So what really is faster...3.8GHZ CPU but NB at 2000 or 3000MHZ?

[demonkevy666]

This should be a sticky

the percent of Nb to core for am3 is higher then Am2+

NB AM3 70%-67% memory should be 40%-50% of Nb speed

NB AM2+ 45%-67% memory should be 40%-50% of Nb speed.

note tight timing are still best in AMD.

[Zeus]

This should be a sticky

That's what i said a couple of posts ago .

Come on mods, sticky please?!

[Tom128]

+1 for sticky. The posts by tony a few pages in were particularly helpful with how to clock a phenom properly for either better clocks or lower volts to reach the clocks.

[radaB]

+2 sticky. helped me lots. thanks for the great data everyone!

[Astennu]

Thanks Tony !!! Good Info !

[Brutuz]

Sorry for the bump, but I saw this wiki page about this and I was wondering if OCZTony could repeat the tests with the RAM at 1600Mhz CL6?

[chew*]

Sorry for the bump, but I saw this wiki page about this and I was wondering if OCZTony could repeat the tests with the RAM at 1600Mhz CL6?

I think the problem that lies with that is that the only "affordable" sticks that can do c6 are d9's and they are 1x1 gig and require high volts and the sticks that are 2x2 gig and low volts are usually not in mainstream AMD users budgets.

The whole point of using AMD is bang for buck not to mention cas 6 offers a small boost in performance "realworld", albeit a decent boost not the $150 more for a set of ram type of boost that is cost effective to most users.

[Brutuz]

I think the problem that lies with that is that the only "affordable" sticks that can do c6 are d9's and they are 1x1 gig and require high volts and the sticks that are 2x2 gig and low volts are usually not in mainstream AMD users budgets.

The whole point of using AMD is bang for buck not to mention cas 6 offers a small boost in performance "realworld", albeit a decent boost not the $150 more for a set of ram type of boost that is cost effective to most users.

True, true... I still think it'd be useful to know, just in case.

Edit: To add to what I mean, it'd be useful for the future when the fast, low latency/good chips are cheap, like it did with DDR2 and DDR1.

[chew*]

True, true... I still think it'd be useful to know, just in case.

Edit: To add to what I mean, it'd be useful for the future when the fast, low latency/good chips are cheap, like it did with DDR2 and DDR1.

yep agreed but sadly prices recently went UP and i don't see this happening to soon.

[Brutuz]

yep agreed but sadly prices recently went UP and i don't see this happening to soon.

Yeah, it's a shame... I wonder if AMDs Bulldozers IMC will react the same.

[FlanK3r]

what is strange, i have always problem with 3GHz at NB (AIR)...WTF? My 955 and now too 965 C3. The stangest is, with x4 955 at 2800+ i needed +0.3V. With C3 only +0.225V. But 3Ghz is impossible. With a luck sometimes im booting at 3000MHz at NB, but superpi crashed or windows frozed.

[Zeus]

3000NB does not come easy by any means. My 940 could do 3GHz NB but only 1M stable, my 955BE C2 could do up 3040 for 1M but this C3 gives in around 2900.

Supercooling the cpu helps a lot for NB clocks.

[FlanK3r]

My 955 cant hit 3000, only about 2870MHz, and now C3 is +- the same (2900+). Only diferent its, 2800 is for my C3 very easy with low voltage. At 3000 dont help me CPU-NB at 1.5V (+0.4V), strange, because 2870 i can hit in Cinebench at 0.25V, for superpi lower.

[Brutuz]

what is strange, i have always problem with 3GHz at NB (AIR)...WTF? My 955 and now too 965 C3. The stangest is, with x4 955 at 2800+ i needed +0.3V. With C3 only +0.225V. But 3Ghz is impossible. With a luck sometimes im booting at 3000MHz at NB, but superpi crashed or windows frozed.

Be thankful, my unlocked 550 BE can't even get 2200Mhz stable at 1.5v.

[richierich]

Still not a sticky???

[Eternalightwith]

You going to update this for Thuban, Tony?

[Peter Burg]

sticky?! ...!