On Fri, 14 Feb 2003 01:31:44 -0800
Andrew Morton <akpm@digeo.com> wrote:
> The main obective of the anticipatory scheduler is not really to improve
> interactivity. It is to increase throughput. Nick is showing some
> impressive benchmark results with this now. Some benchmarking of the
> non-contest variety would be appreciated.
[...]
> anticipatory_io_scheduling.patch
> Subject: [PATCH] 2.5.59-mm3 antic io sched
>
> ant-sched-9feb.patch
> anticipatory scheduler fix
>
> ant-sched-12feb.patch
> Anticipatory scheduler tuning
(I applied those three)
Those are tiobench's results:
in both:
No size specified, using 510 MB
Unit information
================
File size = megabytes
Blk Size = bytes
Rate = megabytes per second
CPU% = percentage of CPU used during the test
Latency = milliseconds
Lat% = percent of requests that took longer than X seconds
CPU Eff = Rate divided by CPU% - throughput per cpu load
diff between both
Sequential reads
File Blk Num Avg Maximum Lat% Lat% CPU
Identifier Size Size Thr Rate (CPU%) Latency Latency >2s >10s Eff
---------------------------- ------ ----- --- ------ ------ --------- ----------- -------- -------- -----
-2.5.60-mm2 510 4096 1 21.23 8.264% 0.183 111.66 0.00000 0.00000 257
-2.5.60-mm2 510 4096 2 7.33 3.506% 1.061 196.42 0.00000 0.00000 209
-2.5.60-mm2 510 4096 4 5.39 2.593% 2.874 275.79 0.00000 0.00000 208
-2.5.60-mm2 510 4096 8 4.45 2.125% 6.846 1008.67 0.00000 0.00000 209
+2.5.60-mm2-ant 510 4096 1 42.07 17.81% 0.092 57.22 0.00000 0.00000 236
+2.5.60-mm2-ant 510 4096 2 22.84 12.08% 0.332 308.44 0.00000 0.00000 189
+2.5.60-mm2-ant 510 4096 4 17.40 9.298% 0.873 695.78 0.00000 0.00000 187
+2.5.60-mm2-ant 510 4096 8 11.62 6.064% 2.484 1334.89 0.00000 0.00000 192
Random Reads
File Blk Num Avg Maximum Lat% Lat% CPU
Identifier Size Size Thr Rate (CPU%) Latency Latency >2s >10s Eff
---------------------------- ------ ----- --- ------ ------ --------- ----------- -------- -------- -----
-2.5.60-mm2 510 4096 1 0.75 0.775% 5.190 57.26 0.00000 0.00000 97
-2.5.60-mm2 510 4096 2 0.62 0.611% 12.433 122.61 0.00000 0.00000 102
-2.5.60-mm2 510 4096 4 0.54 0.524% 28.542 534.94 0.00000 0.00000 102
-2.5.60-mm2 510 4096 8 0.49 0.527% 58.520 941.25 0.00000 0.00000 94
+2.5.60-mm2-ant 510 4096 1 0.71 0.742% 5.522 62.66 0.00000 0.00000 95
+2.5.60-mm2-ant 510 4096 2 0.63 0.574% 12.137 274.13 0.00000 0.00000 109
+2.5.60-mm2-ant 510 4096 4 0.58 0.548% 24.428 282.18 0.00000 0.00000 106
+2.5.60-mm2-ant 510 4096 8 0.63 0.549% 41.976 596.08 0.00000 0.00000 115
Sequential Writes
File Blk Num Avg Maximum Lat% Lat% CPU
Identifier Size Size Thr Rate (CPU%) Latency Latency >2s >10s Eff
---------------------------- ------ ----- --- ------ ------ --------- ----------- -------- -------- -----
-2.5.60-mm2 510 4096 1 26.59 38.97% 0.119 559.77 0.00000 0.00000 68
-2.5.60-mm2 510 4096 2 16.14 29.96% 0.390 6813.03 0.00306 0.00000 54
-2.5.60-mm2 510 4096 4 9.55 18.37% 1.249 16573.28 0.01615 0.00154 52
-2.5.60-mm2 510 4096 8 6.35 11.40% 4.342 45796.21 0.02093 0.01550 56
+2.5.60-mm2-ant 510 4096 1 33.87 49.55% 0.093 370.18 0.00000 0.00000 68
+2.5.60-mm2-ant 510 4096 2 12.56 24.08% 0.531 9524.84 0.00613 0.00000 52
+2.5.60-mm2-ant 510 4096 4 6.68 13.08% 2.071 20422.31 0.03153 0.00231 51
+2.5.60-mm2-ant 510 4096 8 6.90 13.45% 3.086 41464.78 0.02790 0.01163 51
Random Writes
File Blk Num Avg Maximum Lat% Lat% CPU
Identifier Size Size Thr Rate (CPU%) Latency Latency >2s >10s Eff
---------------------------- ------ ----- --- ------ ------ --------- ----------- -------- -------- -----
-2.5.60-mm2 510 4096 1 0.99 1.050% 0.140 33.67 0.00000 0.00000 95
-2.5.60-mm2 510 4096 2 1.02 1.138% 0.043 11.72 0.00000 0.00000 90
-2.5.60-mm2 510 4096 4 1.02 1.423% 0.082 99.74 0.00000 0.00000 72
-2.5.60-mm2 510 4096 8 0.99 1.329% 0.074 74.51 0.00000 0.00000 74
+2.5.60-mm2-ant 510 4096 1 0.96 0.917% 0.027 1.03 0.00000 0.00000 105
+2.5.60-mm2-ant 510 4096 2 1.00 1.472% 0.044 2.42 0.00000 0.00000 68
+2.5.60-mm2-ant 510 4096 4 0.95 1.367% 0.044 1.96 0.00000 0.00000 69
+2.5.60-mm2-ant 510 4096 8 0.99 1.431% 0.079 75.46 0.00000 0.00000 69
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Feb 15 2003 - 22:01:04 EST