Hi Stefan, At Tue, 02 Oct 2001 20:21:05 +0200, stef wrote: > > Takashi Iwai wrote: > > > > > > And if i add some more midi traffic, the sequencer queue slows down. I > > > > > [...] > > > > Hmm.. it's very strange, and not intended at all.. > > > > Do you mean that the calculated value is actually slow down or the > > > > actually playing tempo is slow down? > > > > > > The played music itself slows down. > > > > Ok, then could you tell me the environment you've tested? > > Five songs to a single MTPAV port for each? > > PII-400, Linux 2.2.15pre5, alsa-driver-0.9.0beta8, alsa-lib-0.9.0beta8 > Eight identical songs to one queue to... > ...eight mtpav hardware ports: 140bpm->119bpm slowdown > ...one mtpav hardware port: 140bpm->138bpm slowdown > ...ymfpci midi: no slowdown
Ok, it looks like not directly related to scheduling of sequencer but port access of MTPAV. As I tested also (modified) playmidi1 with shared queues, it works ok. I tested up to 20 clients and no slowdown of playback with emu10k1 wavetable. If it's mtpav hardware, especially midi output problem.. hmm.. so far there is no check that the write buffer of mtpav (hardware) is full or not. Possibly this may block the function somehow. > ...eight mtpav hardware ports, but with rtctimer 1024: same as with system > timer > ...eight mtpav hardware ports, but with rtctimer 128: nearly no slowdown Do you mean 1024Hz and 128Hz? If so, it's interesting that slower interrupts prevent slowdown.. > > So far as I tested, i couldn't reproduce the problem. > > I tested a port sblive wavetable, and six other ports sent to external > > MIDI unconnected (thus discarded) with the standard timer. > > Possibly dependent on the midi file to play? > > Yes, of course! The more notes per second, the slower the queue becomes. > > Next week i'll send you my midi player program together with the midi files i > used for testing. But for now i solved my problem by using only one > mtpav port with rtctimer. The other output port i need comes from the > ymf744b chip inside my sony vaio 505 laptop. I soldered a midi jack to > the unused midi tx pin. > > When i come back next week, i definiteley start to reverse-engineer MOTU's > parallel port protocol to fix the port routing problems of mtpav. > Does Michael T. Mayers (who wrote card-mtpav) still exist? I was trying > to contact him about a year ago, but no response. Do you maintain the code > now? Well, i rewrote and cleaned up the code but i myself don't own the hardware, so it's fairly difficult to maintain it correctly :-< Takashi _______________________________________________ Alsa-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-devel
