cache killer memory death test - 2.2.6 vs andrea2&4

Harvey J. Stein (hjstein@bfr.co.il)
29 Apr 1999 10:00:37 +0300


I just finished testing 2.2.6andrea4. Here are its numbers (along
with the other #s previously quoted), with 50,000 records:

test 1 - boot, login & run dbase on a file.
test 2 - boot, login, startx, run dbase on a file, bring up netscape,
edit in emacs, etc.

See previous msgs for dbase program & details.

Test 1 (noninteractive) Test 2 (interactive)
user sys elapsed worst 2nd user sys elapsed worst 2nd Feel
2.0.36 27.69 7.70 2:37.27 9.73 6.88 28.97 10.21 17:30.68 76.37 75.11 D
28.32 7.33 0:43.87 3.36 2.32
27.52 7.40 0:43.08 3.42 2.22

2.2.5 28.13 6.80 2:50.52 8.00 7.19 29.44 8.70 13:21.34 61.29 64.95 B
28.64 6.61 1:47.39 7.91 6.77
28.55 6.40 1:48.49 7.88 7.11

2.2.5arca12 27.99 6.77 1:24.68 14.93 9.38 30.13 6.91 4:13.60 14.08 13.09 F
28.98 6.69 1:19.67 14.12 9.56
28.18 6.77 1:18.99 13.89 9.39

2.2.6 28.20 6.70 2:47.06 7.47 6.87 31.73 8.92 14:41.84 16.57 16.53 B
28.92 6.17 1:46.01 7.97 6.62
28.69 6.39 1:47.77 7.81 6.76

2.2.6arca1 28.66 7.73 1:24.43 13.85 9.06 30.07 9.62 10:45.50 65.61 45.63 B
29.15 7.82 1:15.55 13.38 8.74
28.37 7.85 1:12.91 13.33 7.24

2.2.6a2** 28.04 7.24 0:42.24 2.82 2.10 29.89 7.33 1:30.70 6.09 4.34 A
28.44 7.06 0:42.37 2.82 2.10 30.45 7.73 1:25.50 5.09 4.27 A
28.02 6.96 0:41.94 2.79 2.09
27.88 7.23 0:42.39 2.80 2.09*

2.2.6a4+ 28.46 6.15 0:41.76 2.65 1.97 29.67 7.07 2:46.26 37.86 19.04 B-
29.46 6.37 0:42.00 2.67 2.00 29.30 6.89 4:04.15 55.05 46.47 B-
28.78 6.06 0:41.05 2.65 1.99 29.99 7.29 3:59.11 42.70 41.36 B-

* - run again directly from a reboot because I couldn't believe the #s!
** - 2.2.6a2 = 2.2.6andrea2.
+ - 2.2.6a4 = 2.2.6andrea4.

Feel = interactive feel = letter grade A=excellent to F=very poor.

For 100,000 records:

(s) - Call gdbm_sync every 10,000 records (which does an fsync)
(ns) - Without gdbm_sync call

2.2.6a2 (s) 85.48 21.33 2:35.95 12.16 9.43 91.16 24.42 10:51.33 51.41 46.82
88.08 21.60 2:38.40 12.06 9.51 87.56 21.85 3:12.71 13.88 9.72
85.13 22.01 2:35.85 12.17 9.41

2.2.6a4 (s) 87.53 18.32 2:32.91 11.95 9.05
86.04 18.33 2:30.59 11.91 8.81
87.64 17.40 2:31.46 11.94 8.85

2.2.6a2 (ns) 85.62 20.93 1:50.11 1.66 1.63 88.50 24.90 17:38.55 97.38 74.78
85.26 20.79 1:49.63 1.64 1.62
84.58 20.98 1:49.11 1.65 1.63

2.2.6a4 (ns) 88.26 17.32 1:48.39 1.63 1.61
84.06 17.66 1:44.51 1.58 1.58
84.63 17.95 1:45.15 1.57 1.58

Basically, andrea2 is a slightly slower than andrea4 in the
noninteractive tests and substantially *faster* in the interactive
tests. Given the poor interactive performance of andrea4 with the
50,000 record test, I didn't bother with a 100,000 record interactive
test on it.

I was surprised by the loss in interactive performance from a2 to a4,
which is why I ran test 2 again for a2 and 2 more times for a4. It's
still substantially better than the stock kernels, though.

It seems like a2 is the best at both avoiding disk activity & wisely
using the disk when necessary. A4 seems to be slightly more efficient
when using memory but much worse at using the disk when necessary.

-- 
Harvey J. Stein
BFM Financial Research
hjstein@bfr.co.il

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/