Dual channel memory (cas3.0) VS. fast (cas2.0) memory

Hi,
what is best to have:
1) 2x256Mb cas3.0 with dual channel memory
2) or a fast 512Mb cas2.0

My mobo is an MSI Delta-L with nForce2 Ultra 400.

Best regards.

"badr" wrote in message
m...
Hi,
what is best to have:
1) 2x256Mb cas3.0 with dual channel memory
2) or a fast 512Mb cas2.0

My mobo is an MSI Delta-L with nForce2 Ultra 400.

Best regards.

Do a Google for "Memory Timings Analysis" Harry Lam 05-16-03. Bottom line:
"There was no significant gain in memory bandwidth from adjisting CAS
latencies". Having said that I get just a bit over 3% improvement on a
Shuttle AN35n Ultra using 2x512GB Kingstom PC3500 in Dual Channel over
Single. Go Dual Channel, but don't expect too much ;-) Post your results!

FRH

"Frank Hagan" wrote in message hlink.net...
"badr" wrote in message
m...
Hi,
what is best to have:
1) 2x256Mb cas3.0 with dual channel memory
2) or a fast 512Mb cas2.0

My mobo is an MSI Delta-L with nForce2 Ultra 400.

Best regards.

Do a Google for "Memory Timings Analysis" Harry Lam 05-16-03. Bottom line:
"There was no significant gain in memory bandwidth from adjisting CAS
latencies". Having said that I get just a bit over 3% improvement on a
Shuttle AN35n Ultra using 2x512GB Kingstom PC3500 in Dual Channel over
Single. Go Dual Channel, but don't expect too much ;-) Post your results!

FRH

Tanks Frank,
this what I need to know. The link for this interesting review by
Harry Lam is http://www.techwarelabs.com/reviews/...gs/index.shtml.

Thanks again,
and thanks to Harry.

badr wrote:
Hi,
what is best to have:
1) 2x256Mb cas3.0 with dual channel memory
2) or a fast 512Mb cas2.0

My mobo is an MSI Delta-L with nForce2 Ultra 400.

There is no such thing as "Dual Channel Memory".

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

badr wrote:
Hi,
what is best to have:
1) 2x256Mb cas3.0 with dual channel memory
2) or a fast 512Mb cas2.0

My mobo is an MSI Delta-L with nForce2 Ultra 400.

There is no such thing as "Dual Channel Memory".

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

" "

There is omly a small increase in performance between 2 sticks in dual channel
mode a a single stick in std mode.

There is no such thing as "Dual Channel Memory".

Ben



Yes, but oddly enough, there is memory that won't do dual
channel. I have a pair of Elixir DDR400 256Meg sticks that
run fine on a NF7-S 2.0 in non-dual channel but totally fail
(memtest86) running dual channel. I recently found a couple
of vendor sites that explicitly state that Elixir is not
suitable for dual channel. I don't understand why, I would
think that with the controller ping-ponging back and forth
between the two sticks, the load on each stick would be less.

Bill Sudbrink wrote:
There is no such thing as "Dual Channel Memory".

Ben



Yes, but oddly enough, there is memory that won't do dual
channel. I have a pair of Elixir DDR400 256Meg sticks that
run fine on a NF7-S 2.0 in non-dual channel but totally fail
(memtest86) running dual channel. I recently found a couple
of vendor sites that explicitly state that Elixir is not
suitable for dual channel. I don't understand why, I would
think that with the controller ping-ponging back and forth
between the two sticks, the load on each stick would be less.

I'd like to have some technical reasoning for it... Dual Channel is exactly
that, two separate channels, so I don't see how RAM can be unsuitable for
Dual Channel, although you're certainly not the first person to experience
that type of issue.

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

There is no such thing as "Dual Channel Memory".

Yes, but oddly enough, there is memory that won't do dual
channel. I have a pair of Elixir DDR400 256Meg sticks that
run fine on a NF7-S 2.0 in non-dual channel but totally fail
(memtest86) running dual channel. I recently found a couple

I'd like to have some technical reasoning for it... Dual Channel is exactly
that, two separate channels, so I don't see how RAM can be unsuitable for
Dual Channel, although you're certainly not the first person to experience
that type of issue.

Ben



Someone just privately emailed this hypothesis: Dual channel works
by taking advantage of the recovery time between channels/modules.
To do this, there must be seperate signal lines to each channel, but
maybe not all of the signal lines are seperate. Maybe "good" RAM
is quiet on these shared signal lines while it is recovering, but for
some reason, "bad" RAM makes noise on those lines. For instance, each
channel could have its own CAS line but they could both share the
address lines. "Good" RAM would be quiet on the address lines while
recovering while "bad" RAM would make noise or somehow interfere with
the address lines while recovering. I think that's the idea. When no
interleaving is going on (single channel), no problem. Nobody cares
about the noise. In dual channel, the noise screws up the communication
to the RAM on the other channel.

Bill Sudbrink wrote:
There is no such thing as "Dual Channel Memory".

Yes, but oddly enough, there is memory that won't do dual
channel. I have a pair of Elixir DDR400 256Meg sticks that
run fine on a NF7-S 2.0 in non-dual channel but totally fail
(memtest86) running dual channel. I recently found a couple

I'd like to have some technical reasoning for it... Dual Channel is
exactly that, two separate channels, so I don't see how RAM can be
unsuitable for Dual Channel, although you're certainly not the first
person to experience that type of issue.

Ben



Someone just privately emailed this hypothesis: Dual channel works
by taking advantage of the recovery time between channels/modules.
To do this, there must be seperate signal lines to each channel, but
maybe not all of the signal lines are seperate. Maybe "good" RAM
is quiet on these shared signal lines while it is recovering, but for
some reason, "bad" RAM makes noise on those lines. For instance, each
channel could have its own CAS line but they could both share the
address lines. "Good" RAM would be quiet on the address lines while
recovering while "bad" RAM would make noise or somehow interfere with
the address lines while recovering. I think that's the idea. When no
interleaving is going on (single channel), no problem. Nobody cares
about the noise. In dual channel, the noise screws up the communication
to the RAM on the other channel.

It's a good hypothesis, but on the nForce, each SLOT (often banks have their
own address) has it's own address bus in order to reduce these exact
problems.

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...