2012/8/9 Renaud <[email protected]>: > Hi, Hi!
> I've been testing Liquidsoap 1.0.0 for some time now. > I've made some tests with the latest version from the sourceforge > repository (1.0.1) and I've been comparing it to the Debian Package on > the official wheezy repositories (1.0.0-4) and I noticed a huge > performance gap between those two. > With the 1.0.1 I compiled, I could barely run 150 liquidsoap processes, > with the 1.0.0-4 from the Debian repos, I could run 245 easily. > I then tried the 1.0.1 package on the Debian Sid repositories (because I > I thought my compilation was at fault) and I noticed the exact same > performance gap. > > Is that behavior considered as normal regarding the code changes between > those versions ? > Do you need more data from me (or me to do specific tests) to try and > understand this ? I don't think that liquidsoap 1.0.1 introduced any change that would induce a performance loss compared to 1.0.0. I would instead look for the architecture, compiler and encoding libraries that you are using. In particular, most processing time it spent encoding, so you should look first at (I suppose) the libmp3lame used in each case. What version? What provider? What compilation options? etc.. BTW, you may also want to test our new fixed-point mp3 encoder: https://github.com/savonet/shine It is supported in the latest HG repository and I suspect that it can lead to small improvements that would be significant when scaling up. Romain ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Savonet-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/savonet-users
