AMD Socket A Dual Channel

[Heffty]

Lets say a person has two mbo's of the same make (ie Asus, Abit, DFI, Biostar...etc), same line (ie Biostar M7NCD and M7NCD Pro), and same chipset (lets say nforce2 Ultra 400) but one has dual channel technology and the other doesn't. Let's also say this guy has 2 1G sticks of Samsung PC3200 DDR RAM that he could use in either board.

What kind of performance increase if any would there be if he used the dual channel board instead of the single channel board? Are there any articles that have been written about tests done to compare the two with conclusive results?

I'd also like to know if there have been any tests done to see if there are any differences between the nforce2 ultra 400 and non-ultra 400 boards.

I am not looking for speculative answers and don't want them to be opinions. I'm looking for solid research that has already been done through tests like whats done in anandtech or toms or something OR I'm looking for someone that has done tests and can back it up with numbers.

Thanks.

 

[ribbon13]

why not dig through toms archive?
http://www17.tomshardware.com/search/se...tml?category=motherboard&words=nforce2

 

[eXx08]

Are you too lazy to do the research yourself or something?

 

[bob4432]

Originally posted by: Heffty
Lets say a person has two mbo's of the same make (ie Asus, Abit, DFI, Biostar...etc), same line (ie Biostar M7NCD and M7NCD Pro), and same chipset (lets say nforce2 Ultra 400) but one has dual channel technology and the other doesn't. Let's also say this guy has 2 1G sticks of Samsung PC3200 DDR RAM that he could use in either board.

What kind of performance increase if any would there be if he used the dual channel board instead of the single channel board? Are there any articles that have been written about tests done to compare the two with conclusive results?

I'd also like to know if there have been any tests done to see if there are any differences between the nforce2 ultra 400 and non-ultra 400 boards.

I am not looking for speculative answers and don't want them to be honest. I'm looking for solid research that has already been done through tests like whats done in anandtech or toms or something OR I'm looking for someone that has done tests and can back it up with numbers.

Thanks.

you don't want honest answers?

 

[Heffty]

First of all, you guys know nothing and that's why you didn't help me. You're probably tech school drop outs or something and the only way you can find some kind of happiness in your lame lives is to post useless comments in a thread. Props to you! Keep em coming, like I care.

Now to the guy who actually tried to help...I've already done some digging myself. I can't seem to find any data that is specific to the issue I want resolved. I do not know much about computers and haven't spent that much time with them. I thought maybe someone in here would know exactly where to look by chance...or would have a lead or two.

 

[Snakexor]

your not gonna get anyhelp now i think


and i didnt pass fortune teller school

 

[ribbon13]

Well, I did used to have an article on that EXACT issue, but that was a few years ago. It gets difficult to find info like that on obsolete hardware. One thing though. The dual-channel compatibility is part of the chipset for socketA systems. So the only way for similar boards to differ only in dual channel would be for one to have the performance chipset and the other to have the value version (ie one nF2 and one nF2 Ultra).

This might be useful
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=1731&p=2

 

[Googer]

Originally posted by: thekillerjks
Are you too lazy to do the research yourself or something?

Sir I do believe that you are correct. :beer:

 

[Googer]

Go to your library and find the old issue of Maximum PC when they first tested this new nForce twin bank chipset. It killed all others in the benchmarks. But I would not spend too much money on it because your socket A is hardly worth spending much money on.

For not much more than you might pay for this new motherboard and ram you can almost get a new and faster PC.

EDIT:

I also doubt that a motherboard of that vintage will be able to address 2Gigabytes of RAM in it's bios.

 

[winr]

09/14/2005 09:35 AM
User is offline View Users Profile Send Private Message Print this message

Author Icon
Spacecomber
Senior Member

Posts: 214
Joined: 04/21/2000


Battlefield 2 seems to be the first game that requires more than 1GB of memory; at least, this seems to be true if you are running a 256mb video card. I supect that it is only the first game, and by no means the last, that is this resource hungry. BF2 is what prompted me to move beyond 1GB for my system.

