DFI LANParty UT NF680i LT SLI-T2R - Mini-Review

[Raja@ASUS]

Moved to proper thread http://www.xtremesystems.org/forums/...d.php?t=146441

[Pedro Rocha]

Sorry, but no more dicussions about stability / stress prime will be allowed in this thread or other thread in XS Reviews

For those you want dicuss what is stable system and if stress prime is the best method I create this thread http://www.xtremesystems.org/forums/...d.php?t=146441 on XOC Q&A, please express your opinions there.

If you want post your comments about this Review or the DFI LANParty UT NF680i LT SLI-T2R, feel free to do it here

Thank you

[Mekrel]

Any chance of some Raid 0 performance benchmarks with HDTach etc?

Thanks

[Pedro Rocha]

Any chance of some Raid 0 performance benchmarks with HDTach etc?

Thanks

HDTach with single HD sure

RAID no, I don't use RAID for many years, I'm to old and lazy for frequent HD formats or backups

Try to find that info other less Overclock oriented review / post , I'm sure you will find it..

[Pedro Rocha]

I have done some memory bandwith test and you may find suprised by the results..

First some important findings:

- Always use yellow slots for much higher bandwith -
- If you use the orange slots with 1T you can OC the memory a bit more (5 to 10Mhz) you will loose on bandwith
- At least in the orange slots if you use more than 2.55v real VDim (around 2.67 set in bios) the board will only detected 1 stick of DDR2 (see capture above)
- As always bandwith depenpends on cup multiplier / strap - for my QX6700 the sweat spot seems to be 11x - do some testing with your cpu by comparing everest mem read.

And now some bandwith

DDR 1160 4-3-3-5 2T


DDR 840 3-3-3-3 1T


I never expetect to see this results, much lower speed with tight beat higher speed with more relaxed timmings

This board needs a lot of testing to find the sweat spot for your cpu / memory so be prepared to spend a few hours

[Raja@ASUS]

the yellow slots were originally stated for 2t clocking to me from Tony, I never checked actual bandwidth performance though, might be worth checking out..

[hipro5]

Pedro, do you have a Core 2 Duo CPU to test how this mobo doing at 1:1 and Cas 3-3-3-x and 1T?.......Thanks M8......

Also how hot is the Digital PWR when working the QUAD @ 3800MHz+...

[Solarfall]

wow really good ram timings with 1T... btw very nice mini revew Pedro Rocha

[Pedro Rocha]

Pedro, do you have a Core 2 Duo CPU to test how this mobo doing at 1:1 and Cas 3-3-3-x and 1T?.......Thanks M8......

Also how hot is the Digital PWR when working the QUAD @ 3800MHz+...

Yes sure, I will test this weekend

Hot of course, but I put a fan over Digital PWR (not a Delta like yours ) - with 40mm fan I would say around 60ºC, is hot but nothing compared with the P5K Dlx cpu voltage regulators than can reach more than 100ºC

For OC, and much specially with Quad Cores, active cooling is almost mandatory on the Digital PWR - se photo at Post #2 http://www.xtremesystems.org/forums/...15&postcount=2

[Pedro Rocha]

Pedro, do you have a Core 2 Duo CPU to test how this mobo doing at 1:1 and Cas 3-3-3-x and 1T?.......Thanks M8......

Your wishes are orders to me , see Core 2 Duo results at post http://www.xtremesystems.org/forums/...15&postcount=2

I suspect you will like to see that numbers

BTW: dam hot here today (near 30ºC) seems that I am in Greece, think I have to go out to buy some Delta Fans

[hipro5]

Your wishes are orders to me , see Core 2 Duo results at post http://www.xtremesystems.org/forums/...15&postcount=2

I suspect you will like to see that numbers

BTW: dam hot here today (near 30ºC) seems that I am in Greece, think I have to go out to buy some Delta Fans

Hmmm......ONLY at up to 420MHz?.....

DELTAs for ever....

[Pedro Rocha]

Hmmm......ONLY at up to 420MHz?.....

DELTAs for ever....

Absolutely not

[hipro5]

Absolutely not

Ahhhh......NOW you're talking....

