Information Technology Reference
In-Depth Information
game engine always pre-allocates enough memory for the maximum number of
supported clients.
Our implementation is capable of running in graphical and console mode.
Table 1 shows an average of the various client types' resource usage. The data
is the average of ten readings from a host with a quad core 2.6Ghz CPU (max
400%) and 4GB memory. The console version utilizes one thread, while the two
others use six. The console client requires significantly less processing power and
memory than the others, while network differences are minimal. The console
version allows up to 57 instances on this specific host, multiple host machines
should be used for large scale testing.
Tabl e 1.
Various Client Types and their System Load
Type
CPU Memory Bandwidth in
Bandwidth out
Regular player
252.42% 79 MB 1.64 Kb/s
5.80 Kb/s
VC Graphical
234.26% 94 MB 2.00 Kb/s
5.14 Kb/s
VC Console
6.94%
29 MB 1.41 Kb/s
6.00 Kb/s
4 Evaluation and Conclusion
Q3A has been modified into a load-generating tool for investigating game system
scalability, by implementing virtual clients that produce server load with authen-
tic network trac. Researchers can implement their concepts into the game and
analyze performance differences by utilizing the virtual client support. This can
be used to improve the credibility of [3] and similar research.
Q3A is old and may not be advanced enough for some researchers, but is likely
a better test platform than small prototypes such as the one utilized in [3]. Vir-
tual clients may be considered complex in comparison to client packet capture
and playback. However, packet traces cannot support evaluation of server pro-
cessing load. The current implementation supports only one virtual client per
instance of the software, wasting resources when launching multiple.
References
1. Halvorsen, S.M., Raaen, K.: Games for Research: A Comparative Study of Open
Source Game Projects. In: an Mey, D., et al. (eds.) Euro-Par 2013. LNCS, vol. 8374,
pp. 353-362. Springer, Heidelberg (2014)
2. Claypool, M., Claypool, K.: Latency and Player Actions in Online Games. Commu-
nications of the ACM - Entertainment Networking 49, 40-45 (2006)
3. Raaen, K., Espeland, H., Stensland, H.K., Petlund, A., Halvorsen, P., Griwodz, C.:
Lears: A lockless, relaxed-atomicity state model for parallel execution of a game
server partition. In: ICPPW 2012: Parallel Processing Workshops, vol. 41, pp. 382-
389. IEEE Computer Society, Washington (2012)
4. Stefyn, D., Cricenti, A.L., Branch, P.A.: Quake III Arena game structures. Swin-
burne University of Technology. Technical reports No. 110209A, Faculty of Informa-
tion and Communication Technologies. Centre for Advanced Internet Architectures
(February 2011)
