You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Stefan Sperling <st...@elego.de> on 2011/04/07 13:29:01 UTC
some more performance benchmarks
Neels has set up his performance benchmarks to run on the svn VM
we have at apache.org.
The tests only use ra_Local and are supposed to measure wc-ng
performance, not httpv2. They focus on edge cases and are thus a
good addition to Mark Phippard's test which focus more on real world usage.
Below are the results of the latest run.
They're probably a bit hard to decipher as is. The plan is to make the
scripts produce more easily digestable results and then send results
to dev@ on a weekly basis.
As far as I can tell, these numbers indicate that trunk is on-par
with 1.6.x except for delete operations.
------8<---------------------------------------------------------------
/home/neels/svnbench/20110407-002451
Started at Thu Apr 7 00:24:51 UTC 2011
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 5x5_1.6
Running svn benchmark in /tmp/tmpIyTMUN
dir levels: 5; new files and dirs per leaf: 5; run 1 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:05:30.543596
N min max avg operation (unit is seconds)
1 330.544 330.544 330.544 TOTAL RUN
139 0.008 1.703 0.021 add
2 0.013 7.599 3.806 checkout
8 6.181 39.123 14.115 commit
1 0.030 0.030 0.030 copy
1 4.294 4.294 4.294 delete
2 8.935 10.962 9.949 merge
164 0.009 0.011 0.010 mkdir
1091 0.009 2.935 0.016 prop mod
41 0.009 0.496 0.038 propdel
2076 0.008 0.221 0.009 proplist
4235 0.009 1.140 0.012 propset
2 0.009 0.010 0.009 resolve
2 0.587 0.919 0.753 resolved
14 0.179 0.678 0.443 status
1 8.425 8.425 8.425 switch
15 0.151 13.000 5.016 update
Running svn benchmark in /tmp/tmp0eRvwQ
dir levels: 5; new files and dirs per leaf: 5; run 2 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:06:39.571401
N min max avg operation (unit is seconds)
2 330.544 399.571 365.057 TOTAL RUN
321 0.008 3.322 0.025 add
4 0.013 11.839 4.866 checkout
16 6.181 45.586 15.683 commit
2 0.030 0.507 0.268 copy
2 4.294 7.881 6.088 delete
4 6.898 21.264 12.015 merge
327 0.009 0.047 0.010 mkdir
2129 0.009 2.935 0.016 prop mod
85 0.009 0.496 0.030 propdel
4156 0.008 0.221 0.009 proplist
8375 0.009 1.592 0.013 propset
4 0.009 0.010 0.010 resolve
4 0.370 0.919 0.609 resolved
28 0.179 0.781 0.472 status
2 8.425 14.073 11.249 switch
30 0.150 16.130 5.375 update
Running svn benchmark in /tmp/tmplxnWwz
dir levels: 5; new files and dirs per leaf: 5; run 3 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:07:38.478860
N min max avg operation (unit is seconds)
3 330.544 458.479 396.198 TOTAL RUN
473 0.008 3.686 0.029 add
6 0.013 17.112 6.098 checkout
24 6.181 45.586 16.385 commit
3 0.030 0.894 0.477 copy
3 2.313 7.881 4.829 delete
6 6.898 21.264 14.159 merge
503 0.009 2.464 0.017 mkdir
3201 0.009 2.935 0.017 prop mod
142 0.009 0.496 0.035 propdel
6294 0.008 0.381 0.009 proplist
12550 0.009 2.034 0.014 propset
6 0.009 0.010 0.009 resolve
6 0.342 1.900 0.780 resolved
42 0.179 0.839 0.466 status
3 8.425 14.363 12.287 switch
45 0.150 24.437 5.951 update
Running svn benchmark in /tmp/tmpj3o2lh
dir levels: 5; new files and dirs per leaf: 5; run 4 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:06:54.330198
N min max avg operation (unit is seconds)
4 330.544 458.479 400.731 TOTAL RUN
653 0.008 3.751 0.030 add
8 0.013 17.112 6.229 checkout
32 6.181 45.586 16.641 commit
4 0.030 0.894 0.370 copy
4 2.313 7.881 5.106 delete
8 6.898 21.264 13.998 merge
683 0.009 2.464 0.016 mkdir
4273 0.009 2.935 0.017 prop mod
182 0.009 0.496 0.034 propdel
8401 0.008 0.381 0.009 proplist
16686 0.009 3.145 0.014 propset
8 0.009 0.010 0.009 resolve
8 0.342 2.710 1.006 resolved
56 0.179 0.958 0.472 status
4 8.425 14.363 12.247 switch
60 0.150 24.437 5.991 update
Running svn benchmark in /tmp/tmpeLkKCX
dir levels: 5; new files and dirs per leaf: 5; run 5 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:06:26.682095
N min max avg operation (unit is seconds)
5 330.544 458.479 397.921 TOTAL RUN
805 0.008 3.751 0.030 add
10 0.013 17.112 6.617 checkout
40 6.181 45.586 16.500 commit
5 0.030 0.894 0.390 copy
5 2.313 7.881 4.596 delete
10 6.898 21.264 13.349 merge
829 0.009 2.464 0.015 mkdir
5303 0.009 2.935 0.017 prop mod
215 0.009 0.496 0.033 propdel
10530 0.008 0.381 0.009 proplist
20743 0.009 3.145 0.014 propset
10 0.009 0.010 0.009 resolve
10 0.334 2.710 0.925 resolved
70 0.179 0.958 0.472 status
5 8.425 14.363 11.515 switch
75 0.150 24.437 6.072 update
Running svn benchmark in /tmp/tmpda5G0S
dir levels: 5; new files and dirs per leaf: 5; run 6 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:09:21.405688
N min max avg operation (unit is seconds)
6 330.544 561.406 425.169 TOTAL RUN
955 0.008 3.820 0.031 add
12 0.013 20.311 7.208 checkout
48 6.181 53.167 17.509 commit
6 0.030 0.894 0.436 copy
6 2.313 7.881 4.588 delete
12 6.898 21.264 14.617 merge
974 0.009 2.464 0.015 mkdir
6384 0.009 2.935 0.018 prop mod
252 0.009 0.496 0.033 propdel
12632 0.008 0.381 0.009 proplist
24856 0.009 5.959 0.015 propset
12 0.009 0.010 0.009 resolve
12 0.334 2.710 0.958 resolved
84 0.179 0.958 0.470 status
6 8.425 15.409 12.164 switch
90 0.150 31.070 6.764 update
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 5x5_trunk
Running svn benchmark in /tmp/tmpovNKl3
dir levels: 5; new files and dirs per leaf: 5; run 1 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:07:41.871009
N min max avg operation (unit is seconds)
1 461.871 461.871 461.871 TOTAL RUN
139 0.011 1.724 0.032 add
2 0.025 6.626 3.325 checkout
8 5.186 23.987 13.493 commit
1 0.038 0.038 0.038 copy