[Pedro Rocha]

As you can see in the photo above I mounted the DFI 680i in the cascade, altought it really needs to recalibrate (drop to -78ºC or more in load) I can have an idea how it compares to a fully moded (vcore, vdrop, vmem) EVGA 680i with serious cooling.

Here is the photo:


After boot up, I noticed that I can not use the 11x with 1:1 - that is really bad since 11x will be the sweat spot for 450Mhz 3-3-3..1T benching

In terms of overclok I would say it overclocks approximately less 100Mhz than the full mod EVGA680i in the same conditions - so the Vdrop mod should be realy needed for extreme overclocking.

Anyway here are the results:




And of course I have to run my favorite game - altought for now with a Single Card fully stock, here are the results



The 2 Lobbies (specially Loby High) and the Car High are quite good for less than 4.900Mhz


I will do some SLI runs later..

Also I think I need to get some post, and discover how to do VDrop mod..


Anyway this was not to bad for fully stock motheboard

[Mekrel]

Thanks for the review, especially the memory findings.

So even with an added 5/10 Mhz at the same timings, the orange slots will be slower than the yellow?

Just came into stock in the UK and I have an itchy finger

Was looking forward to watercooling the NB, but the hooks make it a pain. I think the kind of overclock I'm looking for (on water) means the default heatsink with a very quiet fan will sufice either way.

I'll go look for Raid performance

[Radamanthys]

maybe the Raid is badly compared with the ICHXR . The matrix is absolutely amazing :O .

Very nice board!! Thanks for the mini review

[Pedro Rocha]

Here is a quick SLI run with the 8800 ultras full stock



I think there is margin for improvement, specially after the vdrop mod, anyway not to shabby for a quick run in a HD with Windowd recent installed (some tweaks missing there)

As forecasted 3-3-3...1T make wonders in 3DMark2001, hope the next Bios alllow me the full use of the 11x multiplier to have some 1:1 450Mhz 3-3-3...1T action

[virtualrain]

Pedro,

I ran my memory in both the orange and yellow slots and checked bandwidth in Everest and SiSoft... here are my results.



Comments:

• Strangely, I could not get the board to POST with the RAM running at 1T in the yellow slots.
• At 1100MHz running 2T command rate, there was no difference between the yellow and orange slots with my memory. What am I missing?
• Oddly, when switching RAM between slots, the board requires a CMOS clear to POST. Then after adjusting settings (loading from cmos reloaded profiles for example) and rebooting, it would hang on POST code #75 for 2 minutes before rebooting normally. I'm not sure why it does this, any thoughts?.
• 1T is marginally faster at 800MHz than 2T at 1100MHz (at least according to Everest... Sisoft Sandra has 1T slightly slower).

[I)ickie]

Pedro,

I ran my memory in both the orange and yellow slots and checked bandwidth in Everest and SiSoft... here are my results.

*SNIP*

Comments:

• Strangely, I could not get the board to POST with the RAM running at 1T in the yellow slots.
• At 1100MHz running 2T command rate, there was no difference between the yellow and orange slots with my memory. What am I missing?
• Oddly, when switching RAM between slots, the board requires a CMOS clear to POST. Then after adjusting settings (loading from cmos reloaded profiles for example) and rebooting, it would hang on POST code #75 for 2 minutes before rebooting normally. I'm not sure why it does this, any thoughts?.
• 1T is marginally faster at 800MHz than 2T at 1100MHz (at least according to Everest... Sisoft Sandra has 1T slightly slower).

If the difference in 1T overclocking on the yellow slots is worth Tony saying something my guess is that there is some difference between the treatment of these pairs of slots and the chipset. This would explain the lack of boot in the yellow slots. seem to have lost the link but i too heard tonys statement about the orange slots being the ones to use for 1T. What do you think the chances are of running this board with 4x1gb at 1T?

[virtualrain]

If the difference in 1T overclocking on the yellow slots is worth Tony saying something my guess is that there is some difference between the treatment of these pairs of slots and the chipset. This would explain the lack of boot in the yellow slots. seem to have lost the link but i too heard tonys statement about the orange slots being the ones to use for 1T. What do you think the chances are of running this board with 4x1gb at 1T?

