Difference between GHZ and FSB?

Message boards : Number crunching : Difference between GHZ and FSB?

To post messages, you must log in.

AuthorMessage
Orgil

Send message
Joined: 11 Dec 05
Posts: 82
Credit: 169,751
RAC: 0
Message 50572 - Posted: 11 Jan 2008, 17:02:31 UTC

How come 1.73GHZ T2250 533fsb cpu crunch faster than 3ghz P4 HT 800fsb cpu? both used only one core (for HT it is one thread).
ID: 50572 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 50574 - Posted: 11 Jan 2008, 17:40:24 UTC

Please explain how you determined one is faster then the other. Are you looking at time to complete a WU? Or RAC? Or BOINC benchmark ratings? Other?
Rosetta Moderator: Mod.Sense
ID: 50574 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Luuklag

Send message
Joined: 13 Sep 07
Posts: 262
Credit: 4,171
RAC: 0
Message 50577 - Posted: 11 Jan 2008, 18:02:00 UTC

the good way to do this, is compare benchmark (drystone/whetstone)
ID: 50577 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Orgil

Send message
Joined: 11 Dec 05
Posts: 82
Credit: 169,751
RAC: 0
Message 50578 - Posted: 11 Jan 2008, 18:34:12 UTC
Last modified: 11 Jan 2008, 18:40:33 UTC

I observed these cpu's for Rosetta and Seti and seemingly T2250 is mostly same and some time faster than P4 3ghz. I mean in crunching speed.

In each WU report you can see the cpu time from there I saw this difference.

If I consider ghz and fsb the result supposed to be opposite but in reality the T2250 is beating.
ID: 50578 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
jegs

Send message
Joined: 16 May 07
Posts: 9
Credit: 80,767
RAC: 0
Message 50583 - Posted: 11 Jan 2008, 19:55:58 UTC - in response to Message 50572.  

How come 1.73GHZ T2250 533fsb cpu crunch faster than 3ghz P4 HT 800fsb cpu? both used only one core (for HT it is one thread).


The T2250 uses the new "Core" technology and the P4's use the old "Netburst" technology. The newer processors with "Core" technology can do more work per clock cycle than the old P4's with "Netburst". It's like comparing apples to oranges.
ID: 50583 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 118,331,123
RAC: 25,769
Message 50585 - Posted: 11 Jan 2008, 20:13:45 UTC

drystone/whetstone don't test cache or memory throughput - very basic benchmarks and so aren't much use for determining a processor's crunching ability. As jegs posted, the core-based processors (and their predecessors such as the pentium-m dothans) are much more efficient than the P4 netburst architecture. They also have larger caches which helps.

Core's succeessor (core2) is quite a bit more efficient again - my core2 based E6420 (3.2GHz) is here: https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=481396
to compare the results.

FSB has little impact on Rosetta on the core/core2 platforms AFAIK. I don't think memory throughput is a limiting factor much of the time. That might be different on other architectures though, especially if the CPU has a smaller cache.

HTH
Danny
ID: 50585 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
AlphaLaser

Send message
Joined: 19 Aug 06
Posts: 52
Credit: 3,327,939
RAC: 0
Message 50602 - Posted: 12 Jan 2008, 5:24:26 UTC
Last modified: 12 Jan 2008, 5:25:03 UTC

When comparing CPU's between two different architectures, making a speed comparison based only on clock speeds is no longer valid. Different CPU architectures can have different instructions per clock (IPC) among other factors that make that architecture fast or slow irrespective of the clock.

For example, the Pentium M/Yonah architecture has a short, efficient instruction pipeline whereas NetBurst's is quite long, resulting in slow throughput clock-for-clock. NetBurst was originally designed to overcome this problem by clocking to extremely high frequencies (5 GHz+). Unfortunately NetBurst hits a thermal and power "wall" long before that happens.
ID: 50602 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Orgil

Send message
Joined: 11 Dec 05
Posts: 82
Credit: 169,751
RAC: 0
Message 50615 - Posted: 12 Jan 2008, 16:13:59 UTC

Thanks for the great descriptions, it is too pity that Intel did not release fair descriptions of their products and mislead its non IT professed customers.

Before this architecture talk I was considering ghz and fsb are the main stuffs to measure a cpu power.

ID: 50615 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Gerry Rough
Avatar

Send message
Joined: 2 Jan 06
Posts: 111
Credit: 1,389,340
RAC: 0
Message 50669 - Posted: 13 Jan 2008, 22:20:28 UTC - in response to Message 50615.  

Thanks for the great descriptions, it is too pity that Intel did not release fair descriptions of their products and mislead its non IT professed customers.

Before this architecture talk I was considering ghz and fsb are the main stuffs to measure a cpu power.


For the rest of us who are not IT buffs, when does the fast FSB kick in for crunching speeds? When I upgraded my boxes recently, I chose the faster clock speeds when possible, and wondered about FSB speed for crunching (didn't have any faster options here though).


(Click for detailed stats)
ID: 50669 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Pepo
Avatar

Send message
Joined: 28 Sep 05
Posts: 115
Credit: 101,358
RAC: 0
Message 50670 - Posted: 13 Jan 2008, 22:41:07 UTC - in response to Message 50615.  

Before this architecture talk I was considering ghz and fsb are the main stuffs to measure a cpu power.

This was not true for a very long time already. For instance, a 1GHz P4 (OK, maybe not exactly this frequency, but take it as approx. example) was slower than 733 MHz P3.

Peter
ID: 50670 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Difference between GHZ and FSB?



©2024 University of Washington
https://www.bakerlab.org