NSTL CONCLUSION
NSTL tested RealNetworks' RealSystem G2 and Microsoft's NetShow to determine whether any differences could be found between the two. The methodology was proposed by RealNetworks Inc., and was designed to contrast the ability of both players to handle data streaming, paying particular attention to the issues of rebuffering and packet data loss. The data collected by NSTL during the testing for this project reveals two products with contrasting aptitudes for handling these issues.
During testing, both media players were operated using UDP protocol, thus forcing each media player to deal with packet loss issues without any significant assistance from the protocol. Also, even though the encoding of the song data was for a 28.8 Kbps stream, the modems were operated at three different settings, thus incorporating the fact that actual data stream performance may vary, depending on time, network load, and line conditions.
As the data clearly indicates, RealPlayer G2, by dynamically adjusting its data stream, far surpasses Windows Media Player in its ability to handle its proprietary encoded stream when the provided bandwidth falls below the optimal setting. Not only that but, the lower the data rate the greater the disparity in the results exhibited by the two media players.
RealNetworks' "SureStream" technology, which is used to encode the song data in several parallel streams, allows the media player on the receiving end to choose which stream it will process, based on its monitoring of the available throughput. While RealPlayer G2 was able to keep a consistently reliable stream of song data flowing to the user, Windows Media Player many times would experience either excessive rebuffering or total "loss of connection" (no further content received from the content server). In fact, in a total of seventy-two tests conducted on the two products (thirty-six on each), Windows Media Player actually "lost the connection" five times while RealPlayer G2 maintained every single connection to the end of the song, regardless of how encumbered it was by the data rate.
One significant item of note that does not really appear in the data is the quality of the songs as they are processed through the media player. It is generally agreed that discontinuities in the generated sound, whether caused by the server's load or heavy network load conditions, is undesirable at best. While NSTL was able to count and measure the rebuffering needed during song play, there were a number of times when the music was chopped up by small pauses that essentially could not be measured. While this did not occur for RealPlayer G2, it happened upwards of twenty to thirty times per song when Windows Media Player operated at 16.8 Kbps, and, to a lesser degree, when it operated at 19.2 Kbps. NSTL believes that, by anyone's standard this is unacceptable.
This report was prepared under contract for RealNetworks, and contains highlights of tests executed by NSTL at the client's request, in accordance with parameters provided in the full testing report.
NSTL makes no recommendation or endorsement of any product. NSTL does not warrant or guarantee the adequacy, accuracy of completeness of the data provided in this report. NSTL MAKES NO WARRANTIES, EXPRESS OR IMPLIED, AS TO THE RESULTS TO BE OBTAINED BY ANY PERSON OR ENTITY FROM USE OF THIS DATA.
|
