| From: |
| Con Kolivas <kernel@kolivas.org> |
| To: |
| linux kernel mailing list <linux-kernel@vger.kernel.org> |
| Subject: |
| [BENCHMARK] 2.6.0-test4{-mm2} with contest |
| Date: |
| Thu, 28 Aug 2003 11:38:47 +1000 |
| Cc: |
| Andrew Morton <akpm@osdl.org> |
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Here are contest (http://contest.kolivas.org) benchmarks with the dedicated
osdl (http://www.osdl.org) hardware for 2.6.0-test4 and mm2. mm2 includes my
O18.1 scheduler tweaks.
no_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 1 79 93.7 0.0 0.0 1.00
2.6.0-test4 1 80 93.8 0.0 0.0 1.00
2.6.0-test4-mm2 1 79 93.7 0.0 0.0 1.00
cacherun:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 1 76 97.4 0.0 0.0 0.96
2.6.0-test4 1 75 100.0 0.0 0.0 0.94
2.6.0-test4-mm2 1 75 98.7 0.0 0.0 0.95
process_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 2 109 68.8 64.5 28.4 1.38
2.6.0-test4 2 108 69.4 64.0 28.7 1.35
2.6.0-test4-mm2 2 125 59.2 92.5 38.4 1.58
ctar_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 3 112 71.4 1.0 5.4 1.42
2.6.0-test4 3 117 67.5 1.0 5.1 1.46
2.6.0-test4-mm2 3 115 69.6 1.0 6.1 1.46
xtar_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 3 122 63.1 2.0 4.9 1.54
2.6.0-test4 3 132 58.3 3.0 7.5 1.65
2.6.0-test4-mm2 3 141 54.6 3.3 7.0 1.78
io_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 4 119 64.7 43.8 19.3 1.51
2.6.0-test4 4 119 64.7 43.4 19.2 1.49
2.6.0-test4-mm2 4 110 70.0 35.4 16.4 1.39
io_other:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 2 116 67.2 48.2 20.7 1.47
2.6.0-test4 2 116 67.2 48.0 21.4 1.45
2.6.0-test4-mm2 2 110 70.0 41.0 18.8 1.39
read_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 2 106 74.5 8.3 6.6 1.34
2.6.0-test4 2 109 72.5 8.6 5.5 1.36
2.6.0-test4-mm2 2 106 73.6 8.3 5.6 1.34
list_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 2 95 81.1 0.0 7.4 1.20
2.6.0-test4 2 97 79.4 0.0 8.2 1.21
2.6.0-test4-mm2 2 96 80.2 0.0 7.3 1.22
mem_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 2 98 80.6 54.0 2.0 1.24
2.6.0-test4 2 101 78.2 67.0 3.0 1.26
2.6.0-test4-mm2 2 98 79.6 66.5 3.1 1.24
dbench_load:
Kernel [runs] Time CPU% Loads LCPU% Ratio
2.5.75 4 366 20.8 6.0 58.5 4.63
2.6.0-test4 4 390 19.5 6.2 57.3 4.88
2.6.0-test4-mm2 3 242 31.8 3.0 44.2 3.06
There is virtually no change between 2.5.75 and 2.6.0-test4 (which is expected
and desirable) apart from the nice increase in the amount of work done by
mem_load without any hit to the concurrent kernel compile.
There is a small prolongation of time under process_load (process_load is 16
tasks and the kernel compile is 4) but with an appropriate improvement in
work done by process_load and no wasted cpu time.
Conversely there is a significant drop in time under dbench_load (dbench 16 vs
make -j4) with equivalent cpu usage.
Otherwise slight shifts in the balance only.
Uniprocessor p3 833 with 256Mb Ram & IDE harddisks
Con
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
iD8DBQE/TV0wZUg7+tp6mRURAtQFAJwMXK8fu60D35zQ158eiVU2rwOBIwCfTbux
UTS6f8bMX11BbuippD3tJJI=
=VV5G
-----END PGP SIGNATURE-----
-
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/