1 120.950 120.950 120.950 delete
2 10.837 13.988 12.413 merge
164 0.011 0.573 0.021 mkdir
1091 0.011 0.077 0.017 prop mod
41 0.012 0.076 0.021 propdel
2076 0.010 0.096 0.016 proplist
4235 0.011 0.543 0.017 propset
2 0.010 0.010 0.010 resolve
2 0.922 1.773 1.348 resolved
14 0.252 2.528 0.927 status
1 6.508 6.508 6.508 switch
15 0.289 8.148 2.912 update
Running svn benchmark in /tmp/tmpCwCsGe
dir levels: 5; new files and dirs per leaf: 5; run 2 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:08:04.488720
N min max avg operation (unit is seconds)
2 461.871 484.489 473.180 TOTAL RUN
321 0.011 1.988 0.031 add
4 0.025 8.391 3.767 checkout
16 4.483 28.344 13.952 commit
2 0.038 0.048 0.043 copy
2 120.950 130.853 125.902 delete
4 10.544 13.988 11.731 merge
327 0.011 0.573 0.019 mkdir
2129 0.011 0.109 0.017 prop mod
85 0.011 0.076 0.019 propdel
4156 0.010 0.102 0.016 proplist
8375 0.011 0.543 0.017 propset
4 0.010 0.011 0.011 resolve
4 0.922 2.053 1.514 resolved
28 0.252 2.603 0.976 status
2 5.743 6.508 6.126 switch
30 0.245 9.084 3.006 update
Running svn benchmark in /tmp/tmpaXahGN
dir levels: 5; new files and dirs per leaf: 5; run 3 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:07:42.478464
N min max avg operation (unit is seconds)
3 461.871 484.489 469.613 TOTAL RUN
473 0.011 2.089 0.031 add
6 0.025 8.391 3.789 checkout
24 4.483 28.344 13.883 commit
3 0.038 0.062 0.050 copy
3 120.950 130.853 124.867 delete
6 9.075 13.988 11.258 merge
503 0.011 0.573 0.019 mkdir
3201 0.011 0.316 0.017 prop mod
142 0.011 0.076 0.018 propdel
6294 0.010 0.102 0.016 proplist
12550 0.011 0.543 0.017 propset
6 0.010 0.011 0.011 resolve
6 0.922 2.380 1.625 resolved
42 0.252 2.603 1.004 status
3 5.743 6.508 6.018 switch
45 0.245 9.390 2.998 update
Running svn benchmark in /tmp/tmpww_2_Q
dir levels: 5; new files and dirs per leaf: 5; run 4 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:07:23.774131
N min max avg operation (unit is seconds)
4 443.774 484.489 463.153 TOTAL RUN
653 0.011 2.411 0.031 add
8 0.025 8.391 3.882 checkout
32 3.908 28.344 13.837 commit
4 0.028 0.062 0.044 copy
4 102.673 130.853 119.319 delete
8 8.591 20.045 12.023 merge
683 0.011 0.647 0.019 mkdir
4273 0.011 0.316 0.017 prop mod
182 0.011 0.076 0.017 propdel
8401 0.010 0.102 0.015 proplist
16686 0.011 0.543 0.017 propset
8 0.010 0.011 0.011 resolve
8 0.922 2.380 1.550 resolved
56 0.252 2.932 0.996 status
4 5.743 6.508 6.038 switch
60 0.245 9.390 2.957 update
Running svn benchmark in /tmp/tmpKjP2n0
dir levels: 5; new files and dirs per leaf: 5; run 5 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:06:20.290688
N min max avg operation (unit is seconds)
5 380.291 484.489 446.581 TOTAL RUN
805 0.011 2.411 0.030 add
10 0.025 8.391 3.733 checkout
40 3.908 28.344 13.516 commit
5 0.027 0.062 0.041 copy
5 102.673 130.853 116.274 delete
10 8.591 20.045 11.545 merge
829 0.011 0.647 0.018 mkdir
5303 0.011 0.548 0.016 prop mod
215 0.011 0.076 0.016 propdel
10530 0.010 0.102 0.014 proplist
20743 0.011 0.543 0.016 propset
10 0.010 0.023 0.012 resolve
10 0.922 2.380 1.529 resolved
70 0.252 2.932 0.971 status
5 5.069 6.508 5.844 switch
75 0.245 9.390 2.867 update
Running svn benchmark in /tmp/tmp9OI77O
dir levels: 5; new files and dirs per leaf: 5; run 6 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:06:49.024624
N min max avg operation (unit is seconds)
6 380.291 484.489 440.321 TOTAL RUN
955 0.011 2.411 0.029 add
12 0.024 8.391 3.699 checkout
48 3.864 28.344 13.460 commit
6 0.027 0.116 0.053 copy
6 102.673 130.853 114.028 delete
12 8.591 20.045 11.233 merge
974 0.011 0.647 0.018 mkdir
6384 0.011 0.548 0.016 prop mod
252 0.011 0.076 0.016 propdel
12632 0.010 0.102 0.014 proplist
24856 0.011 0.543 0.016 propset
12 0.010 0.023 0.012 resolve
12 0.922 2.380 1.499 resolved
84 0.252 2.932 0.958 status
6 5.069 6.508 5.722 switch
90 0.244 9.390 2.847 update
Results for dir levels: 5 spread: 5
Timings for 5x5_1.6
N min max avg operation (unit is seconds)
6 330.544 561.406 425.169 TOTAL RUN
955 0.008 3.820 0.031 add
12 0.013 20.311 7.208 checkout
48 6.181 53.167 17.509 commit
6 0.030 0.894 0.436 copy
6 2.313 7.881 4.588 delete
12 6.898 21.264 14.617 merge
974 0.009 2.464 0.015 mkdir
6384 0.009 2.935 0.018 prop mod
252 0.009 0.496 0.033 propdel
12632 0.008 0.381 0.009 proplist
24856 0.009 5.959 0.015 propset
12 0.009 0.010 0.009 resolve
12 0.334 2.710 0.958 resolved
84 0.179 0.958 0.470 status
6 8.425 15.409 12.164 switch
90 0.150 31.070 6.764 update
---
Timings for 5x5_trunk
N min max avg operation (unit is seconds)
6 380.291 484.489 440.321 TOTAL RUN
955 0.011 2.411 0.029 add
12 0.024 8.391 3.699 checkout
48 3.864 28.344 13.460 commit
6 0.027 0.116 0.053 copy
6 102.673 130.853 114.028 delete
12 8.591 20.045 11.233 merge
974 0.011 0.647 0.018 mkdir
6384 0.011 0.548 0.016 prop mod
252 0.011 0.076 0.016 propdel
12632 0.010 0.102 0.014 proplist
24856 0.011 0.543 0.016 propset
12 0.010 0.023 0.012 resolve
12 0.922 2.380 1.499 resolved
84 0.252 2.932 0.958 status
6 5.069 6.508 5.722 switch
90 0.244 9.390 2.847 update
---
COMPARE 5x5_1.6 to 5x5_trunk
1.23|+0.45 means factor=1.23, difference in seconds = 0.45
factor < 1 or difference < 0 means '5x5_trunk' is faster than '5x5_1.6'
min max avg operation
1.151|+49.7471 0.863|-76.9170 1.036|+15.1526 TOTAL RUN
1.322|+0.0026 0.631|-1.4086 0.938|-0.0019 add
1.864|+0.0113 0.413|-11.9194 0.513|-3.5088 checkout
0.625|-2.3169 0.533|-24.8236 0.769|-4.0490 commit
0.899|-0.0030 0.129|-0.7784 0.122|-0.3825 copy
44.382|+100.3599 16.604|+122.9722 24.856|+109.4408 delete
1.245|+1.6928 0.943|-1.2191 0.768|-3.3842 merge
