Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 13
Posts: 13   Pages: 2   [ 1 2 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 1839 times and has 12 replies Next Thread
Alex Turovsky
Cruncher
Israel
Joined: Dec 1, 2005
Post Count: 34
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Processor performance comparison in WOrld Community Grid project

*ADMINS, I think this chart should star in some important palce as I've never seen one before at the Community-maintained FAQs. Thank you.
-------------------------------------------------------------------------------

I made a CPU power comparison in computers under my command.

Results:- Overclocking makes no difference
- Big L2 Cache on your CPU makes all the difference
- Intel new i7 CPUs are crunching beasts since with their Hyper Threading technology they process 8 tasks in any given minute.

Comparison chart:


Here is how I managed to see the scores. (The Q9450 at stock speed is running only half a day, sio it's similar to the overclocked Q9450):

*Please replace your images with appropriate size. tks**tkh
----------------------------------------
Team Leader of HWzone.co.il, the biggest Israeli WCG team.
----------------------------------------
[Edit 3 times, last edit by TKH at Feb 2, 2009 2:58:40 PM]
[Jan 31, 2009 10:19:55 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Sekerob
Ace Cruncher
Joined: Jul 24, 2005
Post Count: 20043
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Hi Alex,

May i ask you to shrink you image a bit. Many do not have very wide screens to view this.

As for the information, results is a debateble measure because the continuous flex of job durations. Credit or points per day would add to the "feel" of the numbers. Many refer to RAC (Recent Average Credit) as a measure of performance.

good going and cheers
----------------------------------------
WCG Global & Research > Make Proposal Help: Start Here!
Please help to make the Forums an enjoyable experience for All!
[Jan 31, 2009 11:30:55 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Alex Turovsky
Cruncher
Israel
Joined: Dec 1, 2005
Post Count: 34
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Regarding the picture, right click on any image>properties>copy the links>paste into browser.

You'll get this:
http://img134.imageshack.us/img134/4954/wcgsurveyji9.jpg

Regarding the results:
You are right, resaults DO vary each day. that's why all my results based on average of a few months.

Like if Q9450 did 20 or 10 a day, I wouldn't count it at all. only, the mainstream numbers.

How does RAC works ?
----------------------------------------
Team Leader of HWzone.co.il, the biggest Israeli WCG team.
[Jan 31, 2009 11:47:56 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Sekerob
Ace Cruncher
Joined: Jul 24, 2005
Post Count: 20043
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

RAC is a sliding 60 day average I think.
----------------------------------------
WCG Global & Research > Make Proposal Help: Start Here!
Please help to make the Forums an enjoyable experience for All!
[Jan 31, 2009 11:59:22 AM]   Link   Report threatening or abusive post: please login first  Go to top 
GIBA
Ace Cruncher
Joined: Apr 25, 2005
Post Count: 5374
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Hi Alex,
off topic could you please copntact me via e-mail (see link below).
Thank you in advance. coffee
----------------------------------------
Cheers ! GIB@ peace coffee
Join BRASIL - BRAZIL@GRID team and be very happy !
http://www.worldcommunitygrid.org/team/viewTeamInfo.do?teamId=DF99KT5DN1

[Jan 31, 2009 3:27:02 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Alex, you picked a bad time for your experiment. With the recent changes to work unit length (from 10 hours to 6), and the unusually large variance in CEP tasks, and the fixed length for NRW, your result counts can't be relied on.

Some of your figures, for example the Q9450, are clearly wrong.

Thank you for your efforts. If the numbers can be made more reliable, then it will be a great resource.

You misunderstood Sekerob's request. We have a rule that images must be no bigger than 600x600, for the convenience of readers (and due to forum limitations).
[Jan 31, 2009 3:55:05 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Information about Recent Average Credit can be found here: http://www.boinc-wiki.info/RAC
[Jan 31, 2009 4:53:01 PM]   Link   Report threatening or abusive post: please login first  Go to top 
twilyth
Master Cruncher
US
Joined: Mar 30, 2007
Post Count: 2130
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

I agree. The concept is excellent and you deserve credit for your efforts but the Q9450 results cannot be correct. Processor efficiency is almost linear with speed (for varying clock speeds on a particular model CPU). A Q6600 running at 3.6ghz (150% of stock) will be approximately 50% faster. You can see this just by running the boinc benchmarking utility from within the boinc task manager.

What would be much more useful in my opinion would be to see average points per day for each model. That would eliminate any issues regarding variations in work units. The points system is supposed to reflect the efficiency of each type of system so average PPD (RAC) should be a good indication of processor speed.

I don't think you need to do 60 days of data though. A week or two should be fine.

In the mean time, could you please run boinc benchmarks (whetstone, drystone) for each machine and post the results. The FLOPs is the most useful information.

Thank you.
----------------------------------------


[Jan 31, 2009 4:54:15 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Alex Turovsky
Cruncher
Israel
Joined: Dec 1, 2005
Post Count: 34
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

Hi Alex,
off topic could you please copntact me via e-mail (see link below).
Thank you in advance. coffee

Can't find your email (?).



(RAC) should be a good indication of processor speed.

could you please run boinc benchmarks (whetstone, drystone) for each machine and post the results. The FLOPs is the most useful information.


1. Here is the table from other boinc site, some centralized project site:
http://statsnstones.tswb.org/UserDetail.aspx?...4579b7b328f9f8f942a799d1a

2. I can't do the test you asked since some computers located in cities near Gaza Strip border at the south west and HAMAS terrorists bombard citizens in these areas with missiles daily (yes even after the military operation Israeli army did there) so until they stop, I won't go there.

3. I have only these results (what site is it anyway ?):
*only Q6600,Q9450& i7 920 are 24\7\100% comparable
http://statsnstones.tswb.org/UserDetail.aspx?...4579b7b328f9f8f942a799d1a

- can I somehow get e-mail notification regarding new posts ? I didn't even know you guys answered until going here manually.
----------------------------------------
Team Leader of HWzone.co.il, the biggest Israeli WCG team.
----------------------------------------
[Edit 3 times, last edit by Alex Turovsky at Feb 2, 2009 10:42:32 AM]
[Feb 2, 2009 10:38:53 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Processor performance comparison in WOrld Community Grid project

I see you have ignored the request above -

We have a rule that images must be no bigger than 600x600,
for the convenience of readers (and due to forum limitations).
[Feb 2, 2009 11:08:06 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 13   Pages: 2   [ 1 2 | Next Page ]
[ Jump to Last Post ]
Post new Thread