Dual Channel

[solo]

I have a pc3500 geil ultra platinum keeping my CPU busy at 280fsb stable. If i get another stick, but it doesn't perform like the first stick, will i be able to use it in dual channel and get optimum performance?

cheers

[cruc1fy]

Originally posted by solo
I 280fsb stable.


Dual Channel memory configurations don't have that much of an effect on the available 32-bit procs (because a lot of the gains are lost in the lengthy front side bus).

[[OC]This]

Here comes another Dual Channel "basher".

I'm getting fed up with these people.


Those people who don't want DC, run your CPU at default.. set your memory timing to default... maybe even Underclock your system. You don't need to overclock your system.


Get my drift.. :cool:

[stan03]

yes you can use dual channel, but you will prolly have to lower the setting on the stick you have now so that they sticks match... but still dual channel benefits should probably make up for it.

[cruc1fy]

Originally posted by [OC]This
Here comes another Dual Channel "basher".

I'm getting fed up with these people.


Who's a "dual channel basher?"

I run a dual channel setup in one of my systems.

My acknowledgement of the fact that it has only a minute effect on system performance does not make me a "basher." I have no doubt that it will become very useful with the 64-bit chipset features (like AMD's hypertransport).

[solo]

so dual channel won't allow me to raise my fsb higher than it already is? because both sticks will be limited by the slower one?

cheers

[method().man]

Originally posted by solo
so dual channel won't allow me to raise my fsb higher than it already is? because both sticks will be limited by the slower one?

cheers

That is correct.

Dual channel mostly helps people who run very high FSBs. In such cases, you can ease off on the clock speed of your memory since dual channel can make up and exceed whatever bandwidth loses you take from your memory not being up to snuff.

[stan03]

Originally posted by solo
so dual channel won't allow me to raise my fsb higher than it already is? because both sticks will be limited by the slower one?

cheers

thats what i was trying to say, sorry for not being clear.

[Flewdefur]

Ok, i have a question which is semi related; i didnt want to start a new thread for it...Does there have to be something special with your ram to run Dual Channel or is it just a motherboard setting you can use when you have two sticks (and follow motherboard instructions blah blah)?

[method().man]

Originally posted by Flewdefur
Ok, i have a question which is semi related; i didnt want to start a new thread for it...Does there have to be something special with your ram to run Dual Channel or is it just a motherboard setting you can use when you have two sticks (and follow motherboard instructions blah blah)?

As long as the chips are relatively identical (ie, same size, density, number of chips), dual channel should work fine.

[Flewdefur]

Thanks for the information! I think i'll get one of the 2x256 DC tested sets anyway.

[Mico]

Don't pay extra money for Dual-Channel RAM. As long as the chips are the same, they will work. I've seen some sites that cost about $20-30 more for the exact same RAM because it was Dual-Channel tested.

[Mustanley]

Originally posted by cruc1fy


Who's a "dual channel basher?"

I run a dual channel setup in one of my systems.

My acknowledgement of the fact that it has only a minute effect on system performance does not make me a "basher." I have no doubt that it will become very useful with the 64-bit chipset features (like AMD's hypertransport).

On AMD systems, you are correct, dual channel does not add much performance and can even hinder your OC limit. But things are different with P4C chips. They suck up all the bandwidth you can throw at them, since the processor is effectively running at 800MHz, or higher if overclocked. So it depends on what platform solo is using as to whether or not he will see any significant performance gains.

