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

Recommended memory for p5k premium MB?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Maverick0984
@ Supershanks, I sure hope it doesn't take you 2.25v to get those to 1200Mhz...

DDR2-800 Ballistix chips can reach 1066 at 2.2v still with slightly looser timings. You shouldn't be recommending those to anyone btw. Unless by some chance your FSB for your CPU is at 600Mhz (which it isn't). Clock your ram back down, try to get to 1:1 ratio with the CPU FSB and lower the timings of the RAM. That is the only place where you might, and a strong might, have a performance boost. Running them at 1200Mhz buys you nothing, other than saying you are running them at that speed. There are no performance gains.

Again, unless you are actually able to achieve a 600FSB on your CPU. Which I assure you, you cannot .
Well posted my results, you got freeagent's attention:), would you care to enlighten us with the knowledge that triggered the above.

Apologies to v8440 if it's a little offtopic, but throw away comments like the above can confuse newbies, hell they confused me.:)
 
I definitely agree w/ what FreeAgent is saying about running the FSB faster using a lower CPU multiplier.

However, using a memory multiplier you are running your memory faster than the FSB. Memory has to travel along the FSB, so I don't see much of an advantage to running it faster than the FSB. I see your above results w/ 400 4-4-4-12 compared to 600 5-5-5-15. I think a fairer comparison would be between 400 4-4-4-12, 500 5-5-5-15, & 600 6-6-6-18. I think the bandwidth will be about the same. I think you'll find that if you run it at 400 and tweak your timings & sub-timings to the absolute lowest you can get them, and the do the same at 600 that the spread b/n them won't be as dramatic. And even though Everest is giving better bandwidth scores in real-world usage you won't see much if any difference.

I have played w/ 900 and 1080. I wasn't able to get 1125 to play, but maybe I just needed to loosen the timings a bit. Anyway my 3DMark06 scores went up a whopping 6 points w/ 1080.

Running the RAM w/ a multiplier may have some benefits w/ some benchmarks, but you aren't going to notice it in real-world applications, and if it causes my RAM to die b/c I have to give it extra volts or causes crashes due to instability then it is definitely not worth it.

Still, I can understand the need/want to experiment w/ it and see what happens. I just feel it is important to keep the RAM 1:1 w/ loose timings while trying to find your max CPU and FSB frequencies. After that have fun, but if you start getting crashes make sure to revert your RAM back to 1:1 to eliminate that as a possible source in your trouble-shooting quest.

Edit: Going back to the original question I think good cheap (~$70) D9 DDR2-800 is the way to go. DDR2-1000+ at $130 or more just isn't worth the extra cost for the very minimal gains unless you just have money to burn.
 
Last edited:
Back