What setup is giving me the best performance?

On Mon, 16 Feb 2004 22:22:54 +0000, Geoff wrote:

 > Gene wrote:
  >> All things remaining the same (board & memory), which chip is giving
  >> me the best overall performance?
  >>
  >> An AMD 1700 DLT3CJIUHB 0310XPMW, L2 cache 256, 512 MB DDR SDRAM @
  >> 333DDR, clocked to 2312 MHz.
  >> OR an AMD 2600 Barton, L2 cache 512, 512 MB DDR SDRAM @ 400DDR,
  >> clocked to 2150 MHz.
  >>
  >> Does the Barton's 512 L2 cache along with the 400DDR compensate for
  >> the 162 MHz difference from the 1700 JIUHB?
  >>
 >
 > seti@home benchmark times for the baove:
 > 3.17 hours (3 hr 10 min) for tbred b @ 2312 cpu 333 ram
 > 2.28 hours (2 hr 17 min) for barton @ 2150 cpu 400 ram
 >
 > no question really

All you've really proved is that Seti makes use of the extra cache. In
other benchmarks, the cpu with the higher clockspeed will win. You also
don't say if you have memory in sync with FSB testing each cpu.
Performance will degrade considerably running the FSB and memory bus out
of sync. This could be another reason the barton shows so good. So
basically your seti benchmarks are worthless.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
<a style='text-decoration: underline;' href="http://mysite.verizon.net/res0exft/cpu.htm" target="_blank">http://mysite.verizon.net/res0exft/cpu.htm</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: What setup is giving me the best performance? 

The Barton is set at a FSB of 187 with a multipier of 11.5
The Throughbred is set at a FSB of 171 with a multiplier of 13.5

With this Abit AT7-Max2's AGP linked to the FSB, I know there are
limitations to what I can do. Until I change boards, I am trying to
determine what chip to run.


"Gene" <gene RemoveThis @nospama.com> wrote in message
news:peqdnZrcirgvq6zdRVn-vw@bright.net...
 > All things remaining the same (board & memory), which chip is giving me
the
 > best overall performance?
 >
 > An AMD 1700 DLT3CJIUHB 0310XPMW, L2 cache 256, 512 MB DDR SDRAM @ 333DDR,
 > clocked to 2312 MHz.
 > OR an AMD 2600 Barton, L2 cache 512, 512 MB DDR SDRAM @ 400DDR, clocked
to
 > 2150 MHz.
 >
 > Does the Barton's 512 L2 cache along with the 400DDR compensate for the
162
 > MHz difference from the 1700 JIUHB?
 >
 > Thanks for any thoughts,
 >
 > Gene
 > Abit AT7-Max2
 >
 >


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.587 / Virus Database: 371 - Release Date: 2/12/04<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: What setup is giving me the best performance? 

Wes Newell wrote:
 > On Mon, 16 Feb 2004 22:22:54 +0000, Geoff wrote:
 >
  >> Gene wrote:
   >>> All things remaining the same (board & memory), which chip is giving
   >>> me the best overall performance?
   >>>
   >>> An AMD 1700 DLT3CJIUHB 0310XPMW, L2 cache 256, 512 MB DDR SDRAM @
   >>> 333DDR, clocked to 2312 MHz.
   >>> OR an AMD 2600 Barton, L2 cache 512, 512 MB DDR SDRAM @ 400DDR,
   >>> clocked to 2150 MHz.
   >>>
   >>> Does the Barton's 512 L2 cache along with the 400DDR compensate for
   >>> the 162 MHz difference from the 1700 JIUHB?
   >>>
  >>
  >> seti@home benchmark times for the baove:
  >> 3.17 hours (3 hr 10 min) for tbred b @ 2312 cpu 333 ram
  >> 2.28 hours (2 hr 17 min) for barton @ 2150 cpu 400 ram
  >>
  >> no question really
 >
 > All you've really proved is that Seti makes use of the extra cache. In
 > other benchmarks, the cpu with the higher clockspeed will win. You
 > also don't say if you have memory in sync with FSB testing each cpu.
 > Performance will degrade considerably running the FSB and memory bus
 > out of sync. This could be another reason the barton shows so good. So
 > basically your seti benchmarks are worthless.

And if those figures are taken from that SETI site somewhere where you input
CPU type, clockspeed, mem speed etc. and hit 'calculate' I can tell you
categorically that it is often wrong, very wrong. I've tried it with several
combos I've run and then compared the result with average times from
SETISpy's log. It varies widely from RL performace.

My TbredB at 2.1GHz takes between 2.1 and 2.7 hours to do a WU, averages 2.4
hours. It is on a 200MHZ FSB though.
--
~misfit~<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: What setup is giving me the best performance?