Hi, On 11/15/2010 07:16 AM, Zooko O'Whielacronx wrote:
> In fact, could you run this a few times and collect multiple > measurements? Maybe it was just a transient "fluke" -- some kind of > noise on your network or one of your machines interfered with the > download. > > If it wasn't a fluke I wonder if it could be the effect of issue > #1223, which is fixed in trunk (== 1.8.1 release candidate). #1223 > should have the impact of slowing downloads, but I'm not sure by how > much. Oh, sorry I made a mistake in the previous mail, the version this server is running is 1.8.0-r4784 which is only 9 patches away from trunk. So the fix for #1223 was actually already in. It seems that encoding parameter has a huge impact on download speed. This is the average time in seconds across three runs that a 100 MB file download takes depending on the encoding parameters. 1-of-3 58 2-of-3 63 3-of-3 58 10-of-30 210 20-of-60 394 > Anyway, is the overall performance good enough for your current uses? As usual, I would love to see better performances in those use cases. But it's not currently a show-stopper for me. François _______________________________________________ tahoe-dev mailing list [email protected] http://tahoe-lafs.org/cgi-bin/mailman/listinfo/tahoe-dev
