Archived from groups: alt.comp.hardware.overclocking.amd (
More info?)
<agustine@benedictsass.com> wrote in message
news:0ivl71hc293btn76e5lhecf274l2ks86u6@4ax.com...
> On Thu, 05 May 2005 02:05:16 GMT, "Peter Harrington"
> <pharring@columbus.rr.com> wrote:
>>Just a follow-up. I was curious if dual channel actually interferes with
>>overclocking and hitting faster FSB speeds. We had problems overclocking
>>a
>>dual channel board when the RAM modules were not matched, but were the
>>same
>>model number and vendor. Going to single channel solved all our woes.
>>
>>Has anyone achieved faster speeds by switching to single channel? I may
>>have to test it might self when I am not so busy.
>
> It's easier to run faster and stable with a singe channel setup.
>
> Dual channel just adds more complications for the memory system.
>
> It's only really a problem if you use cheap memory.
>
> Various tech websites are full of old articles about benchmarked races
> between all kinds of silly technology. single, dual, hyperthreaded,
> with the NEW extensions, with the faster memory, what CAS ????, etc..
>
> and you know what ? The games are so bloated these days none of that
> will get you more than a dozen frames per second difference.
>
>
Yesterday, I did this experiment. I switched my system from dual channel to
single channel. I have two 1 GB ram sticks and overclocking 12x214. I am
using the Matlab bench program, because MATLAB consumes the largest amount
of CPU on my systems. For the FFT test, there was a substantial loss in
performance when I switched to single channel. The FFT is the most
representative test for my computations.
My goal for stability was to hit 12x218 because that is the speed of my RAM.
Both single and dual channel failed the Prime95 torture test in about a
minute at this speed, so I did not see any justification for not using dual
channel.
Best wishes,
Pete