1.258|+0.0023 0.263|-1.8166 1.189|+0.0028 mkdir
1.256|+0.0022 0.187|-2.3871 0.910|-0.0016 prop mod
1.237|+0.0021 0.153|-0.4198 0.480|-0.0174 propdel
1.267|+0.0020 0.267|-0.2795 1.544|+0.0050 proplist
1.267|+0.0023 0.091|-5.4157 1.018|+0.0003 propset
1.148|+0.0013 2.347|+0.0133 1.254|+0.0024 resolve
2.764|+0.5885 0.878|-0.3298 1.565|+0.5413 resolved
1.410|+0.0733 3.061|+1.9744 2.038|+0.4877 status
0.602|-3.3563 0.422|-8.9009 0.470|-6.4426 switch
1.622|+0.0934 0.302|-21.6808 0.421|-3.9173 update
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 100x1_1.6
Running svn benchmark in /tmp/tmpLbeiSz
dir levels: 100; new files and dirs per leaf: 1; run 1 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:12.712391
N min max avg operation (unit is seconds)
1 72.712 72.712 72.712 TOTAL RUN
23 0.009 0.718 0.040 add
2 0.015 6.887 3.451 checkout
8 0.955 1.950 1.330 commit
1 0.027 0.027 0.027 copy
1 0.415 0.415 0.415 delete
2 0.716 1.397 1.056 merge
20 0.009 0.012 0.011 mkdir
64 0.009 0.011 0.010 prop mod
5 0.009 0.099 0.028 propdel
115 0.008 0.010 0.009 proplist
194 0.009 0.247 0.013 propset
2 0.009 0.009 0.009 resolve
2 0.083 0.140 0.112 resolved
14 0.065 0.166 0.118 status
1 4.460 4.460 4.460 switch
15 0.072 6.318 2.686 update
Running svn benchmark in /tmp/tmpT7RIDI
dir levels: 100; new files and dirs per leaf: 1; run 2 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:14.990382
N min max avg operation (unit is seconds)
2 72.712 74.990 73.851 TOTAL RUN
43 0.008 0.718 0.040 add
4 0.013 6.887 3.203 checkout
16 0.814 2.037 1.248 commit
2 0.027 0.160 0.094 copy
2 0.415 0.457 0.436 delete
4 0.716 1.397 0.985 merge
46 0.009 0.032 0.012 mkdir
131 0.009 0.019 0.010 prop mod
7 0.009 0.099 0.030 propdel
248 0.008 0.017 0.009 proplist
371 0.009 0.247 0.012 propset
4 0.008 0.009 0.009 resolve
4 0.083 0.158 0.120 resolved
28 0.065 0.188 0.119 status
2 4.460 5.042 4.751 switch
30 0.072 7.590 2.851 update
Running svn benchmark in /tmp/tmpH32c17
dir levels: 100; new files and dirs per leaf: 1; run 3 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:14.629023
N min max avg operation (unit is seconds)
3 72.712 74.990 74.111 TOTAL RUN
71 0.008 0.718 0.034 add
6 0.013 6.887 3.133 checkout
24 0.782 2.142 1.171 commit
3 0.024 0.160 0.070 copy
3 0.365 0.457 0.413 delete
6 0.716 1.397 1.015 merge
71 0.009 0.032 0.012 mkdir
204 0.009 0.152 0.011 prop mod
7 0.009 0.099 0.030 propdel
386 0.008 0.017 0.009 proplist
558 0.009 0.408 0.012 propset
6 0.008 0.010 0.009 resolve
6 0.083 0.158 0.123 resolved
42 0.065 0.188 0.119 status
3 4.460 5.301 4.934 switch
45 0.072 7.590 2.898 update
Running svn benchmark in /tmp/tmpbepUwG
dir levels: 100; new files and dirs per leaf: 1; run 4 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:13.213827
N min max avg operation (unit is seconds)
4 72.712 74.990 73.886 TOTAL RUN
98 0.008 0.718 0.034 add
8 0.013 6.887 3.110 checkout
32 0.705 2.158 1.170 commit
4 0.024 0.160 0.060 copy
4 0.288 0.457 0.382 delete
8 0.716 1.397 1.000 merge
90 0.009 0.032 0.011 mkdir
266 0.009 0.152 0.011 prop mod
10 0.009 0.099 0.024 propdel
509 0.008 0.017 0.009 proplist
743 0.009 0.408 0.012 propset
8 0.008 0.010 0.009 resolve
8 0.083 0.158 0.124 resolved
56 0.065 0.261 0.120 status
4 4.460 5.301 4.916 switch
60 0.072 7.590 2.900 update
Running svn benchmark in /tmp/tmp3jGNt1
dir levels: 100; new files and dirs per leaf: 1; run 5 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:16.413093
N min max avg operation (unit is seconds)
5 72.712 76.413 74.392 TOTAL RUN
119 0.008 0.718 0.034 add
10 0.013 6.887 3.132 checkout
40 0.705 2.158 1.165 commit
5 0.024 0.160 0.074 copy
5 0.288 0.457 0.387 delete
10 0.716 1.397 0.996 merge
108 0.009 0.032 0.011 mkdir
329 0.009 0.152 0.011 prop mod
13 0.009 0.099 0.020 propdel
642 0.008 0.017 0.009 proplist
917 0.009 0.408 0.012 propset
10 0.008 0.010 0.009 resolve
10 0.083 0.159 0.124 resolved
70 0.065 0.261 0.121 status
5 4.460 5.462 5.025 switch
75 0.072 7.590 2.931 update
Running svn benchmark in /tmp/tmpdbKHJ6
dir levels: 100; new files and dirs per leaf: 1; run 6 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:01:16.609033
N min max avg operation (unit is seconds)
6 72.712 76.609 74.761 TOTAL RUN
141 0.008 0.718 0.034 add
12 0.013 6.887 3.112 checkout
48 0.660 2.239 1.155 commit
6 0.023 0.160 0.066 copy
6 0.288 0.457 0.376 delete
12 0.716 1.397 0.989 merge
131 0.009 0.032 0.011 mkdir
379 0.009 0.152 0.011 prop mod
14 0.009 0.099 0.020 propdel
759 0.008 0.017 0.009 proplist
1099 0.009 0.408 0.012 propset
12 0.008 0.010 0.009 resolve
12 0.083 0.173 0.126 resolved
84 0.065 0.261 0.124 status
6 4.460 5.462 5.095 switch
90 0.072 7.590 2.962 update
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 100x1_trunk
Running svn benchmark in /tmp/tmp5CMQJO
dir levels: 100; new files and dirs per leaf: 1; run 1 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:26.835246
N min max avg operation (unit is seconds)
1 26.835 26.835 26.835 TOTAL RUN
23 0.011 0.149 0.020 add
2 0.025 0.517 0.271 checkout
8 0.326 1.637 0.901 commit
1 0.041 0.041 0.041 copy
1 2.496 2.496 2.496 delete
2 0.782 0.864 0.823 merge
20 0.012 0.019 0.015 mkdir
64 0.011 0.018 0.014 prop mod
5 0.011 0.016 0.013 propdel
115 0.010 0.016 0.013 proplist
194 0.012 0.019 0.015 propset
2 0.010 0.011 0.011 resolve
2 0.103 0.172 0.138 resolved
14 0.062 0.604 0.174 status
1 0.511 0.511 0.511 switch
15 0.067 0.729 0.360 update
Running svn benchmark in /tmp/tmpwlT5VK
dir levels: 100; new files and dirs per leaf: 1; run 2 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:27.925119
N min max avg operation (unit is seconds)