[solo]

ROAR!

so dual channel will technically allow me to get a higher overclock? correct me if im wrong plz

cheers

[Mustanley]

Dual channel will limit the degree to which you can overclock the FSB more than running just single channel memory. However, since the benefit to raising the FSB is basically increasing memory bandwidth, you have to weigh the advantage of dual channel vs. higher FSB. Since dual channel almost effectively doubles memory bandwidth, it beats out a higher overclocked FSB with a P4. An Athlon XP Barton can't really make any use of the extra bandwidth dual channel provides because it runs on 166 or 200 MHz bus (double pumped to equal 333 MHz or 400 Mhz). Running PC3200 memory at 166MHZ DDR (333 MHz effective) or 200 MHz DDR (400 MHz) will effectively fill the bus going to the cpu. Running dual channel may double the memory bandwidth (like running 800 MHz) going to the Barton CPU, but it can only handle 400 MHz worth, so that extra bandwidth is wasted.
The P4 is running 200 MHz quad pumped (800 MHz effective) so running 400 MHz DDR memory dual channel can supply the cpu continually without wasting memory bandwidth.

These numbers are more theoretical than actual. There is overhead to account for within the memory controllers, so the actual bandwidth may only be 70%-80% of the numbers listed.

This was somewhat hard for me to articulate, so I hope it makes sense to you. You never did say whether you are running an AMD or Intel cpu.

[Voodoo_fx]

NO, it will not get you a higher OC, not that NO is like yelling, but thats the answer, what there all trying to say is that Dual Channel will improve your bandwidth say if you are running a 133Mhz chip more than running a 200Mhz chip because the AthlonXP can only use so much of it... so I would stick to what you have. :)

[Voodoo_fx]

lol, I thought he had an AthlonXP, he most likely has a P4 thats why he has that insane FSB.

[Mustanley]

Hah, now I see his FSB, so he is definitely running a P4. You will definitely not be able to overclock further adding a second dimm, in fact you may to drop back the FSB a bit to maintain stability. But the added bandwidth of running dual channel will more than make up for it. So, YES, go dual channel.

[solo]

I found another stick of the same RAM for around 70$, debating whether i should go for it or not. im hoping to hit 300 fsb wiht water cooling making it around 3.6 ghz :D

cheers

[[OC]This]

There is some serious AMD DC basher(one guy, especially) in this thread.

Sounds like a [H]ardcore Basher... joined in some sort of cult.

wow... most pathetic messages I'm seeing here. Originally posted by Mustanley
An Athlon XP Barton can't really make any use of the extra bandwidth dual channel provides because it runs on 166 or 200
what the hell is this..


sounds like he hasn't done any bench comparisons between SCDDR and DCDDR on the amd platforms. Also suggest to run the FSB at 200 or above, and then enable DC. Equivalent to 350mhz of clockspeed boost is not something to be put down.

That's why I said, if you don't want DC, or think DC makes no gain and all wasted, you don't need to overclock your system, at all. This is very good advice.. Do not overclock your system, just run it all at default. Overclocking will make no improvement. [turbo-charged sarcasm] ;)


This thread is being infected by few newbies in here.. even though this guy has Bong cooler setup, he doesn't seem to know "jack" about DC.

wow..



method().man and few others seem to posted pretty good replies here. Correct statements. I appreciate that.


:cool:

[[OC]This]

Rule #1, Learn to appreciate the performance gain you get from DC in the AMD platforms. It seems only highly educated/ highly experienced [H]ardcore overclockers(or top overclockers) know what's good.

On the other hand, I've seen too many incidents such as DC bashing by the newbies, or experienced member with ignorance.

[madcow235]

CALM DOWN!!!

[solo]

rar

so i wont be able to hit a higher overclock? as in speed, not fsb

cheers

[Flewdefur]

from what i gather by running in DC you will gain greater stability and efficency so there should be a slight boost in performance.

[[OC]This]

Originally posted by solo
rar

so i wont be able to hit a higher overclock? as in speed, not fsb

cheers
that is partially correct, but not necessarily.

if you have an older generation Nforce2 board i.e. A3 stepping, you'll have a little difficulty overclocking well beyond 200 FSB 128bit DC data bus while some of them can easily do 200 FSB DC and up. Otherwise, you can perform a small Vdd mod, which isn't difficult.