I think that in theory you could try to find a similar module to what you have now from another manufacturer and it might work fine in dual channel mode, but I really don't have any experience with this. Obviously, the memory makers have made much of the modules needing to be matched, as they sell their memory in kits of matched pairs, and I can't say as I've heard much of people describing how they successfully got DIMMs from different manufacturers to work well together in dual channel mode.

So, it would come down to trial and error on your part, which probably isn't very practical.

Again, the difference between dual channel and single channel in these systems is not that much. I'll list some results comparing my two mushkin modules in dual channel versus one of these plus a 1GB Patriot module running in single channel mode. This is with an Athlon XP running at 2.3GHz on a 210MHz FSB and a memory bus also at 210MHz. Video card is a 6800GT. Timings are 2,3,2,8 for both memory setups. (Actually, 2,3,2,11 would be better for this system, but I used tRAS=8 for some reason, perhaps because it was closer to the SPD timing.) I'll list the 2x512 results, then the 512+1GB results, so you can easily compare them (2x512mb / 512mb+1GB).

Sandra Memory Integer: 3226/3155
Sandra Memory Floating: 3040/2884
Sandra Memory&Cahce, Combined Index: 4678/4092
ScienceMark Memory: 3070/2991
Prime95 2048K FFT Benchmark (lower time is better): 134.86ms/141.693
Aquamark Overall: 56,330/55,467
Aquamark CPU: 8345/8028
Aquamark Graphics Card: 8502/8472

While their is a consistent advantage to running in dual channel mode on the benchmarks that I tested this with, the difference is only a few percentage points. This is why I'm content to have increased my memory beyond 1GB, which smoothed out BF2 game play, even if it meant going to single channel. I'm in no hurry to regain the the performance that running two matched 1GB modules would offer, since I really don't think I'll notice the 1fps increase in my average frame rate as I play.

Space
Reply Quote Top Bottom

 

[Heffty]

thank you very much for your responses...especially that last one.

 

[bob4432]

you can still use less than 1GB of total ram usage (if your machine idles at 240MB or less) and have good graphics with bf2 - http://bf2hardware.sytes.net:6583/

 

[Heffty]

I'm in no hurry to regain the the performance that running two matched 1GB modules would offer, since I really don't think I'll notice the 1fps increase in my average frame rate as I play.

It seems as though what you tested here was if 2 512's in dual channel mode would run equally, better than, or worse than 1.5G of memory in a non-dual channel setup. Your results from this test seemed to convince you that there was no point in going up to 2 gigs for the minimal performance gain with dual channel. What I am most interested in is a comparison between running the two 512's in the single channel setup and two 512's in the dual channel setup...straight up head to head. Or better yet...2 1G sticks in single compared to 2 1G sticks in dual.

I think your rationalizing is probably correct; I'd just like to know what the head to head numbers would be. If I had a dual channel board I'd test it myself...just a point of interest for those morons who think I am just lazy. I'll tell you whats lazy is calling someone out from a keyboard rather than doing it in person.

 

[bob4432]

Originally posted by: Heffty
I'm in no hurry to regain the the performance that running two matched 1GB modules would offer, since I really don't think I'll notice the 1fps increase in my average frame rate as I play.

It seems as though what you tested here was if 2 512's in dual channel mode would run equally, better than, or worse than 1.5G of memory in a non-dual channel setup. Your results from this test seemed to convince you that there was no point in going up to 2 gigs for the minimal performance gain with dual channel. What I am most interested in is a comparison between running the two 512's in the single channel setup and two 512's in the dual channel setup...straight up head to head. Or better yet...2 1G sticks in single compared to 2 1G sticks in dual.

I think your rationalizing is probably correct; I'd just like to know what the head to head numbers would be. If I had a dual channel board I'd test it myself...just a point of interest for those morons who think I am just lazy. I'll tell you whats lazy is calling someone out from a keyboard rather than doing it in person.

i can't remember exact #'s but it is was pretty low, like 5% when using ram bandwidth heavy apps, nothing that requires the actual use of dual channel, other than those wanting the higher performance possible. the p4s were the the ones that actually needed the bandwidth of dual channel due to memory controller and cpu architecture design.

