Le Fri, 09 Jun 2006 15:32:45 -0400, Lee Revell <[EMAIL PROTECTED]> a écrit :
> On Fri, 2006-06-09 at 20:53 +0200, Dominique Michel wrote: > > Le Fri, 09 Jun 2006 11:39:27 +0200, > > Takashi Iwai <[EMAIL PROTECTED]> a écrit : > > > > > At Thu, 8 Jun 2006 23:47:45 +0200, > > > Dominique Michel wrote: > > > > > > > > Le Thu, 08 Jun 2006 12:46:21 +0200, > > > > Takashi Iwai <[EMAIL PROTECTED]> a écrit : > > > > > > > > > At Sat, 3 Jun 2006 00:42:49 +0200, > > > > > Dominique Michel wrote: > > > > > > > > > > > > Le Fri, 2 Jun 2006 15:34:43 +0200, > > > > > > Dominique Michel <[EMAIL PROTECTED]> a écrit : > > > > > > > > > > > > > Hi all, > > > > > > > > > > > > > > I have an audigy2, run a 2.6.16-rt22 kernel and the alsa drivers > > > > > > > from this kernel. So, the right driver for this card is emu10k1. > > > > > > > > > > > > > > When I load a soudfont of 61 Mb in the card, asxfload -M tell me > > > > > > > att I have about 70Mb left. The biggest font I can load is about > > > > > > > 131Mb. > > > > > > > My problem is at I want to load per default a 238 Mb soudfont in > > > > > > > the card. I have 1.5Gb ram in the box. Is it a way to load it in > > > > > > > the card? > > > > > > > I rtfm, but I don't find anything. The audigy need unfragmented > > > > > > > memory to load a soundfont. Is it a way to reserve a amount of > > > > > > > unfragmented memory for this card at boot time? > > > > > > > > > > > > > > I see on the alsa website at the driver for my card have changed > > > > > > > with the last alsa drivers. > > > > > > > > > > > > Not true. No matter, I just installed the last stable version of > > > > > > all the alsa packages and get exactly the same result with asfxload. > > > > > > > > > > You need to increase the value of max_buffer_size module option. > > > > > The default is 128MB. See ALSA-Configuration.txt. > > > > > > > > > > > > > > > Takashi > > > > > > > > Thank you for the answer. > > > > I added "options snd-emu10k1 max_buffer_size=256 max_synth_voices=256" > > > > in /etc/modules.d/alsa and I can load and use the soundfont. > > > > But when doing > > > > > > > > # cat /proc/asound/card0/wavetableD1 > > > > Device: Emu10k1 > > > > Ports: 4 > > > > Addresses: 17:0 17:1 17:2 17:3 > > > > Use Counter: 0 > > > > Max Voices: 64 > > > > Allocated Voices: 0 > > > > Memory Size: 268435456 > > > > Memory Available: 26109738 > > > > Allocated Blocks: 1716 > > > > SoundFonts: 1 > > > > Instruments: 9002 > > > > Samples: 1715 > > > > Locked Instruments: 9002 > > > > Locked Samples: 1715 > > > > > > > > it show at I still have only 64 voices. > > > > Is it normal? > > > > > > Yes. Emu10k1 has max 64 voices playable at the same time. > > > > > > What is the meaning of this option if the card have 64 hardware voices and > > the default for the module is 64 voices? I don't see why we can want less > > as this default value, but maybe at I am wrong- > > > > Because those 64 voices are shared between the wavetable synth and PCM. I found a description for the emu10k1 at http://www.osdever.net/downloads/docs/m2049.zip (to read it, rename the file to m2049.pdf and open with xpdf) It is not a data sheet but we can see at the wavetable synth is a true 64 channel synth. The mixing is done in the mixer and effects processor. The PCM signals are routed directly from the main memory and the ADC to the mixer. It is one channel that route the waveforms (soundfonts) in the main memory directly to the synth. After traitment in the synth, the 64 voices are routed to the mixer. We have 16 pcm output at the output of the mixer. We can see those 16 pcm mixer channels in qjackctl with the appropriate settings, but the 64 voices of the synth doesn't appear. They must be automatiquely mapped in the mixer to a stereo output. I believe at if I want to change the mapping of those 64 synth voices in the mixer, I must use programs as ld10k1 and qlo10k1. But in practice, a stereo output is enough for a MIDI performance, and a combination of muse (or rosegarden), ardour and jamin will be much more easier to use. The EMU10K1 use a special way to mapp the memory for the soundfonts. They are stored in the main memory and loaded by a pci maste and virtual memory mapper into the chip when needed by the synth. I am wondering if the max_synth_voices parameter can affect this mapping and how? I was reading too in a windows forum at it is a hardware limitation when the synth is loading a soundfont between 2 samples. They was talking of 32 MB per saamples. In my system, if I use a little soudfont, the synth work very well with a very little latency, but when using a hugue soundfont, I must increase the latency if I want to get no xrun when the synth is loading some soundfonts.So, I believe at, by increasing the latency, the synth is abble to load very big soundfonts, even bigger as 32 MB per sample. Is it true? Dominique > > Lee > > > Best, > > Dominique > > > > > > > > > I have other questions too. Is it a limitation beside the amount of ram > > > > in my system with max_buffer_size? > > > > > > The hardware limitation is 2GB (or 1GB? I forgot). > > > > > > > Is it other limitation such as hardare limitation with this card? > > > > > > The kernel might be not able to allocate the memory if you have > > > already used for other purposes. > > > > > > > > > Takashi > > > > > > _______________________________________________ > > Alsa-user mailing list > > Alsa-user@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/alsa-user > _______________________________________________ Alsa-user mailing list Alsa-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/alsa-user