If you have the newer generation nforce2 A1 ultra400 stepping, you are highly likely be able to do beyond 200 FSBs DC without a fuss. As long as each of your memory is able to do 200FSB(if that is where you want it at), you'll be fine. Place both sticks in the appropriate dimm slot locations by your motherboard's manual for a pure 128 bit Dual Channel configuration. You may not even need good cooling on the NorthBridge chipset, which makes it even easiler for ya. ;) and don't get confused... there are ways to run SCDDR aka 64bit memory data bus if you place the memory sticks in the wrong dimm slots.

Here is the most important part of the entire thread, [b]you want to find every little bit and parts of source that will help boost your system performance aside from raw FSB speeds and memory timings.

Here are just some of the factors that will get you high performance out of your system:

- High FSBs
- Fairly tight memory timings
- 128 bit Dual Channel configuration
- CPU Interface enabled (equivalent to Fast CPU Decode)


These are the real juice and will only boost your system effectively, and humanly possible.

You can do all that without giving up on the FSB much.

You add here and there, you get a whopping grand peformance increase. That's the key. Yes, DCDDR plays an important role in here. and yes, I'm talking about the AMD system. don't forget.... ;)


Treat DC with some respect.

:cool:

[IceMan3928]

What were trying to say is that if you run dual channel there is a chance that you will have to back off your overclock to keep it stable, however sometimes (depending on your current bus speed and the type of chip your using) the lower FSB frequency is offset and sometimes outweighed by the increase in performance gained from using both channels.

and [OC]This, just relax and take a few deep breaths man. Flaming someone because of their hard headedness will get you no where, perhaps some links to a few benchmarks would help set the record straight for everyone. :)

[[OC]This]

talk about 3dmarks, there's a guy, who is a one of the best video card tweaker in the GF3 series, is scoobs. I do respect him.. for his 3d benchmark masterpieces.. same goes for another highly known member spliffs. Me, scoobs, and spliffs compete each other very hard in 3dmark. Often scoobs beats my ass.. and spliffs next. When I moved on to the very first generation Nforce2 platform, i did some testing with memory DC and SC DDR configurations.. as well as the rest of tweaks in the BIOS. The mad ass power came from the DCDDR and CPU Interface. All of us have some form of Fast CPU decode enabled for maximal performance. I finally beat scoobs and he asked me the other day, how do you get better performnace out of your system even though your video card is clocked far less than mine.. It was the DCDDR that helped me boost 350 points compare to a SCDDR config. SCDDR(with two sticks) is no better than a single stick, one channel SCDDR or the SCDDR on a KT266/333/400 systems based on my best judgement and observations. 350 points of difference is about 330mhz of system clockspeed boost(with the FSB at constant) on a scale of GF3 series and a TbredXP processor running at around 2.xx ghz sytem. 100 mhz of raw clockspeed(with the FSB constant) is about 105~ 110 points in 3dmark. 330mhz of bonus is not something to be ignored. That is not a small gain, what-so-ever.

My buds that I used to/am associate with all use DCDDR. Caprid uses it. Qwerty, aka MickeyMouse uses it. Shadco uses it. Mdzcpa use it. OPP uses it. RacerX uses it. If you mention DCDDR is no gain in performance and no good using it, he will say you'd be crazy to not use it. I liked that comment... that was close to a year ago. There are more members that know what's good. I respect all these people, because they know how to get most out of your system. DCDDR was one of the major ones that is considered "required", as a standard. I am more aggressive personality. You would get such response from me. Where there's some sort of non-sense talk going on, I cannot keep quiet.. It gets me off... get me pi.ssed off. I've been very patient.. and this kind of incident happened many times before.. I tried to keep quiet and let it go and which i did... but there are still more peeps that just don't get it... like this one. Somebody has to "refresh" their mind some .. to address these things.