2 26.835 27.925 27.380 TOTAL RUN
43 0.011 0.169 0.021 add
4 0.025 0.548 0.279 checkout
16 0.326 2.141 0.899 commit
2 0.021 0.041 0.031 copy
2 2.496 2.683 2.589 delete
4 0.741 0.945 0.833 merge
46 0.012 0.019 0.015 mkdir
131 0.011 0.018 0.015 prop mod
7 0.011 0.018 0.014 propdel
248 0.010 0.022 0.013 proplist
371 0.011 0.019 0.015 propset
4 0.010 0.011 0.010 resolve
4 0.103 0.183 0.142 resolved
28 0.062 0.638 0.180 status
2 0.511 0.551 0.531 switch
30 0.067 0.772 0.372 update
Running svn benchmark in /tmp/tmpXqDK9d
dir levels: 100; new files and dirs per leaf: 1; run 3 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:27.232233
N min max avg operation (unit is seconds)
3 26.835 27.925 27.331 TOTAL RUN
71 0.011 0.169 0.020 add
6 0.024 0.548 0.278 checkout
24 0.326 2.141 0.872 commit
3 0.021 0.041 0.028 copy
3 2.496 2.754 2.644 delete
6 0.741 0.945 0.835 merge
71 0.012 0.020 0.015 mkdir
204 0.011 0.018 0.014 prop mod
7 0.011 0.018 0.014 propdel
386 0.010 0.022 0.013 proplist
558 0.011 0.019 0.015 propset
6 0.010 0.011 0.010 resolve
6 0.103 0.190 0.145 resolved
42 0.062 0.638 0.180 status
3 0.511 0.551 0.535 switch
45 0.067 0.772 0.375 update
Running svn benchmark in /tmp/tmp7qnuJz
dir levels: 100; new files and dirs per leaf: 1; run 4 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:27.175945
N min max avg operation (unit is seconds)
4 26.835 27.925 27.292 TOTAL RUN
98 0.011 0.169 0.020 add
8 0.024 0.548 0.278 checkout
32 0.326 2.141 0.863 commit
4 0.021 0.041 0.027 copy
4 2.496 2.839 2.693 delete
8 0.741 0.952 0.848 merge
90 0.012 0.020 0.015 mkdir
266 0.011 0.019 0.014 prop mod
10 0.011 0.018 0.013 propdel
509 0.010 0.022 0.013 proplist
743 0.011 0.019 0.015 propset
8 0.010 0.011 0.010 resolve
8 0.103 0.190 0.146 resolved
56 0.062 0.638 0.180 status
4 0.511 0.551 0.534 switch
60 0.067 0.811 0.377 update
Running svn benchmark in /tmp/tmpA1fnxG
dir levels: 100; new files and dirs per leaf: 1; run 5 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:27.974462
N min max avg operation (unit is seconds)
5 26.835 27.974 27.429 TOTAL RUN
119 0.011 0.169 0.021 add
10 0.024 0.558 0.281 checkout
40 0.326 2.141 0.873 commit
5 0.021 0.041 0.026 copy
5 2.496 2.839 2.686 delete
10 0.741 0.967 0.856 merge
108 0.012 0.020 0.015 mkdir
329 0.011 0.019 0.014 prop mod
13 0.011 0.018 0.013 propdel
642 0.010 0.022 0.013 proplist
917 0.011 0.020 0.015 propset
10 0.010 0.011 0.010 resolve
10 0.103 0.190 0.146 resolved
70 0.062 0.653 0.179 status
5 0.511 0.570 0.541 switch
75 0.067 0.811 0.381 update
Running svn benchmark in /tmp/tmpJ4XWJb
dir levels: 100; new files and dirs per leaf: 1; run 6 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:30.905283
N min max avg operation (unit is seconds)
6 26.835 30.905 28.008 TOTAL RUN
141 0.011 0.169 0.021 add
12 0.024 0.830 0.305 checkout
48 0.326 2.141 0.888 commit
6 0.021 0.059 0.032 copy
6 2.496 2.839 2.710 delete
12 0.741 1.026 0.870 merge
131 0.012 0.055 0.016 mkdir
379 0.011 0.058 0.015 prop mod
14 0.011 0.018 0.013 propdel
759 0.010 0.061 0.013 proplist
1099 0.011 0.074 0.015 propset
12 0.010 0.011 0.010 resolve
12 0.103 0.192 0.147 resolved
84 0.062 0.658 0.181 status
6 0.511 0.599 0.551 switch
90 0.067 0.891 0.391 update
Results for dir levels: 100 spread: 1
Timings for 100x1_1.6
N min max avg operation (unit is seconds)
6 72.712 76.609 74.761 TOTAL RUN
141 0.008 0.718 0.034 add
12 0.013 6.887 3.112 checkout
48 0.660 2.239 1.155 commit
6 0.023 0.160 0.066 copy
6 0.288 0.457 0.376 delete
12 0.716 1.397 0.989 merge
131 0.009 0.032 0.011 mkdir
379 0.009 0.152 0.011 prop mod
14 0.009 0.099 0.020 propdel
759 0.008 0.017 0.009 proplist
1099 0.009 0.408 0.012 propset
12 0.008 0.010 0.009 resolve
12 0.083 0.173 0.126 resolved
84 0.065 0.261 0.124 status
6 4.460 5.462 5.095 switch
90 0.072 7.590 2.962 update
---
Timings for 100x1_trunk
N min max avg operation (unit is seconds)
6 26.835 30.905 28.008 TOTAL RUN
141 0.011 0.169 0.021 add
12 0.024 0.830 0.305 checkout
48 0.326 2.141 0.888 commit
6 0.021 0.059 0.032 copy
6 2.496 2.839 2.710 delete
12 0.741 1.026 0.870 merge
131 0.012 0.055 0.016 mkdir
379 0.011 0.058 0.015 prop mod
14 0.011 0.018 0.013 propdel
759 0.010 0.061 0.013 proplist
1099 0.011 0.074 0.015 propset
12 0.010 0.011 0.010 resolve
12 0.103 0.192 0.147 resolved
84 0.062 0.658 0.181 status
6 0.511 0.599 0.551 switch
90 0.067 0.891 0.391 update
---
COMPARE 100x1_1.6 to 100x1_trunk
1.23|+0.45 means factor=1.23, difference in seconds = 0.45
factor < 1 or difference < 0 means '100x1_trunk' is faster than '100x1_1.6'
min max avg operation
0.369|-45.8771 0.403|-45.7037 0.375|-46.7532 TOTAL RUN
1.343|+0.0029 0.235|-0.5490 0.625|-0.0128 add
1.797|+0.0106 0.121|-6.0566 0.098|-2.8065 checkout
0.494|-0.3340 0.956|-0.0983 0.768|-0.2675 commit
0.920|-0.0018 0.369|-0.1012 0.483|-0.0339 copy
8.650|+2.2071 6.214|+2.3825 7.211|+2.3340 delete
1.035|+0.0251 0.734|-0.3708 0.879|-0.1196 merge
1.263|+0.0024 1.715|+0.0230 1.438|+0.0050 mkdir
1.251|+0.0022 0.380|-0.0941 1.437|+0.0046 prop mod
1.208|+0.0019 0.179|-0.0809 0.657|-0.0067 propdel
1.263|+0.0021 3.673|+0.0447 1.523|+0.0045 proplist
1.251|+0.0022 0.182|-0.3340 1.276|+0.0033 propset
1.177|+0.0015 1.040|+0.0004 1.148|+0.0013 resolve
1.243|+0.0202 1.110|+0.0190 1.169|+0.0213 resolved
0.956|-0.0029 2.521|+0.3971 1.467|+0.0577 status
0.115|-3.9488 0.110|-4.8625 0.108|-4.5442 switch
0.923|-0.0056 0.117|-6.6983 0.132|-2.5707 update
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 1x100_1.6
Running svn benchmark in /tmp/tmp40sQWe
dir levels: 1; new files and dirs per leaf: 100; run 1 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:05.536515
