheet Sitemap LatestNews

Room 1530, Zhongguancun MOOC Times Building,Block B, 18 Zhongguancun Street, Haidian District,Beijing, China Tel:(010)82350740 Postcode:100190

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号