Archived from groups: alt.comp.hardware.overclocking.amd (
More info?)
"Key-Bored" <f@home.com> writes:
>On my Shutte AN35N Ultra I can oc my 2100+ to 2.17 (217x10) with my 2 sticks
>(256 each) of pc3200 ram in single-channel mode.
>However, in dual channel mode, it gets unstable past 200 and the highest I
>can get is 200x10.5.
>My question is, which is better: 2170 w/single channel memory, or 2100
>w/dual-channel memory?
Most folks with typical applications seem to see well under 10% gain with
dual channel memory, maybe somewhere between 5 and 7% seems typical.
Most folks who don't have some sort of stopwatch to measure the difference,
or who don't put two systems side by side so they can visually subtract the
two times, will likely never notice any gain less than 30%, and only be
able to notice something that small for a very short period of time.
If you don't believe that it would be easy for someone to set up a carefully
controlled experiment to test whether you can dependable detect <30%. Let
you look at the system, send you outside, toss a coin to either speed it up
or not, bring you back in an hour and ask you to look at it and decide.
Repeat that a few times and you will probably have a lot less appreciation
for the people who are going crazy to get an extra 4% out of their system.
In one case I actually had a user who did not notice a 90% increase in
speed, when they were not given any hints that this was what they were
supposed to be looking for, but that was likely a special case.