N min max avg operation (unit is seconds)
1 5.537 5.537 5.537 TOTAL RUN
1 0.184 0.184 0.184 add
2 0.013 0.102 0.058 checkout
8 0.087 0.242 0.173 commit
1 0.022 0.022 0.022 copy
1 0.098 0.098 0.098 delete
2 0.182 0.303 0.242 merge
31 0.010 0.146 0.015 prop mod
4 0.011 0.011 0.011 propdel
48 0.009 0.010 0.009 proplist
98 0.010 0.017 0.011 propset
2 0.009 0.010 0.010 resolve
2 0.011 0.017 0.014 resolved
14 0.011 0.020 0.016 status
1 0.099 0.099 0.099 switch
15 0.013 0.183 0.053 update
Running svn benchmark in /tmp/tmp8HfWGt
dir levels: 1; new files and dirs per leaf: 100; run 2 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:04.559893
N min max avg operation (unit is seconds)
2 4.560 5.537 5.048 TOTAL RUN
2 0.051 0.184 0.117 add
4 0.013 0.102 0.056 checkout
16 0.080 0.242 0.163 commit
2 0.021 0.022 0.021 copy
2 0.089 0.098 0.093 delete
4 0.182 0.303 0.228 merge
51 0.010 0.146 0.013 prop mod
4 0.011 0.011 0.011 propdel
85 0.009 0.010 0.009 proplist
176 0.010 0.017 0.011 propset
4 0.009 0.011 0.010 resolve
4 0.011 0.022 0.016 resolved
28 0.011 0.020 0.016 status
2 0.099 0.103 0.101 switch
30 0.013 0.189 0.053 update
Running svn benchmark in /tmp/tmp72L_AW
dir levels: 1; new files and dirs per leaf: 100; run 3 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:05.026687
N min max avg operation (unit is seconds)
3 4.560 5.537 5.041 TOTAL RUN
3 0.051 0.245 0.160 add
6 0.013 0.102 0.056 checkout
24 0.080 0.242 0.162 commit
3 0.019 0.022 0.021 copy
3 0.089 0.098 0.094 delete
6 0.182 0.303 0.226 merge
62 0.010 0.146 0.013 prop mod
4 0.011 0.011 0.011 propdel
115 0.009 0.048 0.010 proplist
272 0.010 0.017 0.011 propset
6 0.009 0.011 0.010 resolve
6 0.011 0.022 0.015 resolved
42 0.011 0.038 0.016 status
3 0.099 0.119 0.107 switch
45 0.013 0.189 0.054 update
Running svn benchmark in /tmp/tmpB4PDsE
dir levels: 1; new files and dirs per leaf: 100; run 4 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:05.190180
N min max avg operation (unit is seconds)
4 4.560 5.537 5.078 TOTAL RUN
4 0.047 0.245 0.132 add
8 0.013 0.124 0.059 checkout
32 0.080 0.258 0.164 commit
4 0.019 0.024 0.021 copy
4 0.089 0.098 0.093 delete
8 0.182 0.303 0.227 merge
79 0.010 0.146 0.012 prop mod
4 0.011 0.011 0.011 propdel
143 0.008 0.048 0.010 proplist
369 0.010 0.055 0.011 propset
8 0.009 0.011 0.010 resolve
8 0.011 0.028 0.018 resolved
56 0.011 0.048 0.017 status
4 0.093 0.119 0.104 switch
60 0.013 0.210 0.056 update
Running svn benchmark in /tmp/tmpD2Q5kk
dir levels: 1; new files and dirs per leaf: 100; run 5 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:04.757104
N min max avg operation (unit is seconds)
5 4.560 5.537 5.014 TOTAL RUN
5 0.047 0.315 0.168 add
10 0.013 0.124 0.058 checkout
40 0.080 0.258 0.162 commit
5 0.019 0.024 0.021 copy
5 0.082 0.098 0.091 delete
10 0.182 0.303 0.225 merge
97 0.010 0.146 0.012 prop mod
4 0.011 0.011 0.011 propdel
172 0.008 0.048 0.010 proplist
463 0.010 0.055 0.011 propset
10 0.008 0.011 0.009 resolve
10 0.010 0.028 0.017 resolved
70 0.010 0.048 0.016 status
5 0.093 0.119 0.103 switch
75 0.012 0.210 0.055 update
Running svn benchmark in /tmp/tmpDaZ251
dir levels: 1; new files and dirs per leaf: 100; run 6 of 6
svn, version 1.6.17 (dev build), compiled Apr 5 2011, 11:14:17
Done with svn benchmark in 0:00:04.727893
N min max avg operation (unit is seconds)
6 4.560 5.537 4.966 TOTAL RUN
6 0.046 0.315 0.148 add
12 0.013 0.124 0.057 checkout
48 0.080 0.258 0.161 commit
6 0.019 0.024 0.021 copy
6 0.082 0.098 0.090 delete
12 0.177 0.315 0.228 merge
122 0.010 0.146 0.011 prop mod
6 0.010 0.011 0.011 propdel
212 0.008 0.048 0.010 proplist
543 0.010 0.055 0.011 propset
12 0.008 0.032 0.011 resolve
12 0.010 0.028 0.017 resolved
84 0.010 0.048 0.016 status
6 0.093 0.119 0.102 switch
90 0.012 0.210 0.056 update
Hi, going to run a Subversion benchmark series of 6 runs...
Going to write results to new file 1x100_trunk
Running svn benchmark in /tmp/tmpCKc2Cj
dir levels: 1; new files and dirs per leaf: 100; run 1 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:06.887158
N min max avg operation (unit is seconds)
1 6.887 6.887 6.887 TOTAL RUN
1 0.042 0.042 0.042 add
2 0.027 0.159 0.093 checkout
8 0.101 0.394 0.262 commit
1 0.023 0.023 0.023 copy
1 0.092 0.092 0.092 delete
2 0.193 0.228 0.210 merge
31 0.011 0.045 0.014 prop mod
4 0.011 0.037 0.020 propdel
48 0.010 0.049 0.014 proplist
98 0.011 0.021 0.012 propset
2 0.011 0.011 0.011 resolve
2 0.035 0.076 0.056 resolved
14 0.013 0.066 0.025 status
1 0.094 0.094 0.094 switch
15 0.026 0.148 0.066 update
Running svn benchmark in /tmp/tmp05MP1c
dir levels: 1; new files and dirs per leaf: 100; run 2 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:05.343976
N min max avg operation (unit is seconds)
2 5.344 6.887 6.116 TOTAL RUN
2 0.041 0.042 0.041 add
4 0.026 0.159 0.089 checkout
16 0.092 0.416 0.244 commit
2 0.022 0.023 0.022 copy
2 0.092 0.093 0.093 delete
4 0.166 0.228 0.194 merge
51 0.011 0.045 0.013 prop mod
4 0.011 0.037 0.020 propdel
85 0.010 0.049 0.013 proplist
176 0.011 0.021 0.012 propset
4 0.010 0.011 0.010 resolve
4 0.026 0.076 0.043 resolved
28 0.013 0.066 0.023 status
2 0.094 0.099 0.096 switch
30 0.020 0.148 0.060 update
Running svn benchmark in /tmp/tmpTID1L2
dir levels: 1; new files and dirs per leaf: 100; run 3 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:05.405396
N min max avg operation (unit is seconds)
3 5.344 6.887 5.879 TOTAL RUN
3 0.038 0.042 0.040 add
6 0.025 0.159 0.088 checkout
24 0.090 0.416 0.237 commit
3 0.021 0.023 0.022 copy
3 0.090 0.093 0.092 delete
6 0.156 0.228 0.187 merge
