Thread 'CPU Benchmarks'

Message boards : BOINC client : CPU Benchmarks
Message board moderation

To post messages, you must log in.

AuthorMessage
Edward

Send message
Joined: 10 Apr 08
Posts: 2
United Kingdom
Message 16567 - Posted: 10 Apr 2008, 10:09:24 UTC

Hi guys. The CPU performance measures for my machine show much higner figures for when it had windows vista on it than at the minute, using ubuntu linux. Any ideas why? Do BOINC projects run slower on linux?
ID: 16567 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5144
United Kingdom
Message 16569 - Posted: 10 Apr 2008, 10:26:12 UTC - in response to Message 16567.  
Last modified: 10 Apr 2008, 10:41:52 UTC

Hi guys. The CPU performance measures for my machine show much higner figures for when it had windows vista on it than at the minute, using ubuntu linux. Any ideas why? Do BOINC projects run slower on linux?

Not necessarily, but possibly.

There are two quite separate issues.

a) The benchmark figure reported by BOINC. This is, frankly, meaningless for resolving questions like this. Ignore it.

b) The actual speed your CPU is running. If you've been running Vista, you presumably have a reasonably modern and powerful CPU in your computer. These usually have a power-saving or cool-running slow mode, and it's possible that Linux has selected this mode for the low-priority science applications launched by BOINC.

We had a long discussion about this at SETI recently: that link drops you in the middle, at the best answer: Toby says to use

cat /proc/cpuinfo

to find the current CPU speed, and

sudo cpufreq-selector -g performance

to fix it if it turns out to be slow.

Edit - I did in fact put in a request for this information to be added to the BOINC FAQ, so that all help-desk advisors would be aware of it. Not unreasonably I got a reply that someone with personal experience, and understanding of the CPUs/Linux versions affected, would be best placed to do the write-up. I don't have the experience to be that person: any takers?
ID: 16569 · Report as offensive
ProfileKSMarksPsych
Avatar

Send message
Joined: 30 Oct 05
Posts: 1239
United States
Message 16574 - Posted: 10 Apr 2008, 11:57:22 UTC - in response to Message 16569.  

cat /proc/cpuinfo

to find the current CPU speed, and

sudo cpufreq-selector -g performance

to fix it if it turns out to be slow.

Edit - I did in fact put in a request for this information to be added to the BOINC FAQ, so that all help-desk advisors would be aware of it. Not unreasonably I got a reply that someone with personal experience, and understanding of the CPUs/Linux versions affected, would be best placed to do the write-up. I don't have the experience to be that person: any takers?


*shrug*

Fedora 7 with KDE has a GUI to do this. But I don't think I ever changed what it was set at (dynamic FWIW, choices are Performance, Dynamic and Powersave). On second glance, it's probably specific to KDE because it's in KPowersave. I'd guess that Gnome has something similar though. I'll need to boot back to Windows this weekend, so before I go back to Fedora I'll poke around in Ubuntu/Gnome and see what I can find.

Kathryn :o)
ID: 16574 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5144
United Kingdom
Message 16576 - Posted: 10 Apr 2008, 12:38:28 UTC - in response to Message 16574.  

Fedora 7 with KDE has a GUI to do this. But I don't think I ever changed what it was set at (dynamic FWIW, choices are Performance, Dynamic and Powersave). On second glance, it's probably specific to KDE because it's in KPowersave. I'd guess that Gnome has something similar though. I'll need to boot back to Windows this weekend, so before I go back to Fedora I'll poke around in Ubuntu/Gnome and see what I can find.

It probably also matters what SpeedStep / Cool'n'Quiet controls are available in your BIOS, and what they're currently set at.

I think this is a difficult class of problem for volunteer helpers to deal with. It's most likely to be reported at a project level (and with a complaint of RAC falling, rather than benchmarks as here): and I would actually expect BOINC benchmarks to run at full speed (normal priority), and the CPU speed to drop only when the computer is otherwise unused and the project science apps are the only significant thing running.

But the issue wouldn't be specific to any one particular project: it's genuinely a BOINC issue, because it arises from BOINC assigning low CPU priority ("nice") to project science apps. That raises the question of the relationship between BOINC, and project, helpdesks and discussion forums. I think I'll just leave that one hanging.
ID: 16576 · Report as offensive
Conan

Send message
Joined: 2 Apr 08
Posts: 4
Australia
Message 16577 - Posted: 10 Apr 2008, 12:49:50 UTC

Benchmarks with 5.10.45 on one of my AMD Opteron 285 @2.6GHz is over 20% less than with 5.10.21, on my Linux Fedora Core 3 system.

This has no bearing on processing speed, but does affect how much work you can get. That is until Boinc Manager works out that speed is better than benchmark for the computer, and then you will get more work.

Benchmarks under Windows XP running 5.10.38 are over 10% higher than Linux computer with the same specs, Opteron 285 running 5.10.21.
ID: 16577 · Report as offensive
Edward

Send message
Joined: 10 Apr 08
Posts: 2
United Kingdom
Message 16739 - Posted: 19 Apr 2008, 19:34:49 UTC

Thank you! As it turns out the CPU wasnt quite running to its maximum potential:

edward@SCIMITAR:~$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 15
model name : Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz
stepping : 11
cpu MHz : 1998.000
cache size : 4096 KB
physical id : 0
siblings : 2
core id : 0
cpu cores : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 10
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
bogomips : 5323.57
clflush size : 64

processor : 1
vendor_id : GenuineIntel
cpu family : 6
model : 15
model name : Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz
stepping : 11
cpu MHz : 1998.000
cache size : 4096 KB
physical id : 0
siblings : 2
core id : 1
cpu cores : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 10
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
bogomips : 5319.99
clflush size : 64

edward@SCIMITAR:~$ sudo cpufreq-selector -g performance
[sudo] password for edward:
edward@SCIMITAR:~$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 15
model name : Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz
stepping : 11
cpu MHz : 2664.000
cache size : 4096 KB
physical id : 0
siblings : 2
core id : 0
cpu cores : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 10
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
bogomips : 5323.57
clflush size : 64

processor : 1
vendor_id : GenuineIntel
cpu family : 6
model : 15
model name : Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz
stepping : 11
cpu MHz : 1998.000
cache size : 4096 KB
physical id : 0
siblings : 2
core id : 1
cpu cores : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 10
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr lahf_lm
bogomips : 5319.99
clflush size : 64
ID: 16739 · Report as offensive

Message boards : BOINC client : CPU Benchmarks

Copyright © 2025 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.