 

[Googer]

Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

 

[bob4432]

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

 

[Boogak]

nForce2: single-channel mode vs. dual-channel mode

 

[Googer]

Originally posted by: bob4432

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

I do know tualatin chips were out perorming newer chips. That is why I still have it.
I Have not had time to play it lately, but I was running it at 1280x1024 and 1152x864.

 

[bob4432]

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

I do know tualatin chips were out perorming newer chips. That is why I still have it.
I Have not had time to play it lately, but I was running it at 1280x1024 and 1152x864.

impressive to say the least....do you hit the pagefile at all?

 

[Googer]

Originally posted by: bob4432

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

I do know tualatin chips were out perorming newer chips. That is why I still have it.
I Have not had time to play it lately, but I was running it at 1280x1024 and 1152x864.

impressive to say the least....do you hit the pagefile at all?

Not really much, it happend once or twice but I was running folding at home (uses 100+ MB of RAM) while I was playing. Battlefield 2 is entirely GPU dependant, it relies very little on System RAM and CPU to get it's jobs done. Some benchmarks from a while ago comfirmed this- A P4 1.6GHz generated the same number of frames per second as an AMD FX55. Also worth noting is there was no differance in the FPS when moving from 512MB to 1024MB of RAM. It may improve the loading time of MAPS but it does not give any extra FPS.

Also I have done a Few optimizations to Windows 2000 to reduce it's memory footprint.

 

[bob4432]

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

I do know tualatin chips were out perorming newer chips. That is why I still have it.
I Have not had time to play it lately, but I was running it at 1280x1024 and 1152x864.

impressive to say the least....do you hit the pagefile at all?

Not really much, it happend once or twice but I was running folding at home (uses 100+ MB of RAM) while I was playing. Battlefield 2 is entirely GPU dependant, it relies very little on System RAM and CPU to get it's jobs done. Some benchmarks from a while ago comfirmed this- A P4 1.6GHz generated the same number of frames per second as an AMD FX55. Also worth noting is there was no differance in the FPS when moving from 512MB to 1024MB of RAM. It may improve the loading time of MAPS but it does not give any extra FPS.

Also I have done a Few optimizations to Windows 2000 to reduce it's memory footprint.

i remember seeing some benches that were close to what you are saying, but still that gpu is a little on the low side for bf2. not knocking the card because i had a 9800pro that worked very well for me. just out of curiosity, what does your machine idle at ram wise? there is no doubt that your cpu owns a 1.6 P4, it just seems that you are getting very good fps from older hardware and the people i know that are running 9800pros regardless of cpu are playing at 800x600 or a max of 1024x768 with no AA and every so often when we are on teamspeak they will complain about serious studder that appears to be limited to their machine as others on the same server will have no issues, but have higher end gpus.

i had put this site to give others information i had concluded as i was messing around with different setting to keep my machine that idles at ~240MB under 1GB when gaming, as i don't want to buy another 1GB just for a game. as you can see i was able to get it down to 940MB so i was more than happy with that and still be over 60fps according to fraps with excellent image quality even in the most intense smoke, firefight, dusty scenes.

 

[Googer]

Originally posted by: bob4432

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer

Originally posted by: bob4432

Originally posted by: Googer
Battlefield 2 plays fine on My computer-
Pentium III Server version 1.4GHz Tualatin with full speed (vs half speed) 512k cache
One Stick of CAS2 PC133- 512 Megabytes
ATI 9700 PRO.
Hitachi 7k250

you have to understand that those Tualatin S chips were very underated in and although their clock speed says 1.4GHz, they were killing the early P4s @ 1.8GHz. i don't know where you got that chip or what you paid for it, but you can get a 3000Venice cheaper than one of those these days. i don't think that many were made compared to the other cpus and the fact that they were dual capable makes them even more valueable. i would love to get a pair of these and a good dual skt370 m/b for a nice server with tons of power.

also you have to put in what resolution and settings you are playing it at, we could all play at the lowest setting with all the setting at low and no AA.

and if you have choose between 800x600 2AA vs 1024x768 No AA, imo 800x600 2AA looked much better.

last have you used fraps to see what your average fps is? imo, 60+fps average is required and with my rig that is what i get when playing at the specs i listed on the page i referrenced earlier.

