Interbench run on a quad core, so the default jobs on compile and burn are equal to the number of cpus (4). Default values for both schedulers. Note this is also a fairness benchmark as well as a latency benchmark, so lower across the board is not the be all and end all of the results. Using 2990639 loops per ms, running every load for 30 seconds Benchmarking kernel 2.6.30-bfs at datestamp 200909022012 --- Benchmarking simulated cpu of Audio in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0.004 +/- 0.00436 0.006 100 100 Video 0.008 +/- 0.00879 0.015 100 100 X 0.006 +/- 0.0067 0.014 100 100 Burn 0.005 +/- 0.00563 0.009 100 100 Write 0.005 +/- 0.00887 0.16 100 100 Read 0.006 +/- 0.00696 0.018 100 100 Compile 0.007 +/- 0.00751 0.019 100 100 --- Benchmarking simulated cpu of Video in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0.004 +/- 0.00451 0.007 100 100 X 0.006 +/- 0.00648 0.014 100 100 Burn 0.005 +/- 0.00581 0.008 100 100 Write 0.006 +/- 0.00808 0.138 100 100 Read 0.005 +/- 0.00556 0.024 100 100 Compile 0.176 +/- 1.68 16.7 100 99 --- Benchmarking simulated cpu of X in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0 +/- 0.000507 0.006 100 100 Video 0 +/- 0.00112 0.014 100 100 Burn 3.13 +/- 7.89 34 68.5 60 Write 0.006 +/- 0.0817 1 100 99.3 Read 0.003 +/- 0.0577 1 100 99.7 Compile 3.59 +/- 8.78 34 66.8 57.7 --- Benchmarking simulated cpu of Gaming in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU None 0 +/- 0 0 100 Video 0 +/- 0 0 100 X 0 +/- 0 0 100 Burn 24.8 +/- 25.3 32.8 80.1 Write 0 +/- 0 0 100 Read 0 +/- 0 0 100 Compile 26.8 +/- 27.4 42.3 78.8 Using 2990639 loops per ms, running every load for 30 seconds Benchmarking kernel 2.6.30 at datestamp 200909022029 --- Benchmarking simulated cpu of Audio in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0.005 +/- 0.00562 0.007 100 100 Video 0.003 +/- 0.00333 0.009 100 100 X 0.003 +/- 0.00409 0.01 100 100 Burn 0.004 +/- 0.00415 0.006 100 100 Write 0.005 +/- 0.00592 0.021 100 100 Read 0.004 +/- 0.00463 0.009 100 100 Compile 0.003 +/- 0.00426 0.014 100 100 --- Benchmarking simulated cpu of Video in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0.005 +/- 0.00564 0.009 100 100 X 0.004 +/- 0.00508 0.012 100 100 Burn 0.004 +/- 0.00416 0.006 100 100 Write 0.004 +/- 0.0105 0.326 100 100 Read 0.006 +/- 0.00892 0.15 100 100 Compile 0.029 +/- 1.07 45.2 99.9 99.9 --- Benchmarking simulated cpu of X in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met None 0 +/- 0.00051 0.006 100 100 Video 0.003 +/- 0.0577 1 100 99.7 Burn 11 +/- 26.1 115 42.1 34.5 Write 0.003 +/- 0.0577 1 100 99.7 Read 0.006 +/- 0.0817 1 100 99.3 Compile 0.33 +/- 2.37 21 98.7 96.3 --- Benchmarking simulated cpu of Gaming in the presence of simulated --- Load Latency +/- SD (ms) Max Latency % Desired CPU None 0 +/- 0 0 100 Video 0 +/- 0 0 100 X 0.02 +/- 0.35 6.06 100 Burn 0.3 +/- 3 29.9 99.7 Write 0 +/- 0.000847 0.01 100 Read 0 +/- 0.000577 0.01 100 Compile 6.17 +/- 24.4 130 94.2 Comments: The "Burn" load is 4 continuous cpu consuming processes. The "Compile" load is 4 continuous cpu consuming processes running concurrently with a continuous write and continuous read. Theoretically the disk i/o should not interfere with the cpu consumption and give approximately the same cpu usage. With BFS in both the X and gaming runs, the amount of cpu given to burn vs compile is almost identical in both causes (67% given to X and 80% given to gaming). This shows very good fairness maintained for the background cpu user, while maintaining a reasonable amount of cpu for the foreground (X and gaming) consistent with what would be fair. The gaming test uses a fully cpu bound foreground task against 4 fully cpu bound background tasks, so it should get 1/5 of the total cpu, and on 4 cpus that is 1/5 of 4, so 0.8 (80%, spot on). The X test uses a sweep of cpu from 0 to 100% and back again, and yes, the cpu proportion is correct for bfs yet again. Disturbing things about the cfs values are the fluctuating cpu usage of burn vs compile, the uneven/unfair cpu distribution, and the wildly varying maximum latencies. THE MICROSECOND LATENCIES ARE IRRELEVANT AND WAY BELOW HUMAN PERCEPTION. Only anything larger than about 5ms really matters.