> Hmm, I had to think about this a bit, but I do agree now tha > In the long run, this should probably be done by ALSA (or at least tuned > to some value reported by ALSA), but the patch makes a lot of sense to > me, and certainly more sense that the watermark value that we currently > stay away from the write index. Thanks. It looks like in some cases the information on DMA burst is available to the kernel, e.g. if you run lspci -vvnn you will see some information on what the HDaudio controller does. However in other cases this is not necessarily known, isn't reported, etc. In short it might be a while before this works in ALSA across all possible hardware. - Pierre