I think you answered your own question ...with my experience not being able to POST with 1T in the yellow says to me it's going to be damn near impossible to run four sticks at 1T on this board.

BTW, Pedro, if you are around, I'm still interested in understanding how you get more bandwidth from the yellow slots.

[I)ickie]

I think you answered your own question ...with my experience not being able to POST with 1T in the yellow says to me it's going to be damn near impossible to run four sticks at 1T on this board.

BTW, Pedro, if you are around, I'm still interested in understanding how you get more bandwidth from the yellow slots.

Naive hope that perhaps the board would exhibit different behavior when all four slots were loaded.
Thanks for the reply

[Pedro Rocha]

BTW, Pedro, if you are around, I'm still interested in understanding how you get more bandwidth from the yellow slots.

I'm in my laptop now, think I have some screen capture that evidence the diference in my bench disk, altought small there is diference.

What sticks you have ??

With my Corsair - 8500C5 and 6400C3 - I can run at 1T with no problem at all in booth the EVGA and 680i with no problem at all.

Also remember a basic law of physic - less cycle (1T) need much more voltage .. for 410 3-3-3-1T, 2.40V in Dimms is the miminum for 1T benching ..of course my 450 3-3-3 1T (see above) I was using around 2.6V.

Two things you guys should remember.
- 2.6 or even 2.5v should be only use for benching - NOT to 24/7
- Always use active cooling over the DDR2 sticks with higger voltages.


A final note regarding some comments from you :

My reviews are Overclocked Oriented Reviews - they are not by any mean intend to reproduce a 24/7 use environment!!
If you want stock level or moderate OC reviews, please look elsewhere, here in XS or all other over the internet.. there are hundreds for stock level or moderate OC reviews.

I do these reviews on pure hobby basis, I am not and I am not intended to be a professional reviewer.

I just want to leave these 2 things quite clear

[virtualrain]

Pedro, good grief, I'm not trying to be difficult for you in any way. I don't know why you are defensive... what did I say that offended you?

BTW, I think hobby reviews are the most valueable and I've found your work here interesting and I've posted a link to your review and some of your tips on my blog to share with others with full credit to you of course. I mean no disrespect... let me be clear about that! (I may be jealous of the resources you have but that's the extent of my ill feelings! )

Anyway, back to this board...

When you said that you could get more memory bandwidth from the yellow slots I wasn't sure whether that meant that clock for clock the yellow slots were faster or you could get higher clocks from the yellow slots. Can you elaborate?

As you can see a few posts up, your comments encouraged me to try both slots to see if I could improve my memory performance in anyway. The results speak for themselves... they were statistically the same for me at 2T and 1T would not POST in the yellow slots.

I'm using Mushkin PC2-8500 2x2GB (rated for 5-4-4-12 at 1066) which is a very good kit of 4GB... maybe the best... but still no match for 2x1GB. I've found this kit very stable at 2.3V at 800MHz 4-3-3-8 1T on this board for 24-7 use. I could run it as high as 900MHz 1T which is amazing for 2x2GB sticks, but I'm FSB limited on my quad and find that I get the best memory performance running linked/synced as in my sig.

Cheers!

[Pedro Rocha]

Hi virtualrain

Sorry I lead you to misunderstand me, neither I'm at defensive, neither you offend me.

This a general observation for all used, not for you in particular

Back on topic : I only compared 1T speed versus 1T speed on yellow / orange slots. Didnt even compared 2T speeds.

I can do some test with 2T, maybe with 2T results are similar.

But for the me the main advantages of the DFI over the EVGA is the much higger overclock you can achieve in 1T with lower timmings.

[virtualrain]

Back on topic : I only compared 1T speed versus 1T speed on yellow / orange slots. Didnt even compared 2T speeds.

I can do some test with 2T, maybe with 2T results are similar.

But for the me the main advantages of the DFI over the EVGA is the much higger overclock you can achieve in 1T with lower timmings.

Ok... I see what you are saying about yellow/orange now. I'm still perplexed as to why I couldn't even post in the yellow slots at 1T

BTW, I don't think testing with 2T on this board is worthwhile, as you say, 1T is where the performance is at on this board.