mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2007-05-18, 22:27   #23
Cruelty
 
Cruelty's Avatar
 
May 2005

31348 Posts
Default

Quote:
Originally Posted by S485122 View Post
Ok, the conclusion is then that a QuadCore is worth 3 cores with the P965 chipset, 2.6 with 975 and only 2 cores with the nVidia chipsets. Since memory is the bottleneck, memory capable of running 2:1 that is 8500 or 1067 MHz is a must .
The tests each of us has performed are not comparable so your claim is not valid I'm afraid
Cruelty is offline   Reply With Quote
Old 2007-05-19, 02:22   #24
dsouza123
 
dsouza123's Avatar
 
Sep 2002

2·331 Posts
Default

The Barcelona (server), or Phenom (desktop)
quad cores speeds are to be 2.6, 2.7, 2.8 and 2.9 GHz.

The AM2 socket for 2.6, 2.7, 2.9 and the 2.8 Quad FX on socket F.

The prefetch will be 32 bytes, doubling the 16 bytes from previous CPUs.
dsouza123 is offline   Reply With Quote
Old 2007-05-19, 07:59   #25
S485122
 
S485122's Avatar
 
"Jacob"
Sep 2006
Brussels, Belgium

24·109 Posts
Default

Quote:
Originally Posted by Cruelty View Post
The tests each of us has performed are not comparable so your claim is not valid I'm afraid
If I look at the benchmark pages I find iteration times of 44,3 ms for 2048K exponents, my Q67 with the 8500 memory uses an average of 53,5 ms for each of the four instances and 42,9 ms if only one instance is running. So I even get 42,9/53,5*4 = 3,2 cores out of it. My AVERAGE iteration times are better than the BEST times from the benchmark and this is, AFAIK, due to the chipset used.

The same system used with 5400 memory had iteration times of 84 ms. The iteration times are almost exactly proportional to the memory speed. And since the memory used is among the fastest on the market for this architecture*, it is my conclusion that the memory controller/bus needs to be enhanced to provide a sufficient data flow for the Intel Quads.

As George said in a post Prime95 benchmarks are not the best way to benchmark a system for real life throughput, it is better to use real tests.

Jacob

*There is DDR2 10000 memory on the market but the difference with 8500 is only about 17%. DDR3 and DDR4 would be a solution to this problem if that memory really has a triple or quadruple data flow.

Last fiddled with by S485122 on 2007-05-19 at 07:59
S485122 is offline   Reply With Quote
Old 2007-05-19, 08:41   #26
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

11001001010012 Posts
Default

Quote:
Originally Posted by S485122 View Post
*There is DDR2 10000 memory on the market but the difference with 8500 is only about 17%. DDR3 and DDR4 would be a solution to this problem if that memory really has a triple or quadruple data flow.
Unfortunately, DDR3 is double-rate memory, albeit with faster official clock speeds, lower power and a more exciting name. There will be DDR3-1333 memory available soon (but horribly expensive - launch price something like $500 per pair of 1GB modules), which is 10.5GB/sec/channel and goes nicely with the 1333MHz bus which is being introduced soon.

http://www.anandtech.com/memory/showdoc.aspx?i=2989 has a review.

What chip-set are you using with the DDR2 8500 memory?
fivemack is offline   Reply With Quote
Old 2007-05-19, 15:36   #27
S485122
 
S485122's Avatar
 
"Jacob"
Sep 2006
Brussels, Belgium

24·109 Posts
Default

Quote:
Originally Posted by fivemack View Post
What chip-set are you using with the DDR2 8500 memory?
P965 in a Asus P5B-E Plus board.
S485122 is offline   Reply With Quote
Old 2007-05-22, 20:35   #28
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101Γ—103 Posts

3·47·71 Posts
Default

Quote:
Originally Posted by jasong View Post
If Prime95 processes a lot of data(though it becomes unimportant a few hundredths of a second later), and therefore has a bottleneck, would it not be better to seek out a worthy low bandwidth project and strategically distribute that project, along with Prime95, among the cores?
That is why some folks run T-F and L-L concurrently. The T-F can generally stay in cache, while L-L needs main mem.
Uncwilly is online now   Reply With Quote
Old 2007-05-28, 07:37   #29
drew
 
drew's Avatar
 
Jun 2005

2×191 Posts
Default

Quote:
Originally Posted by R.D. Silverman View Post
Time is money. An extra $260 is a pittance when you need the capability NOW.
Some of us have larger price elasticities.
That would be inelasticity.
drew is offline   Reply With Quote
Old 2007-05-28, 08:23   #30
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

22×3×11×83 Posts
Default

Quote:
Originally Posted by drew View Post
That would be inelasticity.
I'm not so sure.

If I want something now, rather than later, the range of prices I'm prepared to pay now is large. If my cost constraints are less flexible (i.e more inelastic) I have to wait until the product is on sale at a price I'm prepared to pay.


Paul
xilman is offline   Reply With Quote
Old 2007-05-28, 08:59   #31
drew
 
drew's Avatar
 
Jun 2005

2·191 Posts
Default

Quote:
Originally Posted by xilman View Post
I'm not so sure.

If I want something now, rather than later, the range of prices I'm prepared to pay now is large. If my cost constraints are less flexible (i.e more inelastic) I have to wait until the product is on sale at a price I'm prepared to pay.


Paul
Price elasticity is a well-defined economic concept. An elastic demand curve means the quantity demanded depends heavily on the price of the product.

Silverman made it clear that he wants a new processor, regardless of the price. This inelasticity means he's not easily influenced by the price of the product.

Last fiddled with by drew on 2007-05-28 at 09:04
drew is offline   Reply With Quote
Old 2007-05-28, 16:49   #32
Richard Cameron
 
Richard Cameron's Avatar
 
Mar 2005

2·5·17 Posts
Default intertemporal consumer choice

Quote:
Originally Posted by R.D. Silverman View Post
Time is money. An extra $260 is a pittance when you need the capability NOW.
this reads to me as a statement of Mr Silverman's time_preference It is his Discount_rate that is very high.
Richard Cameron is offline   Reply With Quote
Old 2007-05-28, 19:48   #33
drew
 
drew's Avatar
 
Jun 2005

2·191 Posts
Default

Quote:
Originally Posted by Richard Cameron View Post
this reads to me as a statement of Mr Silverman's time_preference It is his Discount_rate that is very high.
True, but that's beside the point. The original comment was regarding price elasticity.
drew is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Dual Core to Quad Core Upgrade Rodrigo Hardware 6 2010-11-29 18:48
exclude single core from quad core cpu for gimps jippie Information & Answers 7 2009-12-14 22:04
Quad Core Questions... TomYosho Information & Answers 2 2009-09-14 13:01
Quad Core and P95 sgrupp Hardware 54 2008-01-25 22:01
Optimising work for Intel Core 2 Duo or Quad Core S485122 Software 0 2007-05-13 09:15

All times are UTC. The time now is 21:58.


Fri Oct 22 21:58:16 UTC 2021 up 91 days, 16:27, 0 users, load averages: 4.63, 4.77, 3.77

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

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.
A copy of the license is included in the FAQ.