I had stated earlier in this thread that I never saw any rebuffering issues in 7.3.3., but that was clearly tempting fate. It happened to me yesterday!
I was playing a FLAC-encoded album at the time and immediately switched on the buffering monitoring diagnostic screen saver on my SB3. I noticed that the buffer fullness bar was not able to get to 100%, but was doing a random walk and sometimes emptying completely, thus triggering a rebuffering pause. I then checked what else was going on around the network, and found that one of my sons had a Steam game downloading and was getting excellent download speed. Sufficient speed, in fact, to choke off my bandwidth for streaming FLACs. (I don't currently have any QoS rules in effect on my LINKSYS WRT54GL running Tomato firmware.) My curiosity was aroused to see what streaming bandwidth was being used by the SB3 when playing back FLAC across my wireless network. The attached chart shows what I could find by looking at the Tomato realtime bandwidth monitor. Looking at the graphic you can see three tracks being played in the space of 10 minutes. * The gaps before the peaks are when the end of the previous track is being played out of the SB3's buffer memory and the SB3 has not begun to pre-buffer the next track. * The peaks are when the next track is getting buffered as quickly as possible. * The steady(ish) 700kbps bandwidth is during normal track playback when the buffer is being consumed at the same rate it is being replenished and is close to 100% full. Notice that network conditions were gradually improving during this 10 minute window, indicated by the peak bandwidth during the buffering of the up-coming track. Ideally, this should be as narrow and tall as possible meaning the buffer was refilled in a short period of time. A vulnerability exists if the (supposedly) rapid buffering of the next track is throttled by network congestion. Under these conditions the weakest point is when you have emptied the previous track buffer but can't get sufficiently ahead of the game to gain some safety margin. It seems to me that to safely play back FLACs with an average bandwidth of 700kbps, you need some safe multiple of that bandwidth to be reliably available (at least x2, but preferably x4 or better). It would be better still if the SB3 would start buffering the next track earlier (i.e., alternating between two buffers so you can fill the next buffer while playing out the current one). That gap, followed by a peak is what will surely cause trouble in some marginal connectivity scenarios. +-------------------------------------------------------------------+ |Filename: tomato-bw.jpg | |Download: http://forums.slimdevices.com/attachment.php?attachmentid=7870| +-------------------------------------------------------------------+ -- dsdreamer ---------------------- "Dreamer, easy in the chair that really fits you..." ------------------------------------------------------------------------ dsdreamer's Profile: http://forums.slimdevices.com/member.php?userid=12588 View this thread: http://forums.slimdevices.com/showthread.php?t=65719 _______________________________________________ discuss mailing list discuss@lists.slimdevices.com http://lists.slimdevices.com/mailman/listinfo/discuss