:cool:

[Mustanley]

[OC]This seems to have a pretty substantial chip in his shoulder. Since I have no intention of getting involved in a flame war, I'll refrain from making any negative comments towards said member. If you happen to disagree with my comments, post some tangible evidence to support your position. Just calling me an AMD basher, even though I run two AMD systems, accomplishes nothing. I happen to be running dual channel with my Barton. Why, because I happen to have two 512 MB dimms. I have done plenty of benchmarks which show that running dual channel (sync'd with cpu ) offers up no more than a 5% gain in memory bandwidth. Please check out this thread that's been running in the AMD CPU section of the forums for the last several months.
http://forum.oc-forums.com/vb/showthread.php?s=&postid=1983759#post1983759

solo: You won't be able to overclock any higher, but you will effectively increase the memory bandwidth substantially, which will in turn give a nice perofrmance boost by running dual channel.

[Nolo31]

method().man and few others seem to posted pretty good replies here. Correct statements. I appreciate that.

You would know all about correct statments wouldn't you [OC]This...Saying that black pcb's on memory build up heat and mushkin memory sucks. Blow up motherboards, whatever.

I've done testing, there is not much of a difference at all between DC and SC. In fact, just to prove a point, ill test it with my system. I'll post some screenies in a bit.

[[OC]This]

Originally posted by Mustanley
[OC]This seems to have a pretty substantial chip in his shoulder.
you mean... many, many chips. ;) I suppose you're refering to the NorthBridge chipsets? Almost 10 of the nforce2 boards(both A2, A3, and A1 ultra 400 chipsets) i have worked/tested/ experimented on. I get all great FSB DC result out of every single one of them, well beyond 200 FSB DC. ;)


post some tangible evidence to support your position.
uhh... I am the tangible evidence. I am the information. :eek: :cool:


Just calling me an AMD basher, even though I run two AMD systems, accomplishes nothing.
Read more carefully next time.

I didn't say AMD Basher. I said "AMD DC Basher". Huge difference. ;)



I happen to be running dual channel with my Barton. Why, because I happen to have two 512 MB dimms. I have done plenty of benchmarks which show that running dual channel (sync'd with cpu ) offers up no more than a 5% gain in memory bandwidth.
are you depending/relying on the Sandra memory throughput? Don't rely on that. There's about 3% of gain in "actual/ real-world" applications including 3dmark and 3D gaming performance on a DCDDR@ 200+FSB versus SCDDR@ 200+FSBs based on the sources I've seen. That 3% sounds like a small gain, but it is not. It is often misleading to for our eyes to see such number. That is not a great way to see the comparisons. You will learn to see the difference in performance as you run 3Dmarks and 3d games to see the gain you're actually getting. Every single FPS counts. Gaining such FPS in intensive games@ high res settings would make it a hard to see the big of difference but it does help boost the system performance still. Base it on the 3dmark points, Framerates in games, and other type of office benchmarks to see the greater gains, not the "number 3%". What makes you think a simple overclocking(higher CPU and FSB) will get you? Give you wings?


Please check out this thread that's been running in the AMD CPU section of the forums for the last several months.
http://forum.oc-forums.com/vb/showthread.php?s=&postid=1983759#post1983759That was not out months ago. I've seen that before.

In that thread, let me quote..
[quote]Dual channel with two memory modules, memory buses in parallel

........................... Memory Controller <---- memory bus ----> Memory Module (dimm1/dimm2)
CPU <---- fsb ---->
........................... Memory Controller <---- memory bus ----> Memory Module (dimm3)

singleChannelMemoryBandwidth = memoryBus x 2 x 8 MB/s
maxMemoryBandwidth = 2 x memoryBus x 2 x 8 MB/s
x 2 because data is pumped at twice the bus frequency (using both rising and falling clock edges) in DDR (double data rate) form,
x 8 because the memory data bus is 64 bit or 8 Byte wide.
maybe he didn't put words together correctly or maybe taken wrongly, but DCDDR's memory data bus runs at 128bit, not 64bit. You can run two sticks at 64bit but that is not a true 128bit data bus. Of course you won't get full gain out of DCDDR due to CPU's FSB system, but within the memory data access department, it runs effectively faster, which results in faster overal system performance. He explains it down here.. read below.
Things are actually more complicated than that for nforce2 memory controller, there is some speculative prefetch data cache in the north bridge to make use of the dual channel bandwidth, so that some data are cached in the NB and can get to the CPU faster than getting from the memory. But the net is that until AMD can provide quad rate (or double fsb speed) than now, there is little or no advantage to make use of the dual memory bandwidth for the CPU.
Also pay attention to what he has said. He said "[b]there is little or no advantage to make use of the dual memory bandwidth for the CPU". For the CPU to operate the same as the Dual Channel. He was talking for the CPU, not the entire system overall. ;)

