|
From: <be...@ga...> - 2004-06-05 22:09:28
|
Hi, first of all, very nice work Rui, bravo ! I added a screenshot ryanpg sent me on our screenshots page: http://www.linuxsampler.org/screenshots.html Plus I made some updates in the features page and added developers names if someone that contributed to LS is missing please let me know. If you want your email addressed added near the names (and get more spam :) ) le me know. At time Marek produced a few gifs with the email address of me,Christian and Steve (to avoid spam) but I don't have the time to make such gifs so let me know if I should add I did not try the latest GUI yet so I cannot comment on the glitches Mark described, anyway I'll try to summarize a few things: the stream usage bar should show the buffer percentage fill of the least filled buffer. If you play lots of notes it should constantly jump to 0 and then quickly go up to almost 100% (if no new notes appear). When notes disappear I guess the stream bar remains on the last position where at least one stream was active. as for stream / voice counters, they should almost always show more or less the same value (eg when you fire up a voice a stream is created, when a voice disappears the associated stream disappears too). Just try to play one single note. voice and streams should be 1 and then when you release the note (after the EG has faded out the note) or after the note decays naturally (eg in case of non looped sounds) the voice/stream count number should go to 0. If that does not happen then there is a bug in either the LS engine not exporting the correct value via LSCP or the Qt GUI misinterpreting it. I will try to debug it in the next days but I guess someone will fix this sooner :) Meanwhile I did not sit idle, more benchmarks, ported most of the low level code including optimized file I/O to win32 so which ease a future LS port on Windows. I wrote a benchmark that simulates the stress that a softsampler puts on the disk subsystem. The code is optimized for both linux and win32 (standard win32 I/O calls (buffered) are slow). I'm thinking to wrap it in a GUI and release a crossplatform "sampler benchmark" that would work on Linux, Windows and Mac OS X. Future LS versions will come with such a benchmark builtin so that the user can tune the engine to squeeze out the most of his hardware. But the question is would audio users (in general) like such a benchmark ? I think it would be handy since it would allow people to get an idea how many voices/audio tracks one can get out of his box, how big buffers you need, what's the real world disk seeking performance etc, regardless if a Mac, a Linux box or a Windows box. Ok you could run a softsampler on each box and stress it till it chokes you box but not all softsamplers are crossplatform and often users don't want to buy a softsampler only for benchmarking purposes or after they discovered that their box is not up to the task and they cannot afford a new one. Plus the benchmark could be hosted on this site and if it gets widely used it would give us a bit more visibility :) Anyway I'm pretty excited too, the GUI will bring fresh wind (developers, users, lurkers) to LS :) cheers, Benno http://www.linuxsampler.org ------------------------------------------------- This mail sent through http://www.gardena.net |