62 0.011 0.045 0.013 prop mod
4 0.011 0.037 0.020 propdel
115 0.010 0.049 0.012 proplist
272 0.011 0.021 0.012 propset
6 0.010 0.011 0.010 resolve
6 0.025 0.076 0.039 resolved
42 0.013 0.066 0.023 status
3 0.094 0.099 0.096 switch
45 0.019 0.148 0.058 update
Running svn benchmark in /tmp/tmpLFT1zR
dir levels: 1; new files and dirs per leaf: 100; run 4 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:06.050390
N min max avg operation (unit is seconds)
4 5.344 6.887 5.922 TOTAL RUN
4 0.036 0.042 0.039 add
8 0.025 0.159 0.087 checkout
32 0.090 0.416 0.243 commit
4 0.021 0.023 0.022 copy
4 0.090 0.105 0.095 delete
8 0.156 0.237 0.197 merge
79 0.011 0.045 0.013 prop mod
4 0.011 0.037 0.020 propdel
143 0.010 0.049 0.012 proplist
369 0.011 0.021 0.012 propset
8 0.010 0.046 0.015 resolve
8 0.025 0.076 0.041 resolved
56 0.012 0.066 0.022 status
4 0.094 0.099 0.096 switch
60 0.019 0.148 0.058 update
Running svn benchmark in /tmp/tmpJPYd7z
dir levels: 1; new files and dirs per leaf: 100; run 5 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:05.713691
N min max avg operation (unit is seconds)
5 5.344 6.887 5.880 TOTAL RUN
5 0.036 0.042 0.039 add
10 0.025 0.159 0.086 checkout
40 0.090 0.416 0.241 commit
5 0.021 0.023 0.022 copy
5 0.088 0.105 0.094 delete
10 0.156 0.237 0.197 merge
97 0.011 0.045 0.013 prop mod
4 0.011 0.037 0.020 propdel
172 0.010 0.049 0.012 proplist
463 0.011 0.033 0.012 propset
10 0.010 0.046 0.014 resolve
10 0.025 0.078 0.043 resolved
70 0.012 0.066 0.022 status
5 0.094 0.099 0.096 switch
75 0.018 0.148 0.058 update
Running svn benchmark in /tmp/tmprSVGYt
dir levels: 1; new files and dirs per leaf: 100; run 6 of 6
svn, version 1.7.0-dev (under development), compiled Apr 7 2011, 00:21:31
Done with svn benchmark in 0:00:05.678437
N min max avg operation (unit is seconds)
6 5.344 6.887 5.847 TOTAL RUN
6 0.036 0.053 0.041 add
12 0.024 0.159 0.087 checkout
48 0.090 0.416 0.239 commit
6 0.021 0.027 0.022 copy
6 0.088 0.105 0.093 delete
12 0.156 0.237 0.194 merge
122 0.011 0.045 0.013 prop mod
6 0.011 0.037 0.017 propdel
212 0.010 0.049 0.012 proplist
543 0.011 0.064 0.012 propset
12 0.010 0.046 0.013 resolve
12 0.025 0.078 0.041 resolved
84 0.012 0.066 0.022 status
6 0.092 0.099 0.095 switch
90 0.018 0.153 0.057 update
Results for dir levels: 1 spread: 100
Timings for 1x100_1.6
N min max avg operation (unit is seconds)
6 4.560 5.537 4.966 TOTAL RUN
6 0.046 0.315 0.148 add
12 0.013 0.124 0.057 checkout
48 0.080 0.258 0.161 commit
6 0.019 0.024 0.021 copy
6 0.082 0.098 0.090 delete
12 0.177 0.315 0.228 merge
122 0.010 0.146 0.011 prop mod
6 0.010 0.011 0.011 propdel
212 0.008 0.048 0.010 proplist
543 0.010 0.055 0.011 propset
12 0.008 0.032 0.011 resolve
12 0.010 0.028 0.017 resolved
84 0.010 0.048 0.016 status
6 0.093 0.119 0.102 switch
90 0.012 0.210 0.056 update
---
Timings for 1x100_trunk
N min max avg operation (unit is seconds)
6 5.344 6.887 5.847 TOTAL RUN
6 0.036 0.053 0.041 add
12 0.024 0.159 0.087 checkout
48 0.090 0.416 0.239 commit
6 0.021 0.027 0.022 copy
6 0.088 0.105 0.093 delete
12 0.156 0.237 0.194 merge
122 0.011 0.045 0.013 prop mod
6 0.011 0.037 0.017 propdel
212 0.010 0.049 0.012 proplist
543 0.011 0.064 0.012 propset
12 0.010 0.046 0.013 resolve
12 0.025 0.078 0.041 resolved
84 0.012 0.066 0.022 status
6 0.092 0.099 0.095 switch
90 0.018 0.153 0.057 update
---
COMPARE 1x100_1.6 to 1x100_trunk
1.23|+0.45 means factor=1.23, difference in seconds = 0.45
factor < 1 or difference < 0 means '1x100_trunk' is faster than '1x100_1.6'
min max avg operation
1.172|+0.7841 1.244|+1.3506 1.177|+0.8801 TOTAL RUN
0.773|-0.0105 0.169|-0.2621 0.278|-0.1070 add
1.822|+0.0110 1.284|+0.0350 1.529|+0.0299 checkout
1.123|+0.0098 1.612|+0.1578 1.483|+0.0778 commit
1.092|+0.0017 1.133|+0.0032 1.071|+0.0015 copy
1.068|+0.0056 1.070|+0.0068 1.033|+0.0029 delete
0.884|-0.0205 0.751|-0.0783 0.849|-0.0346 merge
1.133|+0.0013 0.307|-0.1012 1.111|+0.0013 prop mod
1.068|+0.0007 3.388|+0.0261 1.603|+0.0064 propdel
1.179|+0.0015 1.021|+0.0010 1.200|+0.0019 proplist
1.129|+0.0012 1.171|+0.0094 1.122|+0.0013 propset
1.156|+0.0013 1.434|+0.0139 1.185|+0.0021 resolve
2.360|+0.0143 2.801|+0.0499 2.466|+0.0243 resolved
1.216|+0.0022 1.363|+0.0174 1.363|+0.0058 status
0.981|-0.0018 0.826|-0.0208 0.934|-0.0067 switch
1.477|+0.0060 0.732|-0.0563 1.024|+0.0014 update
=========================================================================
calculating total of 1.6...
N min max avg operation (unit is seconds)
18 4.560 561.406 168.299 TOTAL RUN
1102 0.008 3.820 0.032 add
36 0.013 20.311 3.459 checkout
144 0.080 53.167 6.275 commit
18 0.019 0.894 0.174 copy
18 0.082 7.881 1.684 delete
36 0.177 21.264 5.278 merge
1105 0.009 2.464 0.014 mkdir
6885 0.009 2.935 0.017 prop mod
272 0.009 0.496 0.032 propdel
13603 0.008 0.381 0.009 proplist
26498 0.009 5.959 0.015 propset
36 0.008 0.032 0.010 resolve
36 0.010 2.710 0.367 resolved
252 0.010 0.958 0.203 status
18 0.093 15.409 5.787 switch
270 0.012 31.070 3.261 update
calculating total of trunk...
N min max avg operation (unit is seconds)
18 5.344 484.489 158.059 TOTAL RUN
1102 0.011 2.411 0.028 add
36 0.024 8.391 1.364 checkout
144 0.090 28.344 4.862 commit
18 0.021 0.116 0.036 copy
18 0.088 130.853 38.944 delete
36 0.156 20.045 4.099 merge
1105 0.011 0.647 0.018 mkdir
6885 0.011 0.548 0.016 prop mod
272 0.011 0.076 0.016 propdel
13603 0.010 0.102 0.014 proplist
26498 0.011 0.543 0.015 propset
36 0.010 0.046 0.012 resolve
36 0.025 2.380 0.563 resolved
252 0.012 2.932 0.387 status
18 0.092 6.508 2.123 switch
270 0.018 9.390 1.098 update
comparing averaged totals...