[[OC]This]

Originally posted by Nolo31


You would know all about correct statments wouldn't you [OC]This...Saying that black pcb's on memory build up heat and mushkin memory sucks. Blow up motherboards, whatever.

I've done testing, there is not much of a difference at all between DC and SC. In fact, just to prove a point, ill test it with my system. I'll post some screenies in a bit.
relax... I was joking about the Mushkin part. You've been holding grudges didn't you? ohh... poor baby. :p

I wouldn't rely on your bench comparison though. You could be biased, and make poor benchmark setup. sorry, but I can't rely on you. No offense intended.

[[OC]This]

Originally posted by Mustanley
solo: You won't be able to overclock any higher, but you will effectively increase the memory bandwidth substantially, which will in turn give a nice perofrmance boost by running dual channel.
That...... sounds a little more like it. Now I'm getting something.

[Mustanley]

Originally posted by [OC]This
[/b]
uhh... I am the tangible evidence. I am the information. :eek: :cool:[/B]
Originally posted by [OC]This
I wouldn't rely on your bench comparison though. You could be biased, and make poor benchmark setup. sorry, but I can't rely on you.
:rolleyes:

210x12 dual channel) Sandra Memory Bandwidth - 3186,3031
3Dmark2001 - 19375

210x12 single channel) Sandra Memory Bandwidth - 3107,2875
3Dmark2001 - 19345

200x12 dual channel) Sandra Memory Bandwidth - 3070,2926
3Dmark2001 - 19140

200x12 singlel channel) Sandra Memory Bandwidth - 2946,2737
3Dmark2001 - 18764

In these two comparisons which I tested this evening, 3dmark2001 shows only a 0.2% difference at 210 FSB, and a 2.0% difference at 200 FSB. The sandra scores show more of a difference than the 3dmark scores - 2.5% and 5.2% at 210 FSB
and then 4.1% and 6.5% at 200 FSB.
However, I have a feeling you don't trust my scores since I "could be biased and make poor benchmark setup".

This whole diversion is irrelevant in Solo's case because he is running 280 FSB, which obviously indicates a Pentium 4. The nforce2 dual channel thing has nothing to do with his system.

[solo]

:p correct, i am running a p4 2.4c at 280 fsb and im looking to go higher

DC will only improve bandwidth, and benchmark scores, and won't increase my FSB

thats what ive leearned so far. thanks for some informative posts lol

cheers

[[OC]This]

Originally posted by Mustanley


:rolleyes:

210x12 dual channel) Sandra Memory Bandwidth - 3186,3031
3Dmark2001 - 19375

210x12 single channel) Sandra Memory Bandwidth - 3107,2875
3Dmark2001 - 19345
LOL... that is most aweful benchmark I've seen in a long time, lol!! 30 pts difference, ha ha ha.. I think you have unstable system(performance wise) seems to fluctuate a lot, or just not a stable setup.

200x12 dual channel) Sandra Memory Bandwidth - 3070,2926
3Dmark2001 - 19140

200x12 singlel channel) Sandra Memory Bandwidth - 2946,2737
3Dmark2001 - 18764
Once again, I think your overclock system is not very stable. Seems a little more like where it should be in this overclock settings but the first one you posted here at 210 FSB seems out of range. Maybe your BIOS setting had been changed, or maybe your video card is not very stable i'm sorry to say.