I do know tualatin chips were out perorming newer chips. That is why I still have it.
I Have not had time to play it lately, but I was running it at 1280x1024 and 1152x864.

impressive to say the least....do you hit the pagefile at all?

Not really much, it happend once or twice but I was running folding at home (uses 100+ MB of RAM) while I was playing. Battlefield 2 is entirely GPU dependant, it relies very little on System RAM and CPU to get it's jobs done. Some benchmarks from a while ago comfirmed this- A P4 1.6GHz generated the same number of frames per second as an AMD FX55. Also worth noting is there was no differance in the FPS when moving from 512MB to 1024MB of RAM. It may improve the loading time of MAPS but it does not give any extra FPS.

Also I have done a Few optimizations to Windows 2000 to reduce it's memory footprint.

i remember seeing some benches that were close to what you are saying, but still that gpu is a little on the low side for bf2. not knocking the card because i had a 9800pro that worked very well for me. just out of curiosity, what does your machine idle at ram wise? there is no doubt that your cpu owns a 1.6 P4, it just seems that you are getting very good fps from older hardware and the people i know that are running 9800pros regardless of cpu are playing at 800x600 or a max of 1024x768 with no AA and every so often when we are on teamspeak they will complain about serious studder that appears to be limited to their machine as others on the same server will have no issues, but have higher end gpus.

i had put this site to give others information i had concluded as i was messing around with different setting to keep my machine that idles at ~240MB under 1GB when gaming, as i don't want to buy another 1GB just for a game. as you can see i was able to get it down to 940MB so i was more than happy with that and still be over 60fps according to fraps with excellent image quality even in the most intense smoke, firefight, dusty scenes.

It does not bother me what you say about my PC, love it or hate it; your comments do not bother me. At one time all of our PC's were considered king of the hill.

The 9700 and it's later revision the 9800 are very similar in terms of performance. The 9800 is only a few FPS faster in Quake than the 9700 but not by very much that you would likly notice while playing.

With some of the new software I am running like my firewall, AV scanner, and some of the newer hardware related programs like setpoint. I was able to get it down to 100-130MB USED, but now I am around 180-200 USED because things My Logitech 518 require certain programs to run in order to take advantage of the functinality of that piece of equipment. And MS also requres certain memory eating services remain running at all times that I did not have to have running a few years ago.

I have seen frame rates jump as high as 92-93FPS when looking down a mountain, but that is not a sustainable frame rate. I tend to stay around a playable 30FPS as measured by Fraps with all the in game video settings set as low as they can go execpt for the View Distance Scale. Battlefield 2 is very good a managing limited resources and that is reflected in the low sysem requirements on the games box. I have also instructed the ATI driver to keep the eye candy at a minimal level using RADLink.

 

[Spike]

Originally posted by: Googer
Go to your library and find the old issue of Maximum PC when they first tested this new nForce twin bank chipset. It killed all others in the benchmarks. But I would not spend too much money on it because your socket A is hardly worth spending much money on.

For not much more than you might pay for this new motherboard and ram you can almost get a new and faster PC.

EDIT:

I also doubt that a motherboard of that vintage will be able to address 2Gigabytes of RAM in it's bios.

I am currently running 2GB (2z1gb) of OCZ plat 3200 @ 2.5-2-3-5 in my Abit NF7-S 2.0 with my socket A Barton. It would not run stable at cl 2 but upping it to 2.5 made it rock solid.

-spike

 

[Sensai]

NF2 Ultras are capable of addressing 3 GBs with 3 dimm slots at 1 GB each, though with all 3 populated, u can only run at ddr-333 speeds. The regular NF2 boards can technically run at ddr-400 also, but by default of 333 in the BIOS. The Ultra is more of a marketing hype to than a true feature as regular boards already has support for 400 or just needs a BIOS update. Also, expect up to a max of 8-10% performance with dual channel.