Recent Reaim results (STP)

From: cliff white
Date: Tue Feb 10 2004 - 13:13:46 EST



There hasn't been much big change in reaim runs lately,
but just to show we're Not Dead Yet, I
did some quick graphs for some recent kernels:
( 2.6.3-rc2 vs 2.6.2-rc3-mm1 )


http://developer.osdl.org/cliffw/reaim/compares/index.html

All tests were run on ext3. System descriptions here:
http://www.osdl.org/stp

Noteworthy: 2.6.3-rc2 is a tiny,tiny,itsy bit slower than 2.6.2.

2.6.3-rc1-mm1 is 5.79 percent faster than 2.6.3-rc2 on the
2-way compute run. This is new - usually the results of
the compute test do not differ between mainline and -mm.

Numbers:


1-way
Database load
Kernel JPM Lusers Pct Change
patch-2.6.3-rc2 1001.21 17 0.00
2.6.2-rc3-mm1 982.31 17 -1.89
linux-2.6.2 1000.03 17 -0.12

Compute load
patch-2.6.3-rc2 1025.48 17 0.00
2.6.2-rc3-mm1 1006.36 17 -1.86
linux-2.6.2 1024.82 17 -0.06

Fileserver load
patch-2.6.3-rc2 3741.72 62 0.00
2.6.3-rc1-mm1 3712.67 62 -0.78
linux-2.6.2 3748.21 63 0.17

2-way
Database load
patch-2.6.3-rc2 1285.31 22 0.00
2.6.3-rc1-mm1 1321.30 22 2.80
linux-2.6.2 1297.90 22 0.98

Compute load
patch-2.6.3-rc2 1446.99 26 0.00
2.6.3-rc1-mm1 1530.82 26 5.79
linux-2.6.2 1455.38 26 0.58

Fileserver load
patch-2.6.3-rc2 6190.24 104 0.00
2.6.3-rc1-mm1 6123.20 102 -1.08
linux-2.6.2 6173.62 104 -0.27

4-way
Database load
patch-2.6.3-rc2 5288.33 88 0.00
2.6.2-rc3-mm1 5293.70 88 0.10
linux-2.6.2 5246.54 88 -0.79

Compute
patch-2.6.3-rc2 5169.71 88 0.00
2.6.2-rc3-mm1 5159.36 88 -0.20
linux-2.6.2 5206.06 88 0.70

Fileserver load
patch-2.6.3-rc2 10343.18 172 0.00
2.6.2-rc3-mm1 10254.39 168 -0.86
linux-2.6.2 10351.30 176 0.08

8-way
Database load ( Deadline scheduler )
patch-2.6.3-rc1 8591.23 88 0.00
2.6.2-rc3-mm1 9082.65 144 5.72
linux-2.6.2 8565.25 128 -0.30

Compute load
patch-2.6.3-rc1 9640.14 160 -0.11
2.6.2-rc3-mm1 9645.81 160 -0.05
linux-2.6.2 9711.96 160 0.63

cliffw
OSDL
--
The church is near, but the road is icy.
The bar is far, but i will walk carefully. - Russian proverb
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/