[First Look] DFI 680i LT NF-680i LT SLI (56K Warning~~~)

[Borgod]

First thing I did when I got home was update to the latest BIOS. I like the small tweaks they've made.

Went straight to 295x12 3550Mhz and it crashed PrimeStable almost immediately.

I increased vcore and NBvolt a little and changed to 3510Mhz and it was Prime stable 6.5 hours.

Funny thing when I got up this morning to check it I turned my LCD on, moved the mouse to turn off screen saver and when the LCD displayed it showed a BSOD that only occured moments earlier. It seems coming out of screensaver caused the BSOD. Moments before I turned on my LCD I checked my CPU useage with my G15 Keyboard and noticed all cores running 100%. CoreTemp logs confirmed it had been running for 6.5 hours.

Funny that. I'll be sure to turn off my screensaver tonight.

Anyhow to me this confirms it was not my PSU that caused the shutdown. With increased voltage to CPU and NB, the Galaxy had no problem running Prime for 6.5 hours overnight. From memory vcore was 1.4625 and NB was 1.4375 (close to anyway).
NB temps are screaming at over 70degrees now, so I'll be rotating the NB HSF 180 and will put less TIM on it soon.

I'm still happy with progress. GTL tweaks will need to be made soon.
Considering I haven't lapped my CPU and I have a fair bit of headroom in my vcore to play with, I still have a bit to go

NB: I really enjoy the new option with the latest BIOS to modify the CPU temp delta. Now, on load Guardian is the same as Coretemp for me gg DFI

On another note, looking through the hwbot rankings, my 1024M and 32M wPrime times are the fastest for my CPU frequency. Infact my times are faster than others that run their QX6700 at a higher frequency. This is with a multiplier of 12, and 'basic' RAM timings of 4-4-4-1T, using Vista

I don't know what motherboards or OS the others are using, but this MB damn fast

[owcraftsman]

Raja I will get some Ceramique & give that a go on the weekend along with ensuring the heatsink is flat.

The rig is running Orthos small fft's at the moment (2 1/2 hours so far) at 462.5 x 8 3.7GHz with a 70mm fan on the PWM heatsink & a 120mm blowing down between the CPU & NB toward the long chips in front of the PWM heatsink. The heatsink temp is mid 20's & the long chips are low 30's.

One thing I now notice is Smartguardian is running very close to Coretemp (high 60's low 70's) maybe its the fans maybe its the later 521 BIOS I will investigate later & post my findings.

Regards

Craig.

I haven't hooked up my lp 680i lt yet just got it this week but I do believe the bios enables such an action. I will see if i can find the info I speak of. Unfortunately I think I saw this at DFI-Street is now DIY and not all was archived after the debacle. Beside that, I speak of the NF4/AMD platform. I'm yet to be baptized in Intel/DDR2 O'cing so this may be off base.

[Jobeo]

anyone have experience removing the PWM heatsink?? i think i recall someone mentioned it here and found some very tough stuff underneath... was hard to remove.

[Borgod]

I won't ever remove my PWM heatsink, not after the dramas I had with the Abit QuadGT.

[virtualrain]

On another note, looking through the hwbot rankings, my 1024M and 32M wPrime times are the fastest for my CPU frequency. Infact my times are faster than others that run their QX6700 at a higher frequency. This is with a multiplier of 12, and 'basic' RAM timings of 4-4-4-1T, using Vista

I don't know what motherboards or OS the others are using, but this MB damn fast

Borgod, where are the wPrime times you are referring to? I'd like to check them out. What are your times?

[virtualrain]

BTW, there are some $10 off coupons for this board from Motherboardpro.com in the deals forum.

[vincey2kr1]

Still eager to know what the new bios brought? Probably flash my bios tonight especially if it keeps the saved cmos.

[Raja@ASUS]

major changes I have found,

Unlocked upward multi for XE edition cpu's

Temp threshold adjustment/rearrangement of naming to reflect NB and sb temps..

This is what I myself have found..

regards
Raja

[Borgod]

virtualrain: goto www.hwbot.org or more information in the XS Forums <here>
Join i4memory.com team

Vincey: BIOS changes I found:

-Unlocked Multiplier beyond 10 for Extreme CPU's
-Once you have selected your multiplier, you can now lock it
-Added CPU Temp delta settings, so you can match it to CoreTemp etc. Note the scaling is still different, so idle temps in Guardian/BIOS will be different to idle temps in CoreTemp if they're both the same on load

Also, yes the BIOS's are saved, but something is screwy with the FSB if you load your old BIOS settings. eg I loaded an old BIOS setting and although it was set to 10x320, for some reason it calculated this to be a MUCH bigger FSB, so my final frequency was set to something like 28000Mhz (yes really)
Make sure you check this, as you dont want to boot your PC set at this frequency of course.
This occured with all 3 old saved BIOS settings.

Another bug, all saved BIOS's will save vdimm at 2.88v
Make sure you change vdimm to correct voltage before saving and rebooting.

NB and SB temps are now displayed properly

[Raja@ASUS]

I have not had the vdimm over voltage save since the older bios's, I find it stays where I save it to. However it would be wise to check before reboot just to make sure.

Also do not use previously saved cmos bank settings after a bios flash, make sure you over write these with new ones..

regards
Raja

[Borgod]

I have not had the vdimm over voltage save since the older bios's, I find it stays where I save it to. However it would be wise to check before reboot just to make sure.

Also do not use previously saved cmos bank settings after a bios flash, make sure you over write these with new ones..

Yes it stays when I tell it to, but if I load a BIOS Usersave from the OLD bios, the vdimm voltage is overwritten with 2.8vdimm.

Indeed overight with new ones.

Aways double check before loading a BIOS - especially OC'd.

