• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

2.0-3-2-6 or 2.5-2-2-6?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

TWY

Member
Joined
Jul 17, 2002
Location
Singapore
2.0-3-2-6 or 2.5-2-2-6 would be faster?

Anyone did any testing with these 2 sets of RAM timings? :-/
 
BTW, can those Samsung TCCD 4ns chips run 2.5-2-2-6 at about DDR466? Seems like review sites test them at 2.0-3-3-6 instead.
 
Thanks.. :)

Does anyone have any idea whether the Samsung TCCD 4ns chips could do a 2-2 timing on the RAS-CAS Delay and Precharge at DDR466?
 
You want the two middle timings as low as possible. 2.5-2-2-5 is only a little faster at 3DMark2001 than 3-2-2-5 (like 110 points in the 15k area).

Similarly, 3-2-3-5 will beat 2.5-3-3-5. And 2-3-3-6 is a tad slower than 3-2-3-5.

While 3-3-3-5 is ugly, overall it provides excellent performance. 255-265, 1:1 is possible with mT -5B C ram fairly easy, depending on board. 260, 1:1, 3-3-3-5 will outperform 260, 5:4, 2-2-2-5.

2.5-4-3-6 is amazingly slow. About the only thing it beats is 2.5-4-4-7. If you are used to BH-5, I wouldn't get worked up about 2.5-3-3-5 either.
 
I agree with highlander and clevor. I have kingston ddr400 wich I can get down to 2.5-2-2-5 though 2.5-2-2-6 is faster.

Also clevor mentioned that a sync with high timings will outperform a devider with low timings. This is because the devider actually has a latency of its own, not to mention all the other bottlenecks and freewheels it causes.
 
i like the new samsung tccd chips, they are really sick! Like crucial's ballistix ram...
 
CLEVOR

Read your PMs.
Sorry to steal the thread.....but I can add that when I ran a strong 3Dmark run at 1:1 263 MHz fsb 2.5-4-4-7 I got about 700 points less than running 5:4 2-2-2-5 at same fsb. Lower latency with a divider is best for 3DMark than a sync run with low timings. However, benches like PCMark2002 gave serious memory scores with the 1:1 ratio vesus using a divider.
 
A big CPU counts for a lot too. I just finished testing a bunch of 3.0Cs. I was running Vitesta at 1:1, 3-4-4-7 as I didn't want the ram to be a factor in stability. With a 3.0C at 260 fsb or 3.9 gig, I got a 12,211 PCMark2002 score and a 36 sec SuperPI 1M time. Hell, CAS timings don't get any slower than that :p. Wait till I put that chip in a P4C800-E and run 260, 1:1, 3-3-3-5.
 
Back