Is Dual Channel BS??

Foosanator · Jan 29, 2004, 07:56 PM

Alright, I have an Asus A7V8X rev 2.0 non-deluxe board with a xp 3200. When I first put the 3200 in I was running 2x256 sticks of Geil Golden Dragon pc 3200. When I tried to run them in DC they would create a ton of errors in MEMTEST. I sent them back to Geil under warrenty. In the meantime I had a stick of Cheapy Ocz pc3200 512 stick laying around so I tossed it in. I then played around with overclocking the ocz stick/cpu. I was able to hit 11x227 stable (around 2.5 gig). I got my Geil back yesterday and put that in. I was immediatly able to run 11x225 in Dual Channel. (FYI Geil is cas 2-3-3-6 and Ocz is 3-4-4- I went ahead and benched the geil in Dual Channel and got a lower score in UT2K3 bench than I did with the cheapo Ocz. The Ocz Scored considerably better at the same overclock (11x225) in single channel Configuration? So I am a little confused here.... Also, There were no FSB Fluxuations during either of the tests. Any input here before I give up on this Geil Ram???

Thanks,

~~craig588~~ · Jan 29, 2004, 08:45 PM

Dual channel on AXP boards doesnt do anything, they are already at their FSB limit. (ram running at the same speed as the FSB, and no extra BW tweaks in the FSB)

On Intel it will be signifigantly better because usually you will want to run you ram with a devider so already the ram is at 4/5 of the speed of the FSB, intel also has a quad pumped FSB which runs at a theoretical 4xFSB speed, so with an Intel system you have lots of wasted bandwidth, but using dual channel sticks doubles the availible bandwidth so it makes up for it for the most part.

With an AXP mobo none of the availible bandwidth is wasted in single channel mode. (Maybe 100MBs, but nowhere near as high as intels 3000MBs)

Matth · Jan 30, 2004, 03:40 PM

My recollection of Nforce 2 dual/single channel tests is that there was VERY LITTLE gain, since the AMD Socket A processors can only take data at the rate of one normal DDR channel anyway.

It definitely gives a significant gain if using onboard graphics, but Geforce 4 MX (low model) class graphics are hardly something to get excited about.

The other case, is when the RAM does not match the FSB, and dual channel tends to alleviate the performance drain.

BWX · Jan 30, 2004, 04:05 PM

My experience with dual channel is that if you have to run a low FSB speed because the memory isn't that fast you get about a 5% speed increase in bench tests and a few hundred points in 3dmark01.
I'm getting some new matched sticks of XMS in a few days, so I'll be able to test between dual channel and single channel with a 200mhz fsb- I'm thinking it won't give that great of a boost, but enough to make it worth running dual channel. 5% is still 5%.

Foosanator · Jan 30, 2004, 10:32 PM

BWX I am anxious to see what your reslults yeild as I am really dissapointed with my own.

BWX · Jan 31, 2004, 04:53 AM

Quote:

Originally posted by Foosanator
BWX I am anxious to see what your reslults yeild as I am really dissapointed with my own.

Teusday I should have it- Monday if I'm lucky.

Jan 29, 2004, 07:56 PM	#1
Foosanator DriverHeaven Junior Member Join Date: Dec 2003 Posts: 41 Rep Power: 0	Is Dual Channel BS?? Alright, I have an Asus A7V8X rev 2.0 non-deluxe board with a xp 3200. When I first put the 3200 in I was running 2x256 sticks of Geil Golden Dragon pc 3200. When I tried to run them in DC they would create a ton of errors in MEMTEST. I sent them back to Geil under warrenty. In the meantime I had a stick of Cheapy Ocz pc3200 512 stick laying around so I tossed it in. I then played around with overclocking the ocz stick/cpu. I was able to hit 11x227 stable (around 2.5 gig). I got my Geil back yesterday and put that in. I was immediatly able to run 11x225 in Dual Channel. (FYI Geil is cas 2-3-3-6 and Ocz is 3-4-4- I went ahead and benched the geil in Dual Channel and got a lower score in UT2K3 bench than I did with the cheapo Ocz. The Ocz Scored considerably better at the same overclock (11x225) in single channel Configuration? So I am a little confused here.... Also, There were no FSB Fluxuations during either of the tests. Any input here before I give up on this Geil Ram??? Thanks,

Jan 30, 2004, 04:05 PM	#4
BWX unplugged Join Date: Nov 2002 Location: USNY Posts: 19,669 Rep Power: 110 System Specs	My experience with dual channel is that if you have to run a low FSB speed because the memory isn't that fast you get about a 5% speed increase in bench tests and a few hundred points in 3dmark01. I'm getting some new matched sticks of XMS in a few days, so I'll be able to test between dual channel and single channel with a 200mhz fsb- I'm thinking it won't give that great of a boost, but enough to make it worth running dual channel. 5% is still 5%. __________________

Jan 29, 2004, 08:45 PM	#2
~~craig588~~ Banned Join Date: Nov 2002 Location: In clothing Posts: 3,510 Rep Power: 0	Dual channel on AXP boards doesnt do anything, they are already at their FSB limit. (ram running at the same speed as the FSB, and no extra BW tweaks in the FSB) On Intel it will be signifigantly better because usually you will want to run you ram with a devider so already the ram is at 4/5 of the speed of the FSB, intel also has a quad pumped FSB which runs at a theoretical 4xFSB speed, so with an Intel system you have lots of wasted bandwidth, but using dual channel sticks doubles the availible bandwidth so it makes up for it for the most part. With an AXP mobo none of the availible bandwidth is wasted in single channel mode. (Maybe 100MBs, but nowhere near as high as intels 3000MBs)

Jan 30, 2004, 03:40 PM	#3
Matth Flash Banner Hater Join Date: Jun 2002 Location: UK Posts: 3,426 Rep Power: 93 System Specs	My recollection of Nforce 2 dual/single channel tests is that there was VERY LITTLE gain, since the AMD Socket A processors can only take data at the rate of one normal DDR channel anyway. It definitely gives a significant gain if using onboard graphics, but Geforce 4 MX (low model) class graphics are hardly something to get excited about. The other case, is when the RAM does not match the FSB, and dual channel tends to alleviate the performance drain.

Jan 30, 2004, 10:32 PM	Thread Starter #5
Foosanator DriverHeaven Junior Member Join Date: Dec 2003 Posts: 41 Rep Power: 0	BWX I am anxious to see what your reslults yeild as I am really dissapointed with my own.