i just finished testing several different patched and unpatched linux kernels, and was surprised by the results: all basically the same.
alsa range = 33.4% to 33.8% 1.336 ms to 1.352 ms diff = 0.016 ms oss-emu range = 91.6% to 91.9% 3.664 ms to 3.676 ms diff = 0.012 ms as you can see, the differences are negligible and probably irrelevant if more tested were done, and a standard 2.4.12 kernel is just as good as any patched kernel. one note: when i did a real-world test of latency by looping a signal through my card and dividing the difference in time by 2 (because of input and output latency), i got 3.675 ms latency using kernel 2.4.8-ll with oct 1 alsa-cvs, and that corresponds with the oss-emu latency for some reason instead of the alsa latency. i don't know why, and i'll test this on other kernels soon. my system is an amd athlon 900 mhz running at 1006 mhz with redhat 7.1 and an m-audio delta audiophile 2496 soundcard. i used alsa-cvs from october 20 and takashi iwai's modified version of latencytest0.42+alsa. the kernels tested are 2.4.8-ll, 2.4.9-ll, 2.4.10-ll, 2.4.10-pe, 2.4.12, 2.4.12-ac2, 2.4.12-ac2-ll, 2.4.12-ac3, and 2.4.12-ac3-pe. ll = low-latency pe = pre-emptive ac2 and ac3 are alan cox's patches latencytest was ran as follows: ./latencytest -t alsa -d hw -q none 3 256 ./latencytest -t oss -q none 3 256 -dave details below: ----------------------------------------------------------------------- sample rate = 48000 NUMBER of OVERRUNS = 0 PIXEL_PER_MS=112 fragment latency = 1.333333 ms buffer latency= 4.000 ms fragment size=256 total buffer size=768 cpu_load= 80.0000% cpu latency = 1.066667 ms with alsa, 100% of samples are within 1ms of buffer used alsa-cvs from oct 20, oss = alsa's oss-emu ====================2.4.10-ll==================== output type = ALSA max latency=1.3 ms factor=33.4 % of buffer output type = OSS max latency=3.7 ms factor=91.6 % of buffer 1MS factor=83.467993 2MS factor=99.937850 ====================2.4.10-pe==================== output type = OSS max latency=3.7 ms factor=91.9 % of buffer 1MS factor=83.367698 2MS factor=99.965636 output type = ALSA max latency=1.3 ms factor=33.6 % of buffer ====================2.4.12-ac2-ll==================== output type = ALSA max latency=1.3 ms factor=33.7 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.400810 2MS factor=99.932524 ====================2.4.12-ac2==================== output type = ALSA max latency=1.3 ms factor=33.4 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.372922 2MS factor=99.966067 ====================2.4.12-ac3==================== output type = ALSA max latency=1.4 ms factor=33.8 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.361392 2MS factor=99.971942 ====================2.4.12-ac3-pe==================== output type = ALSA max latency=1.3 ms factor=33.5 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.390216 2MS factor=99.943117 ====================2.4.12==================== output type = ALSA max latency=1.3 ms factor=33.4 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.372828 2MS factor=99.960506 ====================2.4.8-ll==================== output type = ALSA max latency=1.3 ms factor=33.6 % of buffer output type = OSS max latency=3.7 ms factor=91.6 % of buffer 1MS factor=83.402836 2MS factor=99.972199 ====================2.4.9-ll==================== output type = ALSA max latency=1.3 ms factor=33.4 % of buffer output type = OSS max latency=3.7 ms factor=91.7 % of buffer 1MS factor=83.385093 2MS factor=99.948240 _______________________________________________ Alsa-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-devel
