These are patches designed to improve system responsiveness and interactivity
with specific emphasis on the desktop, but suitable to any workload.
Apply to 2.6.35:
http://www.kernel.org/pub/linux/kernel/people/ck/patches/2.6/2.6.35/2.6.35-ck1/patch-2.6.35-ck1.bz2
Broken out tarball:
http://www.kernel.org/pub/linux/kernel/people/ck/patches/2.6/2.6.35/2.6.35-ck1/2.6.35-ck1-broken-out.tar.bz2
Discrete patches:
http://www.kernel.org/pub/linux/kernel/people/ck/patches/2.6/2.6.35/2.6.35-ck1/patches/
Web:
http://kernel.kolivas.org
All -ck patches:
http://www.kernel.org/pub/linux/kernel/people/ck/patches/
Each discrete patch contains a brief description of what it does at the top of
the patch itself.
The most significant change is an updated BFS cpu scheduler. See:
http://lkml.org/lkml/2010/8/6/620
The rest is a resync only since 2.6.34-ck1.
Patch series:
2.6.35-sched-bfs-323.patch
sched-add-above-background-load-function.patch
mm-make_swappiness_really_mean_it.patch
mm-zero_swappiness.patch
mm-enable_swaptoken_only_when_swap_full.patch
mm-drop_swap_cache_aggressively.patch
mm-kswapd_inherit_prio-1.patch
mm-background_scan.patch
mm-idleprio_prio-1.patch
mm-lru_cache_add_lru_tail.patch
kconfig-expose_vmsplit_option.patch
hz-default_1000.patch
hz-no_default_250.patch
hz-raise_max.patch
preempt-desktop-tune.patch
cpufreq-bfs_tweaks.patch
ck1-version.patch
Enjoy!
--
-ck
On Sat, 7 Aug 2010 10:39:31 am Con Kolivas wrote:
> These are patches designed to improve system responsiveness and
> interactivity with specific emphasis on the desktop, but suitable to any
> workload.
>
>
> Apply to 2.6.35:
> http://www.kernel.org/pub/linux/kernel/people/ck/patches/2.6/2.6.35/2.6.35-
>ck1/patch-2.6.35-ck1.bz2
As discussed on the BFS release announce, here are some interbench results,
performed on a lower spec machine. Hyperthreading was enabled, so this was an
SMP config with 2 CPUs. Use a fixed width font to make it readable.
See you all next major 3 point kernel release.
Comparison of 2.6.35 v 2.6.35-ck1 on a pentium 4HT with 1GB ram.
Note the significantly higher number of deadlines met on ck1, and when
deadlines are not met, the worst case latencies are mostly better. These
are the two most important parameters to compare. Asterisk placed where
values are significantly better. >= 5% or 5ms is considered significant.
Using 2043662 loops per ms, running every load for 30 seconds
Benchmarking kernel 2.6.35-ck1 at datestamp 201008082239
--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 0.0 +/- 0.0 0.0 100 100
Video 0.0 +/- 0.0 0.0 100 100
X 0.0 +/- 0.0 0.0 100 100
Burn 0.0 +/- 0.0 0.0 100 100
Write 0.0 +/- 0.0 0.4 * 100 100
Read 0.0 +/- 0.0 0.1 100 100
Compile 0.0 +/- 0.0 0.3 * 100 100
Memload 0.0 +/- 0.0 0.1 * 100 100 *
--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 0.0 +/- 0.0 0.0 * 100 100 *
X 0.0 +/- 0.0 0.0 * 100 100 *
Burn 0.0 +/- 0.0 * 0.0 * 100 100 *
Write 0.1 +/- 1.5 22.7 * 100 * 99.3 *
Read 0.0 +/- 0.0 0.1 * 100 100 *
Compile 2.1 +/- 6.0 * 23.7 * 100 * 87.5 *
Memload 0.2 +/- 1.7 23.7 100 * 99 *
--- Benchmarking simulated cpu of X in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 0.5 +/- 2.9 * 39.0 * 89.3 * 86.4 *
Video 17.3 +/- 31.1 80.0 * 32.3 22.8
Burn 30.3 +/- 47.1 * 104.0 * 23.1 * 13.4 *
Write 5.6 +/- 13.2 54.0 * 60.1 51.5
Read 1.6 +/- 5.5 55.0 77.7 * 72.1 *
Compile 43.1 +/- 62.9 * 150.0 * 18 8.63
Memload 2.7 +/- 7.4 * 35.0 * 71.6 * 63.7 *
--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU
None 72.4 +/- 94.7 166.6 58
Video 76.1 +/- 76.5 * 82.0 * 56.8 *
X 67.6 +/- 74.1 * 104.3 * 59.7 *
Burn 103.3 +/- 103.9 * 107.5 * 49.2 *
Write 38.4 +/- 41.3 81.8 72.3
Read 13.1 +/- 14.1 * 45.2 88.4
Compile 125.9 +/- 127.3 * 215.3 * 44.3
Memload 30.3 +/- 34.7 * 145.0 * 76.7
Using 2043662 loops per ms, running every load for 30 seconds
Benchmarking kernel 2.6.35 at datestamp 201008082220
--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 0.0 +/- 0.0 0.0 100 100
Video 0.0 +/- 0.0 0.0 100 100
X 0.0 +/- 0.0 0.0 100 100
Burn 0.0 +/- 0.0 0.0 100 100
Write 0.1 +/- 0.3 5.0 100 100
Read 0.0 +/- 0.0 0.1 100 100
Compile 0.1 +/- 0.8 11.0 100 100
Memload 0.1 +/- 2.0 50.0 99.8 99.8
--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 0.2 +/- 1.9 16.7 100 98.6
X 1.8 +/- 5.4 16.7 100 89.5
Burn 11.5 +/- 13.9 16.7 100 31.1
Write 1.4 +/- 4.9 40.3 99.9 92.1
Read 0.2 +/- 1.8 16.7 100 98.9
Compile 9.2 +/- 14.0 46.3 89.9 49.2
Memload 1.2 +/- 4.4 22.7 100 93.3
--- Benchmarking simulated cpu of X in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU % Deadlines Met
None 5.9 +/- 15.3 104.0 57.8 48.8
Video 13.4 +/- 26.1 100.0 36.4 27.7 *
Burn 40.4 +/- 60.6 150.0 18.8 9.84
Write 7.0 +/- 16.0 84.0 58.5 47.4
Read 2.5 +/- 6.5 30.0 * 74.8 66.1
Compile 55.0 +/- 76.4 288.0 18.1 7.28
Memload 8.7 +/- 26.7 290.0 51.6 43.6
--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load Latency +/- SD (ms) Max Latency % Desired CPU
None 32.7 +/- 52.2 * 167.0 75.4 *
Video 114.6 +/- 116.5 161.0 46.6
X 88.3 +/- 96.6 166.9 53.1
Burn 139.1 +/- 141.5 167.0 41.8
Write 33.1 +/- 36.2 * 83.2 75.1
Read 20.1 +/- 20.6 51.3 83.3
Compile 138.8 +/- 142.9 312.8 41.9
Memload 36.7 +/- 46.5 197.4 73.1
--
-ck