Timings for total_1.6
N min max avg operation (unit is seconds)
18 4.560 561.406 168.299 TOTAL RUN
1102 0.008 3.820 0.032 add
36 0.013 20.311 3.459 checkout
144 0.080 53.167 6.275 commit
18 0.019 0.894 0.174 copy
18 0.082 7.881 1.684 delete
36 0.177 21.264 5.278 merge
1105 0.009 2.464 0.014 mkdir
6885 0.009 2.935 0.017 prop mod
272 0.009 0.496 0.032 propdel
13603 0.008 0.381 0.009 proplist
26498 0.009 5.959 0.015 propset
36 0.008 0.032 0.010 resolve
36 0.010 2.710 0.367 resolved
252 0.010 0.958 0.203 status
18 0.093 15.409 5.787 switch
270 0.012 31.070 3.261 update
---
Timings for total_trunk
N min max avg operation (unit is seconds)
18 5.344 484.489 158.059 TOTAL RUN
1102 0.011 2.411 0.028 add
36 0.024 8.391 1.364 checkout
144 0.090 28.344 4.862 commit
18 0.021 0.116 0.036 copy
18 0.088 130.853 38.944 delete
36 0.156 20.045 4.099 merge
1105 0.011 0.647 0.018 mkdir
6885 0.011 0.548 0.016 prop mod
272 0.011 0.076 0.016 propdel
13603 0.010 0.102 0.014 proplist
26498 0.011 0.543 0.015 propset
36 0.010 0.046 0.012 resolve
36 0.025 2.380 0.563 resolved
252 0.012 2.932 0.387 status
18 0.092 6.508 2.123 switch
270 0.018 9.390 1.098 update
---
COMPARE total_1.6 to total_trunk
1.23|+0.45 means factor=1.23, difference in seconds = 0.45
factor < 1 or difference < 0 means 'total_trunk' is faster than 'total_1.6'
min max avg operation
1.172|+0.7841 0.863|-76.9170 0.939|-10.2402 TOTAL RUN
1.322|+0.0026 0.631|-1.4086 0.879|-0.0039 add
1.827|+0.0108 0.413|-11.9194 0.394|-2.0951 checkout
1.123|+0.0098 0.533|-24.8236 0.775|-1.4129 commit
1.092|+0.0017 0.129|-0.7784 0.206|-0.1383 copy
1.068|+0.0056 16.604|+122.9722 23.120|+37.2592 delete
0.884|-0.0205 0.943|-1.2191 0.777|-1.1794 merge
1.258|+0.0023 0.263|-1.8166 1.212|+0.0031 mkdir
1.256|+0.0022 0.187|-2.3871 0.930|-0.0012 prop mod
1.237|+0.0021 0.153|-0.4198 0.494|-0.0163 propdel
1.267|+0.0020 0.267|-0.2795 1.537|+0.0050 proplist
1.267|+0.0023 0.091|-5.4157 1.028|+0.0004 propset
1.160|+0.0014 1.434|+0.0139 1.196|+0.0019 resolve
2.360|+0.0143 0.878|-0.3298 1.533|+0.1956 resolved
1.216|+0.0022 3.061|+1.9744 1.904|+0.1837 status
0.981|-0.0018 0.422|-8.9009 0.367|-3.6645 switch
1.477|+0.0060 0.302|-21.6808 0.337|-2.1622 update
Had started at Thu Apr 7 00:24:51 UTC 2011,
done at Thu Apr 7 02:03:08 UTC 2011
/home/neels/svnbench/20110407-002451
real 5896.90
user 3366.21
sys 1374.08
Re: some more performance benchmarks
Posted by Daniel Shahaf <da...@elego.de>.
Neels Hofmeyr wrote on Thu, Apr 07, 2011 at 16:42:43 +0200:
> [[[
> COMPARE 5x5_1.6 to 5x5_trunk
> 1.23|+0.45 means factor=1.23, difference in seconds = 0.45
> factor < 1 or difference < 0 means '5x5_trunk' is faster than
> '5x5_1.6'
> min max avg operation
> 1.151|+49.7471 0.863|-76.9170 1.036|+15.1526 TOTAL RUN
> 1.322|+0.0026 0.631|-1.4086 0.938|-0.0019 add
> 1.864|+0.0113 0.413|-11.9194 0.513|-3.5088 checkout
> 0.625|-2.3169 0.533|-24.8236 0.769|-4.0490 commit
> 0.899|-0.0030 0.129|-0.7784 0.122|-0.3825 copy
> 44.382|+100.3599 16.604|+122.9722 24.856|+109.4408 delete
> 1.245|+1.6928 0.943|-1.2191 0.768|-3.3842 merge
> 1.258|+0.0023 0.263|-1.8166 1.189|+0.0028 mkdir
> 1.256|+0.0022 0.187|-2.3871 0.910|-0.0016 prop mod
> 1.237|+0.0021 0.153|-0.4198 0.480|-0.0174 propdel
> 1.267|+0.0020 0.267|-0.2795 1.544|+0.0050 proplist
> 1.267|+0.0023 0.091|-5.4157 1.018|+0.0003 propset
> 1.148|+0.0013 2.347|+0.0133 1.254|+0.0024 resolve
> 2.764|+0.5885 0.878|-0.3298 1.565|+0.5413 resolved
> 1.410|+0.0733 3.061|+1.9744 2.038|+0.4877 status
> 0.602|-3.3563 0.422|-8.9009 0.470|-6.4426 switch
> 1.622|+0.0934 0.302|-21.6808 0.421|-3.9173 update
> ]]]
...
> I'm playing with the thought of making graphical charts of these
> numbers, but that might be overkill :)
Disagree. If you're going to post long columns of numbers and expect
people to make any sense of them, a chart would help *a lot*.
Or, at least, make the output be in unary rather than decimal :-)
Daniel
Re: some more performance benchmarks
Posted by Greg Stein <gs...@gmail.com>.
On Thu, Apr 7, 2011 at 10:42, Neels Hofmeyr <ne...@elego.de> wrote:
>...
> Any ideas/opinions about adding to tools/dev/ would be appreciated.
Commited in r1092569 as tools/dev/benchmarks/suite1. Having them in
source control is better than worrying too much about the naming.
We can now evolve these for other platforms, people can run them on
their OS, etc.
Cheers,
-g
Re: some more performance benchmarks
Posted by Daniel Shahaf <da...@elego.de>.
Neels Hofmeyr wrote on Thu, Apr 07, 2011 at 16:42:43 +0200:
> I have no idea about other loads on our vm. Maybe we could use the
> user/sys output of 'time -p' instead of the real-world time measurements
> it is taking right now, although my intention was to stick to the actual
> real world time. Of course that works best on an otherwise idle system.
> danielsh, or anyone, do you have any idea at what times-of-day (or
> something) the VM host is most likely to idle for ~2 hours, so we can
> schedule the tests then?
It's two-fold:
* When is the VM itself inactive
* When is the host inactive
The first question is easy: just tell artagnon and danielsh "don't use
svn-qavm at the moment". Just edit /etc/motd, or edit /topic on IRC,
or something.
The second question is harder. The host runs over 30 other VMs, and
I am not sure offhand what they are all used for. If you care about
this level of detail, you'd have to talk to Gavin or Tony from infra ---
try catching them on infrastructure@ or #asfinfra.
Re: some more performance benchmarks
Posted by Neels Hofmeyr <ne...@elego.de>.
On Thu, 2011-04-07 at 08:03 -0500, Hyrum K Wright wrote:
> This is great work, though as you admit the results are bit
> incomprehensible at this stage. Would it make sense to somehow
Hmm, I think stsp was a bit early to post, I'm still in the sanity
checking phase ;)
E.g.,:
[[[
Timings for 5x5_1.6
N min max avg operation (unit is seconds)
6 330.544 561.406 425.169 TOTAL RUN
]]]
--> variation of 330s to 561s in 6 identical runs?
(later note: I just found the random seeding might be in the wrong
place. I recently refactored a few things and it should prbly go inside
the N loop now.)
I have no idea about other loads on our vm. Maybe we could use the
user/sys output of 'time -p' instead of the real-world time measurements
it is taking right now, although my intention was to stick to the actual
real world time. Of course that works best on an otherwise idle system.
danielsh, or anyone, do you have any idea at what times-of-day (or
something) the VM host is most likely to idle for ~2 hours, so we can
schedule the tests then?
> combine the scripts to prevent the bifurcation of benchmark suites?
For the curious, I attached the current state of my benchmark scripts,
but I'm still editing.
> Do you have the script of commands, and the accompanying repo?
ATM my scripts are in an elego-private repo, they were initially going
to be just a quick test to have real timing results in an article about
1.7.
Actually, I think they have matured enough to be committed in tools/dev.
If Mark does the same we could have a collective look at them and
decide. I haven't actually seen Mark's results nor scripts yet. I have
posted mine before, older versions, a few months ago...
So where to put it?
tools/dev/benchmark/A <--- mark's benches
tools/dev/benchmark/B <--- my benches
???
I'm reluctant to label the dir with my name...
It'd be nice to find two descriptive words to distinguish between the
benchmarks, "by type".
In reply to CMike, an outline of what my test does:
I have a little python magic I use to call svn subcommands, calling the
cmdline svn client from py. Each call is being timed and added to a set
of N calls of the same subcommand.
So it has no understanding of whether it got 'svn delete one/file.c' or
'svn delete huge/tree/*', it simply adds another sample. So that's
almost madness.
But it goes on and runs the same identical sequence a number of times,
currently 6. Their values are averaged, and the min/max are retained to
have some sense of the real-world meaning of the average.