In these two comparisons which I tested this evening, 3dmark2001 shows only a 0.2% difference at 210 FSB, and a 2.0% difference at 200 FSB. The sandra scores show more of a difference than the 3dmark scores - 2.5% and 5.2% at 210 FSB
and then 4.1% and 6.5% at 200 FSB.
If you go back and read what i've said in my previous reply to you, I've said don't rely on your Sandra memory scores. They are purely synthetic. I would advice you in 3dmark and games. Like mentioned before, your 210 FSB DCDDR/SCDDR benchmark is messed up, LOL. This is out of range.


However, I'm have a feeling you don't trust my scores since I "might be biased and make poor benchmark setup".
Not quite... like i've said, maybe your overclocked setting isn't very stable. Your benchmark is messed up, im afraid to say lol.

This whole diversion is irrelevant in Solo's case because he is running 280 FSB, which obviously indicates a Pentium 4. This whole nforce2 dual channel thing has nothing to do with his system.
AMD is AMD, P4 is P4. But you talking down of AMD DC performance based on your poor benchmark comparison doesn't make you right. Just because you got bad result at 210 FSB in AMD doesn't mean they hardly give you performance gain. LOL

[[OC]This]

Originally posted by solo
:p correct, i am running a p4 2.4c at 280 fsb and im looking to go higher

DC will only improve bandwidth, and benchmark scores, and won't increase my FSB

thats what ive leearned so far. thanks for some informative posts lol

cheers
No problem. :)

[Mustanley]

http://forum.oc-forums.com/vb/showthread.php?s=&threadid=218689
Originally posted by [OC]This[
Qwerty my man, do you believe in the performance of DC?

I am a DC junkie and i appreciate the performance of DCDDR.

There are few [H]ardcore AMD DC bashers over at memory forum section. Please contribute your response there. These idiots(two of them in that thread) don't know what's good. Tell them what system you're running and that you have DC enabled for best performance. They say DC really makes no gain in performance. What a turbocharged bullshi.t that is.

So now you are trying to convince other forum members to come to this thread and attack my statements? Ain't gonna happen. Your comments are very abrasive and it's not going to earn you much respect on this forum.

[[OC]This]

Originally posted by Mustanley
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=218689


So now you are trying to convince other forum members to come to this thread and attack my statements? Ain't gonna happen. Your comments are very abrasive and it's not going to earn you much respect on this forum.
not quite....

The member Qwerty is not a troll. He is too good for you. He is ... way up there, in the high-class overclocker. He doesn't have to attack you. He can just simply state/report his findings about the 128bit DC. Besides, if you read the thread carefully, I said I've asked him to contribute his response. I didn't say anything about attacking/ flame throwing. I'm too good for that.

:cool:

whether it's respect or not, I do not acquire respect. I don't care for such. I'm not here to help people. I have told that to the mods long way back. I've told them I'm here to "show off", not to help people. I may use some words and do some explanations if I really feel the necessity, but mostly, no i'm sorry to say. :eek:


:cool:

[[OC]This]

You shouldn't be taking this too badly. I'm not here to attack you. I have no interest to attack you. Hope you get what i'm saying.

btw.. you like my pic? eh? :D

[james.miller]

you need to calm down [OC]This. try reading the forum rules, and try following them.

after you've done that, read this:
http://home.earthlink.net/~deeppow/OC_Guide/Mem_Mobo_CPU_Overclocking_Guide.htm#dualddr

dual channel has a substancial effect on amd platforms runinng <166mhz but as you pump up the fsb, the gains become smaller and smaller. When over 200mhz fsb, the gains are as small as 2% - which Mustanley, and the link i posted, have proven.
http://homepage.ntlworld.com/james.miller700/pics/dc_vs_sc.jpg
is he wrong aswell?

[[OC]This]

Originally posted by james.miller
you need to calm down [OC]This. try reading the forum rules, and try following them.

after you've done that, read this:
http://home.earthlink.net/~deeppow/OC_Guide/Mem_Mobo_CPU_Overclocking_Guide.htm#dualddr

dual channel has a substancial effect on amd platforms runinng <166mhz but as you pump up the fsb, the gains become smaller and smaller. When over 200mhz fsb, the gains are as small as 2% - which Mustanley, and the link i posted, have proven.
http://homepage.ntlworld.com/james.miller700/pics/dc_vs_sc.jpg
is he wrong aswell?
uh huh, yep.


not just wrong, but aweful, and poorly prepared.



