You are on page 1of 2

Jose Garcia

Prince Sanchez
Tyler Fregoso
Taylor Beimel

Research Project Group Summary

In the article 'A Cloud Gaming Framework for Dynamic Graphical Rendering Towards
Achieving Distributed Game Engines,' James Bulman and Peter Garraghan of Lancaster
University discuss the evolution of cloud gaming and the challenges it faces in maintaining
consistent performance due to issues with virtual machine hardware, network connectivity, and
data caps. To address these challenges, a new cloud gaming framework is introduced, enabling
the distribution of game engine execution across cloud virtual machines and devices. This
framework leverages diverse graphic APIs and implements frame interlacing techniques to
improve game frame rates by up to 33%. In the introduction, the significant growth of the video
game industry is highlighted, with cloud gaming emerging as a promising solution due to
benefits like faster installation, reduced hardware costs, and improved graphics quality. However,
limitations arise from reliance on cloud instances and network connections, leading to quality
degradation and service interruptions. The proposed solution involves dynamically distributing
game engine subsystems across cloud and client systems to manage performance and
connectivity issues. The paper presents empirical evidence demonstrating the framework's ability
to enhance game performance and fault tolerance, marking a 33% improvement over traditional
rendering methods.
Now let's go over why this issue is important. Cloud gaming is increasingly popular,
offering players the ability to play high-quality games on less powerful devices without the need
for extensive downloads and installations. However, the quality of the gaming experience
heavily depends on the continuous high performance of cloud servers and stable internet
connectivity. Any disruption in these elements can lead to poor game quality, including lag,
lowered frame rates, and even complete service outages. Addressing these issues is essential not
only for improving user satisfaction but also for expanding the accessibility of advanced gaming
experiences to a broader audience. This is particularly significant in regions where the internet
may not be as strong or for users who may not have access to the latest hardware. Enhancing the
power and flexibility of cloud gaming systems ensures a more reliable and inclusive gaming
experience for all users, regardless of their local system capabilities or network conditions.
The authors are proposing a new framework that improves cloud gaming on multiple
fronts. To do this the authors spread the game engine process across devices and cloud Virtual
Machines. One major part of this framework is separating the graphical render from the game
engine itself. All graphic API calls are now done with generic commands. These commands are
later processed and translated by CloudRend. CloudRend uses two components. The game
instance manager converts and translates the commands and then mitigation is used to
dynamically render the graphics based on current limitations. The authors also introduce ways to
employ interlacing and hot swapping,. Interlacing allows for collaborative rendering between
cloud VMs and client devices, dynamically adjusting the distribution of rendering tasks based on
performance metrics and network conditions. Hot swapping allows the framework to implement
real time adjustments to the rendering without actual impact on the game. This helps the
rendering dynamically change even during major events like network disruptions. Essentially,
the authors are proposing a way to simplify game development using generic commands, while
also improving the adaptability and rendering capabilities of the game.
The authors tested the metrics of their proposed cloud gaming framework by using a
single cloud machine (i7-7700HQ, GTX 1050 GPU running OpenGL/Vulkan) and a client
machine (Raspberry Pi 4, Quad Core Cortex-A72 which could only run OpenGL). They were
connected over a variable latency network (between 16 - 100 ms). The experiments involved
running a low polygon scene & high polygon scene to test graphical API computational
requirements; each test ran for 10,000 frames, and they measured frames-per-second & network
utilization. They also ran experiments using ranging frame interlacing, from 100% on local to
100% cloud instance, while hot-swapping the graphical API per frame on the cloud. Their results
show that by using collaborative rendering evenly across the cloud & client, you get consistently
higher frame rates; this could be further improved by using a video encoder. Overall, they
demonstrated a 33% performance improvement in frame rates with their proposed cloud gaming
framework.

You might also like