PDA

View Full Version : Please post you results here

Graaf_van_Vlaanderen
October 15th, 2009, 14:39
I'm collecting some data from benchmarks/ubench running on Freebsd.
Feel free to post your results, including which CPU, motherboard and RAM.

Here are two machines I tested:

1. HP DL360 G4, One Xeon 3.4GHZ (1MB) cache, 2GB PC2700 ECC.


%ubench
Unix Benchmark Utility v.0.3
Copyright (C) July, 1999 PhysTech, Inc.
Author: Sergei Viznyuk <sv@phystech.com>
http://www.phystech.com/download/ubench.html
FreeBSD 7.2-RELEASE-p4 FreeBSD 7.2-RELEASE-p4 #0: Fri Oct 2 08:22:32 UTC 2009 root@amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC amd64
Ubench CPU: 181412
Ubench MEM: 114904
--------------------
Ubench AVG: 148158

2. HP DL360 G4, Two 3GHz (1MB) cache, 8GB PC2700:


Unix Benchmark Utility v.0.3
Copyright (C) July, 1999 PhysTech, Inc.
Author: Sergei Viznyuk <sv@phystech.com>
http://www.phystech.com/download/ubench.html
FreeBSD 7.2-RELEASE-p4 FreeBSD 7.2-RELEASE-p4 #0: Fri Oct 2 08:22:32 UTC 2009 root@amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC amd64
Ubench CPU: 308282
Ubench MEM: 124371
--------------------
Ubench AVG: 216326
DutchDaemon
October 15th, 2009, 15:02
I'm not too sure about ubench. During the CPU test it uses 100% of one CPU, and only a fraction (and mostly none) of the other CPU. This is on a 'true' dual-core system.


14410 root 1 118 0 3388K 1096K CPU1 0 1:43 100.00% ubench
14405 root 1 44 0 3388K 1096K piperd 0 0:19 0.00% ubench


In fact: ubench never goes over 100% total CPU. It starts with about 50% on either CPU, and then maxes out one CPU and forgets about the other. A true dual CPU test should max out both CPUs (2 * 100%), no?

And this is what happens during the MEM test ...


14433 root 1 114 0 88380K 86192K CPU1 1 0:09 85.89% ubench
14432 root 1 114 0 14652K 12284K RUN 0 0:10 84.08% ubench
14405 root 1 49 0 4412K 2796K piperd 0 0:31 22.56% ubench


More CPU power in use than during the CPU test ..

Either I misunderstand these tests, or they're not right.
Graaf_van_Vlaanderen
October 15th, 2009, 15:28
I'm not too sure about ubench. During the CPU test it uses 100% of one CPU, and only a fraction (and mostly none) of the other CPU. This is on a 'true' dual-core system.


14410 root 1 118 0 3388K 1096K CPU1 0 1:43 100.00% ubench
14405 root 1 44 0 3388K 1096K piperd 0 0:19 0.00% ubench


In fact: ubench never goes over 100% total CPU. It starts with about 50% on either CPU, and then maxes out one CPU and forgets about the other. A true dual CPU test should max out both CPUs (2 * 100%), no?

And this is what happens during the MEM test ...


14433 root 1 114 0 88380K 86192K CPU1 1 0:09 85.89% ubench
14432 root 1 114 0 14652K 12284K RUN 0 0:10 84.08% ubench
14405 root 1 49 0 4412K 2796K piperd 0 0:31 22.56% ubench


More CPU power in use than during the CPU test ..

Either I misunderstand these tests, or they're not right.


That's indeed strange, since it works fine on a Q6600 and Opteron 275.
This is what I see in top, 2 CPU's and HT enabled:


last pid: 33930; load averages: 3.57, 2.10, 1.20 up 0+01:00:52 16:22:46
35 processes: 5 running, 30 sleeping
CPU 0: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 1: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 2: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 3: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
Mem: 21M Active, 457M Inact, 442M Wired, 1032K Cache, 399M Buf, 6939M Free
Swap: 4096M Total, 4096M Free