Take a good look at his nforce2 "guide".

First of all, hope he knows which dimm slots configuration is DCDDR for the Epox 8RDA board and the which configuration for the Abit NF7 series board. They don't have the same dimm arrangement as the Epox.

He includes "single 128bit controller" in the memory channel. There is no "single 128bit controller". When one stick, or two sticks populated in the proper dimm slots(accordingly for the manufacturer) for SCDDR will get you "64bit" memory data bus. "128bit" memory data bus is enabled when the two (or three sticks) sticks memory is populated accordingly for the DCDDR. Let's not get this confused before we go any further.


His grammar is poor. He continously says "mobo". Some of his words are a bit confusing(beside mobo). Doesn't look good to be a reliable, reputable review. And the message below explains why. Read the quotes and my response below.


He numbered the CPU speed wrong to begin with. He says "CPU FSB speed of 2.2Mhz (CPU multiplier at 11).". 2.2Mhz?? Not only that, but he has made same mistake repeatedly later on. :rolleyes:


Most of his benchmarking programs are Synthetic. Sisoft Sandra and other synthetic benches which I find it rediculous to even rely on. Only descent program he did include is a 3Dmark.


What's more? He doesn't seem to know what he is doing when overclocking and making good benches. He says

"After many hours of testing I found that the maximum CPU FSB that would run all the tests for a long period of time was 194Mhz. Thus the results use this as a maximum FSB and also use a constant CPU speed of 2.328Mhz (CPU multiplier at 12)..."

Note that he has made another mistake "2.328Mhz". What's worse? He says 194Mhz is the highest FSB he could run the benches. Running his system to its max does NOT make a stable system and to depend on performance result. When he puts himself to the edge, the system performance will fluctuate highly. The 3dmark score will jump up and down as if they are out of control. This is not very smart.. When his system is pushed to his max FSB, the system will not be reliable in getting accurate data on the performance result. Also,He mentions nothing about cooling and if the vmods were done to it/them. Improper cooling will get you problems, too. The system will not perform steady as he up the FSB to a such level.


This also explains that he ran repetitive runs(5 runs) on few of the overclocked configurations instead of all of them. What kind of garbage benchmark comparison is this??? Why run only few of them with 5 runs but not the rest?? Sounds like he just ran one bench for the other overclocked configurations. What he is doing is totally unprofessional. His tests are not consistent and seem to base a lot on his thoughts and guesses. He can't even overclock his system to a rock solid stable 200mhz FSB. Running unstable systems is not the way to go.


This is one of the most poorly done review/bench comparison source I've seen in years. You show me this as a source? Garbage?

[james.miller]

garbage? lol

well, ok. you calling that garmabe is a little unfair, especially when he cant defend himself:rolleyes:







oh wait, he can: Deepow (http://forum.oc-forums.com/vb/member.php?s=&action=getinfo&userid=17244)

and while your at it, why dont you rant and rave about maklin01 and ---X---, both of whom are well respected member here.

[L337 M33P]

Ok, time to don a flame-retardent suit as this thread is toasty.
[OC]This, your posts are wayyy to inflammatory and I too suggest you calm down, read the rules and post positive things instead of ranting.

It is true that DCDDR does offer some form of improvement in situations where the FSB is <166 because the Northbridge has functions like speculative data caching. The signal coming from both memory modules also need to be synchronised. This is why at a higher FSB the performance gain in memory benchmarks is negated, because the resources taken up in keeping the signals synchronised are much greater near the limits of the chip.

