Page 1 of 1

CPU-z is claiming I am running dual

Posted: Wed Jan 24, 2007 7:38 pm
by bigbear2007
on 1x512M & 1x1G

Is it possible ?

Re: CPU-z is claiming I am running dual

Posted: Wed Jan 24, 2007 8:55 pm
by RonS
bigbear2007 wrote:Is it possible ?
Nope.

Posted: Thu Jan 25, 2007 1:47 am
by dr_st
Actually it is.

Posted: Thu Jan 25, 2007 1:53 am
by rkawakami
ref: http://en.wikipedia.org/wiki/Dual-channel (which says "yes" it's possible)
http://www.intel.com/support/motherboar ... 011965.htm (which says "no"; note: this says desktops)

Posted: Thu Jan 25, 2007 3:43 am
by laundromatt
so which is it?

Posted: Thu Jan 25, 2007 9:29 am
by Zeus
you are running dual; thinkpads can do it, somehow

Posted: Thu Jan 25, 2007 12:07 pm
by brentpresley
Zeus wrote:you are running dual; thinkpads can do it, somehow
I would like to see some definitive proof for this.

FWIW, CPU-Z normally just looks to see if both DIMM slots are occupied in a dual capable system, and doesn't actually query the chipset to see if it is truly running in dual mode.

Download memtest86, burn it to a CD and boot from it. That program SHOULD be able to ID dual channel properly. (you want version 1.70)

Posted: Thu Jan 25, 2007 1:10 pm
by RonS
I just downloaded and read some of Intel's datasheet for the 945PM, which includes the memory controller.

Three modes are supported:

1. Single channel.
2. Dual-channel symmetric. This is the fastest mode. Both memory modules must be the same size.
3. Dual-channel asymmetric. Performance is just slightly better than single-channel.

http://download.intel.com/design/mobile ... 921903.pdf see page 337

So, I stand corrected. Dual-channel is supported with mismatching memory modules, although in a mode that provides almost no performance boost over single-channel

Posted: Thu Jan 25, 2007 1:35 pm
by brentpresley
RonS wrote:Dual-channel is supported with mismatching memory modules, although in a mode that provides almost no performance boost over single-channel
So why even call it Dual-Channel? It is effectively Single Channel w/ 2 dimms, right?

Posted: Thu Jan 25, 2007 1:59 pm
by RonS
Based on my scant reading, dual-channel asymmetric mode gets you greater than single-channel performance only when software reads memory that alternates between the two modules, which Intel says is "unlikely."

Posted: Thu Jan 25, 2007 5:13 pm
by dr_st
RonS wrote:So, I stand corrected. Dual-channel is supported with mismatching memory modules, although in a mode that provides almost no performance boost over single-channel
That's OK, any dual channel offers only marginal performance boost anyway.

I remember these discussions from about a year, a year and a half ago, when everyone was surprised to see that Dual Channel can be run with non-identical modules, because we were so used to desktop chipsets, where there is no such option.