PID UID THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
33914 1001 1 118 0 5732K 1196K CPU1 1 1:20 100.00% ubench
33916 1001 1 118 0 5732K 1196K RUN 3 1:12 100.00% ubench
33917 1001 1 118 0 5732K 1196K CPU2 2 1:10 100.00% ubench
33915 1001 1 114 0 5732K 1196K CPU0 0 1:20 84.47% ubench
33913 1001 1 -8 0 5732K 1196K piperd 2 0:12 0.00% ubench
DutchDaemon
October 15th, 2009, 15:30
Maybe it can only reliably handle multiple CPUs, not multiple cores.
nimnod
October 15th, 2009, 15:47
Dell R200, One Pentium Dual E2220 @2.4Ghz, 2GB DDR2 800Mhz

Ubench CPU: 596122
Ubench MEM: 183245
--------------------
Ubench AVG: 389683

IBM System x3200, One Pentium D @2.8Ghz, 1GB DDR2 667Mhz

Ubench CPU: 249049
Ubench MEM: 139555
--------------------
Ubench AVG: 194302

I will soon lay my hands on some a dual QC Xeons 5000 series, running FreeBSD, will post further results then.
DutchDaemon
October 15th, 2009, 15:59
This is what I get on a single-core CPU with HT enabled during the CPU test:

65331 root 1 117 0 3388K 836K CPU1 1 0:31 100.00% ubench
65332 root 1 118 0 3388K 836K RUN 0 0:31 100.00% ubench
65252 root 1 44 0 3388K 836K piperd 1 0:22 2.10% ubench

Which is what one might expect. No idea why this doesn't show up on a dual-core CPU.
vivek
October 15th, 2009, 16:05
I don't think so it is useful as software is not updated in ages.
sbe
October 15th, 2009, 16:06
Noname, AMD Athlon 64 X2 Dual Core Processor 5200+, 3GB DDR2 667Mhz

Ubench CPU: 282880
Ubench MEM: 140008
--------------------
Ubench AVG: 211444
Oxyd
October 15th, 2009, 16:13
AMD Athlon(tm) 64 X2 Dual Core Processor 5400+ (2812.82-MHz 686-class CPU), 3 GB (1 + 1 + 0.5 + 0.5) DDR2 533 MHz RAM:

Ubench CPU: 251910
Ubench MEM: 148350
--------------------
Ubench AVG: 200130

It did max out both my CPU cores, according to top -P. Also, this is a 7.2-RELEASE-p4 i386.
nimnod
October 15th, 2009, 16:21
I don't think so it is useful as software is not updated in ages.
Yet still it provides a measure to compare hardware configurations, which I find useful. The problem we may face, is ubench may not make use of new instruction sets thus producing false results while comparing contemporary CPUs.
Graaf_van_Vlaanderen
October 15th, 2009, 17:19
Two dual core Opteron 275, 4GB ECC PC3200 memory.
Motherboad: Tyan S2892
The memory benchmark is much lower than PC2700 (HP DL360 G4).
Perhaps this is because the Xeons make use of an external memory
controller.


%ubench
Unix Benchmark Utility v.0.3
Copyright (C) July, 1999 PhysTech, Inc.
Author: Sergei Viznyuk <sv@phystech.com>
http://www.phystech.com/download/ubench.html
FreeBSD 8.0-RC1 FreeBSD 8.0-RC1 #0: Thu Sep 17 18:50:57 UTC 2009 root@mason.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC amd64
Ubench CPU: 472636
Ubench MEM: 64692
--------------------
Ubench AVG: 268664



last pid: 1785; load averages: 5.04, 2.81, 1.61 up 0+00:13:38 18:07:24
36 processes: 8 running, 28 sleeping
CPU 0: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 1: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 2: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
CPU 3: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
Mem: 14M Active, 15M Inact, 95M Wired, 120K Cache, 41M Buf, 3806M Free
Swap: 4096M Total, 4096M Free

PID UID THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
1782 1001 1 113 0 5920K 1340K RUN 3 0:46 78.66% ubench
1784 1001 1 111 0 5920K 1340K CPU3 3 0:38 70.75% ubench
1781 1001 1 108 0 5920K 1340K CPU2 2 0:41 55.66% ubench
1780 1001 1 107 0 5920K 1340K RUN 1 0:31 52.49% ubench
1779 1001 1 107 0 5920K 1340K CPU1 1 0:32 52.39% ubench
1785 1001 1 107 0 5920K 1340K RUN 0 0:32 51.66% ubench
1783 1001 1 107 0 5920K 1340K RUN 0 0:34 49.46% ubench
1777 1001 1 44 0 5920K 1340K piperd 2 0:16 0.00% ubench