EDIT Maybe you should look further down the page. In those graphs it shows the raw benchmark scores for memory and the 3Dmark 2k1 scores for the relevant memory speed. As you can see the performance advantage reduces in DCDDR mode at higher FSB speeds with the raw mem benchmarks, but 3dMark scores with DCDDR actually take the lead over single channel at high FSB.

[deeppow]

Ah yes, it is [OC]This again.

You folks have to realize that he likes to jerk threads and folks around to get a reaction. Here is an example of his handy work, http://forum.oc-forums.com/vb/showthread.php?s=&threadid=217362. Knock yourself out dude.

Be :cool:

[L337 M33P]

LOL 'nuff said.

Please stop slandering everything [OC]This, it's only making you look silly.

[[OC]This]

Originally posted by james.miller
garbage? lol

well, ok. you calling that garmabe is a little unfair, especially when he cant defend himself:rolleyes:







oh wait, he can: Deepow (http://forum.oc-forums.com/vb/member.php?s=&action=getinfo&userid=17244)

and while your at it, why dont you rant and rave about maklin01 and ---X---, both of whom are well respected member here.
mm! mm!! ..... excuses and excuses.. running out of words? you've lost already, sorry.


;)

[[OC]This]

Originally posted by L337 M33P
EDIT Maybe you should look further down the page. In those graphs it shows the raw benchmark scores for memory and the 3Dmark 2k1 scores for the relevant memory speed. As you can see the performance advantage reduces in DCDDR mode at higher FSB speeds with the raw mem benchmarks, but 3dMark scores with DCDDR actually take the lead over single channel at high FSB.
Haven't you read the previous messages posted?

How many times do I need to say stop comparing with the synthetic Sandra memory bandwidth scores?

This is one of the most idiodic posts I'm seeing more here. Not reading my messages and just spewing makes you look dumb.
remember this?? you say:Please stop slandering everything [OC]This, it's only making you look silly.
Look who's talking. :eek:

[[OC]This]

Originally posted by deeppow
Ah yes, it is [OC]This again.

You folks have to realize that he likes to jerk threads and folks around to get a reaction. Here is an example of his handy work, http://forum.oc-forums.com/vb/showthread.php?s=&threadid=217362. Knock yourself out dude.

Be :cool:
relax.. as mentioned before, I was joking about the Mushkin part. I did admit it was fool of me doing that. I've PM'ed to Nollo(member is this forum) about it. ;)

:cool:

[[OC]This]

Set-in-stone Rule Number 1:


Dual Channel boost insane amount of performance and give you wings and a turbocharger. (now you're pushing around, it is time for me to do some exaggerations). Dual Channel is everything. Without it, everything is pointless. Hopeless. Meaningless.. Lifeless. DC is everything. DC will give you 99% boost of your entire system performance. Your framerate will add another 80+. Your 3dmark score goes through the roof and ... ... never return. :eek:


Exaggeration will be far strenghtened as you are at it. ;)

DC is everything. Worship DC. Get on your knees and bow.

:cool:

[L337 M33P]

^^ Ok my opinion on your validity of posts has just plummeted. I think I got yer with the "Stop slandering stuff" since all you could come up with is "look who's talking" when I have made no slander whatsoever. Oh and BTW, Illiminate is spelt Eliminate.

[[OC]This]

Originally posted by L337 M33P
^^ Ok my opinion on your validity of posts has just plummeted. I think I got yer with the "Stop slandering stuff" since all you could come up with is "look who's talking" when I have made no slander whatsoever.
excuses.. and excuses... :p



Oh and BTW, Illiminate is spelt Eliminate. :o Thanks for correcting my spelling.

Btw, it is spelled "spelled", not "spelt". :rolleyes:

[L337 M33P]

Ok this has gone far enough. I will NOT be drawn into a petty flame war over immature and foolish comments.

[[OC]This]

Originally posted by L337 M33P
Ok this has gone far enough. I will NOT be drawn into a petty flame war over immature and foolish comments.
not quite... read my previous post again. I had to edit it.. noticed your mistake. ;)


No, because now you want out of it. Not so easy **sunshine**.

:cool: