here's a summary on the results. i discarded the 2.4.8-ll and 2.4.9-ll kernels because of the bad disk results (overruns), and just average proc and x11 results. i ranked the results of each catagory (1=*), and the numbers represent the latency difference from the best in the catagory. the nummbers on the right are the total rank (1st+1st+1st+2ns+4th=9), and the average for the five for that kernel. the 2.4.10 ll and pe kernels are clear winners overall, if the harddrive is used, but 2.4.12-ac3-pe is the clear winner if you are just working in ram (like me, with 1 gig of it).
alsa: copy read write proc x11 6.858 / 5 = 1.3716 (best average) -------------------------------------------------------------------- 10-ll 0.002 2 0.019 2 0.035 3 0.050 3 0.044 3 13 1.4016 10-pe 1.366 * 1.348 * 1.357 * 0.027 2 0.088 4 9 1.3946 12 0.054 4 0.019 2 0.033 2 0.549 5 0.256 5 18 1.5538 12-ac2 0.020 3 0.048 4 0.054 4 0.622 7 0.288 6 24 1.5780 12-ac2-ll 0.172 6 0.142 6 0.330 7 0.094 4 0.022 2 25 1.5236 12-ac3 0.068 5 0.049 5 0.315 6 0.594 6 0.312 7 28 1.6392 12-ac3-pe 0.199 7 0.227 7 0.078 5 1.366 * 1.421 * 21 1.4724 oss-emu: copy read write proc x11 13.353 / 5 = 2.6706 (best average) -------------------------------------------------------------------- 10-ll 0.002 2 0.013 4 0.098 4 0.027 3 0.305 6 19 2.7596 10-pe 2.655 * 0.007 2 0.149 6 2.685 * 0.095 2 12 2.7208 12 0.014 4 0.012 3 0.058 3 0.573 7 0.119 3 20 2.8258 12-ac2 0.044 5 0.096 6 2.667 * 0.639 5 0.214 4 21 2.8692 12-ac2-ll 0.057 6 0.104 7 0.139 4 0.076 4 0.280 5 26 2.8018 12-ac3 0.003 3 2.660 * 0.018 2 0.640 6 0.405 7 19 2.8838 12-ac3-pe 0.209 7 0.026 5 0.278 7 0.011 2 2.686 * 22 2.7754 overall: 10-pe 1st, 10-ll 2nd, 12-ac3-pe 3rd. disk: 10-pe 1st, 10-ll 2nd proc/x11: 12-ac3-pe 1st, 10-pe 2nd, 10-ll 3rd. 12-ac3-pe is bad in comparison on disk.