[Jobeo]

ya i just wrote over all my old saved BIOS after i flashed... then again i didn't really have any i wanted to save LOL

also came across possibly the easiest _quad_ overclock i've ever done ... just bumping the multi from 10 to 11 gives a 10&#37; boost and required absolutely no voltage changes to pass OCCT ...

the temps stay low too cause everything is drooping from the stock BIOS voltage settings...

[virtualrain]

I have not had the vdimm over voltage save since the older bios's, I find it stays where I save it to. However it would be wise to check before reboot just to make sure.

Also do not use previously saved cmos bank settings after a bios flash, make sure you over write these with new ones..

regards
Raja

I agree, the whole mapping of settings to values could change with each bios release making it risky to load old settings into a new flash.

I finally have time to flash mine now. Will report later.

Borgod... thanks I'll look into that when I get a minute.

[Raja@ASUS]

Yes it stays when I tell it to, but if I load a BIOS Usersave from the OLD bios, the vdimm voltage is overwritten with 2.8vdimm.

Indeed overight with new ones.

Aways double check before loading a BIOS - especially OC'd.

By old bios you mean settings from an older bios version? If that is the case you should never use older cmos saves on newer bios's, that will create issues such as these.


Each new bios means yo0u must overwrite your user cmos saves and start again, do not use old saves on new bios' flashes !
Using the same bios version with fresh cmos saves I have none of the over voltage issues you mention.

regards
Raja

[virtualrain]

Initial findings on latest BIOS (as most others have previously reported):

I'm using the 2C BIOS on my X3220 as it says it's for Q6600!

PC Health and Status:
- Temperature shutdown defaults to disabled
- NB Temp correctly labeled
- New option for CPU temp offset

FSB:
- Multi option is now available (and works! ) - testing 450x8 now with all other settings the same as before.

Outstanding Issues:
- Voltage set vs. get hasn't changed (still major discrepancies between what you set and what you get)
- GTL entries are still mis-labeled

Hopefully it's not a whole new challenge to find working GTL values at this FSB.

[Borgod]

... snip...

I know mate I agree with you, just giving 'fresh' evidence to support both our recomendations

[Jobeo]

well.. 11 and 12 multi's both proved stable at all the stock settings.

13 multi required me to increase the VID to what the processor is spec'd for around 1.35v but no other changes. this brings me to my highest cpu clock i've achieved on the board of 3467 mhz and it passes OCCT on air which seems incredible.

although i realize the memory throughput is restricted by the lower FSB. i'm not sure how to calculate what FSB is required to fully support 3.4ghz... but regardless this clocking gives me my fastest super PI ever!

[Borgod]

SuperPi and wprime times are largely ineffected by FSB speeds I've found... its the raw CPU frequency and RAM timings that make the difference... in my experience.

[vincey2kr1]

hum updated to beta bios but i cant change the multiplier? You do it in the FSB menu correct?

[virtualrain]

Well, I spent all evening trying to get a good overclock on my x8 multi. I was shooting for 450x8 to equal my previous 400x9. Board/CPU/NB just couldn't do it even with voltages maxed, I couldn't load Vista after 430FSB... tried every GTL setting under the sun and the moon!

So, it would appear something on my system is FSB limited as I can boot Vista at 430x9 but not 440x8.

So, back to x9 multi. I'm going to see if I can increase my previous overclock at all.

BTW, Jobeo, nice OC on air!

What are your voltages (CPU, NB, SB, VTT)?

[C-N]

My rig ran Orthos small fft's for 5 hours last night at 1850fsb/462.5 x 8 = 3.7GHz then I stopped it.

I then selected blend & it shutdown after 45minutes.

I had plenty of air blowing on the PWM & I am fairly convinced the PWM overheating is not the cause of the shutdowns.

The Coretemp/Smartguardian temps ran very close for the 5 hours which is much better than the 15oC difference there used to be with the last BIOS.

Now the temps run closer to each other & coupled with the Coretemp & Orthos logs I am sure its not a CPU temp shutdown problem either.

I was also running 2.3vdimm @ DDR925 CL5 on these sticks which is well within what I have previously tested these sticks to (DDR1000 @ 2.15v) so I am also sure the shutdowns aren't RAM related.

I will try messing around with voltages tonight & see if I can increase the 45min shutdown time.

Regards

Craig.

[Borgod]

Silly question maybe, but is your RAM running in sync?

[C-N]

Silly question maybe, but is your RAM running in sync?

Yes although I have tried unlinked @ DDR900 where I know this RAM has run Orthos 8 hours + on this board with exactly the same 45min shutdown.

Craig.

[Jobeo]

BTW, Jobeo, nice OC on air!

What are your voltages (CPU, NB, SB, VTT)?

tx! nb, sb, and vtt are all stock settings you'd get without any tweaking. the CPU VID is the only thing i had to change its around 1.38v and drooping at idle to 1.34

don't get me wrong i dont think i'm going to win any overclocker of the year awards (unless its the noob oc'er of the year lol).... but, as you related in your GTL experience from last night, i find dealing with the NB, VTT, GTL's a major headache and very frustrating. the board is "extreme" as in extremely finicky when it comes to balancing those things.

like i said with the lower FSB the cores will bottleneck at the MCH if they all start accessing the ram at once. but for now i'm satisfied. i'd like to get my memory clocked up a bit and we'll see how that goes. when i get my stamina up a bit i'll probably start fidgeting with the FSB again but i'm tired of banging my head against the wall.

[virtualrain]

tx! nb, sb, and vtt are all stock settings you'd get without any tweaking. the CPU VID is the only thing i had to change its around 1.38v and drooping at idle to 1.34

Ahh... the benefits of an unlocked multi... Luxury! You have no idea of the frustration level you would be under if you were trying to get that clock stable on a quad at x9 multi!