These 6 identical runs are done once for 1.6.x and once for trunk. They
are identical because I use a pseudo random number generator (to get
"organic" modifications) that is seeded with 0 for each run :P
Finally, the averages/mins/maxes for 1.6.x are compared to those of
trunk, showing a factor and a difference in seconds. (A factor of 10
were negligible if the actual time added was only 2 microseconds...)
Because of this "madness" I mentioned above (variation in single calls),
the *averages* are what's really interesting to look at. Even though an
average for 1.6 may be made up of (60 + 0.06)/2, trunk's average could
be (90 + 0.01)/2, and still shows the worst case pretty visibly (a long
call taking even longer). You can stare at the min/max differences to
make up your mind as to whether the averages are actual timings in
seconds or unitless fantasy ratings. Yet I am pretty confident that they
sufficiently compare the two clients.
In stsp's post, every single run was printed out. The really interesting
places to look though are where it says "Results for"... and
"COMPARE"...
[[[
COMPARE 5x5_1.6 to 5x5_trunk
1.23|+0.45 means factor=1.23, difference in seconds = 0.45
factor < 1 or difference < 0 means '5x5_trunk' is faster than
'5x5_1.6'
min max avg operation
1.151|+49.7471 0.863|-76.9170 1.036|+15.1526 TOTAL RUN
1.322|+0.0026 0.631|-1.4086 0.938|-0.0019 add
1.864|+0.0113 0.413|-11.9194 0.513|-3.5088 checkout
0.625|-2.3169 0.533|-24.8236 0.769|-4.0490 commit
0.899|-0.0030 0.129|-0.7784 0.122|-0.3825 copy
44.382|+100.3599 16.604|+122.9722 24.856|+109.4408 delete
1.245|+1.6928 0.943|-1.2191 0.768|-3.3842 merge
1.258|+0.0023 0.263|-1.8166 1.189|+0.0028 mkdir
1.256|+0.0022 0.187|-2.3871 0.910|-0.0016 prop mod
1.237|+0.0021 0.153|-0.4198 0.480|-0.0174 propdel
1.267|+0.0020 0.267|-0.2795 1.544|+0.0050 proplist
1.267|+0.0023 0.091|-5.4157 1.018|+0.0003 propset
1.148|+0.0013 2.347|+0.0133 1.254|+0.0024 resolve
2.764|+0.5885 0.878|-0.3298 1.565|+0.5413 resolved
1.410|+0.0733 3.061|+1.9744 2.038|+0.4877 status
0.602|-3.3563 0.422|-8.9009 0.470|-6.4426 switch
1.622|+0.0934 0.302|-21.6808 0.421|-3.9173 update
]]]
An interpretation could be: 'svn delete' currently sucks for some reason
(24 times slower???), as well as 'svn status' (twice as slow, but then
again it only added half a second to a run that was >400 seconds long,
on average). The rest looks quite encouraging. IIRC, when I ran this a
few months back I had higher slowness factor numbers than this (and,
notably, a much saner 'svn delete' factor, so there must be some newly
introduced speed bug in trunk's 'svn delete' -- or in my script).
Interesting to note is that the *variation* in time it takes for a run
became somewhat smaller using trunk. So trunk's timings seem to be
slightly more homogeneous than 1.6.x's (looking at the TOTAL min/max).
Now you may note that there are several types of these, i.e. 5x5, 1x100
and 100x1. 5x5 means there is a WC depth of 5 directories, each one
holding 5 subdirs and 5 files. 1x100 is a flat directory with 100 dir
and file entries. 100x1 is a cascade of single subdirs that is 100 dir
levels deep, also with one file on each subdir level.
The 100x1 case is actually where the 1.6 client sucks very badly -- it's
visible in timing but even worse memory wise. Watching `top` while it
runs reveals a memory footprint going towards gigabyte scale.
Consequently, this is where trunk really really ROCKS because it does
the same job with a few mb of RAM, in a fraction of the time.
I'm playing with the thought of making graphical charts of these
numbers, but that might be overkill :)
Any ideas/opinions about adding to tools/dev/ would be appreciated.
Finally, the interesting part that shows what the script tests:
[[[
run_cmd(['svnadmin', 'create', repos])
svn('checkout', file_url, wc)
trunk = j(wc, 'trunk')
create_tree(trunk, levels, spread)
add(trunk)
st(wc)
ci(wc)
up(wc)
propadd_tree(trunk, 0.5)
ci(wc)
up(wc)
st(wc)
trunk_url = file_url + '/trunk'
branch_url = file_url + '/branch'
svn('copy', '-mm', trunk_url, branch_url)
st(wc)
up(wc)
st(wc)
svn('checkout', trunk_url, wc2)
st(wc2)
modify_tree(wc2, 0.5)
st(wc2)
ci(wc2)
up(wc2)
up(wc)
svn('switch', branch_url, wc2)
modify_tree(wc2, 0.5)
st(wc2)
ci(wc2)
up(wc2)
up(wc)
modify_tree(trunk, 0.5)
st(wc)
ci(wc)
up(wc2)
up(wc)
svn('merge', '--accept=postpone', trunk_url, wc2)
st(wc2)
svn('resolve', '--accept=mine-conflict', wc2)
st(wc2)
svn('resolved', '-R', wc2)
st(wc2)
ci(wc2)
up(wc2)
up(wc)
svn('merge', '--accept=postpone', '--reintegrate', branch_url, trunk)
st(wc)
svn('resolve', '--accept=mine-conflict', wc)
st(wc)
svn('resolved', '-R', wc)
st(wc)
ci(wc)
up(wc2)
up(wc)
svn('delete', j(wc, 'branch'))
ci(wc)
up(wc2)
up(wc)
]]]
Note that I still have a few plans for tweaks I'm not discussing because
this mail is long enough.
~Neels
Re: some more performance benchmarks
Posted by Hyrum K Wright <hy...@hyrumwright.org>.
On Thu, Apr 7, 2011 at 6:29 AM, Stefan Sperling <st...@elego.de> wrote:
> Neels has set up his performance benchmarks to run on the svn VM
> we have at apache.org.
>
> The tests only use ra_Local and are supposed to measure wc-ng
> performance, not httpv2. They focus on edge cases and are thus a
> good addition to Mark Phippard's test which focus more on real world usage.
>
> Below are the results of the latest run.
> They're probably a bit hard to decipher as is. The plan is to make the
> scripts produce more easily digestable results and then send results
> to dev@ on a weekly basis.
>
> As far as I can tell, these numbers indicate that trunk is on-par
> with 1.6.x except for delete operations.
>
>...
This is great work, though as you admit the results are bit
incomprehensible at this stage. Would it make sense to somehow
combine the scripts to prevent the bifurcation of benchmark suites?
Do you have the script of commands, and the accompanying repo?
-Hyrum
Re: some more performance benchmarks
Posted by "C. Michael Pilato" <cm...@collab.net>.
On 04/07/2011 07:29 AM, Stefan Sperling wrote:
> Neels has set up his performance benchmarks to run on the svn VM
> we have at apache.org.
[...]
> Below are the results of the latest run.
> They're probably a bit hard to decipher as is. The plan is to make the
> scripts produce more easily digestable results and then send results
> to dev@ on a weekly basis.
Actually, I find them fairly straightforward to interpret. I particularly
like the comparison approach, providing multiplier values to demonstrate
slowerthan/fasterthan and the associated degree of either!
What's missing, I suppose, is an explanation of what each operation is truly
doing. By the varying N values, I'm guessing that there are some number of
high-level scenarios in play, but each step of the scenario is measured.
That is, if the scenario is "merge a branch into the trunk", you will have
measured each add of each file and dir, each commit to built up some
history, the copy to make the branch, and the merge itself, plus maybe some
status/proplist to verify state. Is that (roughly) what's going on? In
that case, explaining each "add" (for example) would be unwieldy. But maybe
explaining the high-level scenarios would be useless?
--
C. Michael Pilato <cm...@collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand