DDR-1200 4-3-3-8 2T @ nv680i with QX6700+Team PC8000

[VictorWang]

NV680I for Intel Refference board, all stock
Intel QX6700es B3, watercooling
Team ddr-1000 Victor ED + 2000rpm 8cm fan
ambient=15c


D9GMH benchable 1200 4-3-3-8 2T with only 2.5v @.@ nv680i's mem-controller = amazing !




since the mobo have fsb issue with my QX6700, so... lower the fsb and raise the m/p.
1.58v 4G pi_1m, need 0.08v more than P5W64 WS Pro, wow~~~!!
pi_1m about 0.15sec slower than asus' 975x mobo




seems bios still not running well with quad-core, so I disable core#2\3.
1.58v 4G pi_32m done ^_^, 0.04v more than P5W64 WS Pro.
pi_32m about 1min slower than asus' 975x mobo, eeeee~~~~~~

[EnJoY]

Nice victor!

Have you tested max 1T at all?

[-Sweeper_]

great results victor
guys, could you tell me if 2Gb cellshock ddr2-1000@4-4-4-12 2.2 D9GKX would be a good deal?
i found one for less than U$400, and i'd like to overclock like hell, with my e6700/q6400.
tks, sorry 4 my english

[Tony]

Victor

680I will clock well with 5-3-3- or 4-3-3- or 5-5-5 or 5-4-4- as the bios seems to have issues. The only setting that seems to work as it should is CAS which cas4 at 1200 is pretty good.

I had Flex at 1300+ with 5-3-3- on EVGA but it performed the same as 5-5-5.

make sure the TRCD and TRP are making a difference, bios may be fooling you.

[bachus_anonym]

I remember, guess2098 was showing same timmings @ 1200 on his 680i sometime ago and I was also wondering if 8-3-3-4.0 had any performance gain over 8-4-4-4.0 that would be worth writing home about or it's just another glitch...

So, I run some of those myself as well at that time and there was in fact some gain, that was repeatable over the course of few SPi32M runs. So, it might be something useful for those, that are looking to squueze out those last extra seconds of their 32M times - in below case it was 9s...

5-3-3-4.0-2T @ 1200 -- vs -- 5-4-4-4.0 @ 1200
(identical secondary mem settings, voltages, methodology and OS conditions)

... so, I think it's definetely interesting... That doesn't change the fact, that still 1200 @ 5-3-3-4.0 on 680i does not produce better 32M times that 1200 @ 8-4-4-4.0 on ASUS P5B Deluxe... Whatever the problem is...

Thanks for sharing, Victor!

[EnJoY]

... so, I think it's definetely interesting... That doesn't change the fact, that still 1200 @ 5-3-3-4.0 on 680i does not produce better 32M times that 1200 @ 8-4-4-4.0 on ASUS P5B Deluxe... Whatever the problem is...

So apparently whether it's in the memory controller itself due to nvidia's design, or in the bios, some of the timings aren't doing what they should, or atleast to the same effect that we're used too. Very strange.

Bachus, have you tried to see if it makes more of an impact with 1T enabled?

[VictorWang]

yup, MC of nv680i is wired.
4-3-3-x, 4-4-4-x, no difference.
but...sub timings really have some effect.
especially tRFC & tRRD, pi_1m about 0.2sec faster after tweak.
any one feel the same ?

[EnJoY]

yup, MC of nv680i is wired.
4-3-3-x, 4-4-4-x, no difference.
but...sub timings really have some effect.
especially tRFC & tRRD, pi_1m about 0.2sec faster after tweak.
any one feel the same ?

What's the fast TRRD for evga 680i?

I know I can get mine all the way down to 1 on my P5B, but apparently 5 is actually faster.

[guess2098]

What's the fast TRRD for evga 680i?

I know I can get mine all the way down to 1 on my P5B, but apparently 5 is actually faster.

in my thread, my TRRD is at 6 i think, rest are all in 1
lol yes! all in 1. like 1-1-1-6-1-1 i remember

http://www.xtremesystems.org/forums/...d.php?t=125062

[VictorWang]

What's the fast TRRD for evga 680i?

I know I can get mine all the way down to 1 on my P5B, but apparently 5 is actually faster.

nv680i refference mobo (evga/bfg/ecs), min=tRRD=1, but not stable with high ram_clk. sure faster than 5 @.@