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

memtest and prime are 2 entirely different beasts, and passing in one will not guarantee passing with the other. Just stressing one area of the board at a time never guarantee's stability upon simultaneous load of CPU, memory and pcie lanes. I don't run prime anymore, as I can generally dial in where I can run 24/7 without it, part of what put me off was looking back over prime history to realise the code has needed slowing down or speeding up based upon new topologies, and at times the changes were not made fast enough. That being said the shutdown issue is something I have seen myself, your best option is just to OC that processor by feel of what you run day to day, once you have been oc'ing for a while, it's not too difficult to do..

...so; what's the verdict on quad core 680i kentsfield overclock on air, or watercooling?

for benching fsb's of 450 are certainly attainable, for 24/7 use, I would be aiming around 400-420 and use 3d/games or real world encoding software to find where I can run my system..

i reckon 400x8 on air for kentsfield..unless you get a miraculous cpu that clocks high on low v.

...those ones i never get :rolleyes:


looks like there is more tweaking to be done on dfi680i

Quote:

---

Originally Posted by raju

memtest and prime are 2 entirely different beasts, and passing in one will not guarantee passing with the other. Just stressing one area of the board at a time never guarantee's stability upon simultaneous load of CPU, memory and pcie lanes. I don't run prime anymore, as I can generally dial in where I can run 24/7 without it, part of what put me off was looking back over prime history to realise the code has needed slowing down or speeding up based upon new topologies, and at times the changes were not made fast enough. That being said the shutdown issue is something I have seen myself, your best option is just to OC that processor by feel of what you run day to day, once you have been oc'ing for a while, it's not too difficult to do..

---

I don't know if we are on the same page here, but I'll try and explain again why i don't believe it is memory. Memory is running as we speak with the E6600 at 950 4-3-3-6-2t 2.2v with prime running in background. I have used this same memory on 6 EVGA's, 2 Stikers, 2 Gigabytes 680i's, 2 Abit 680's and a foxconn 680i lt. Never any shutdowns. I stated in previous post it even shuts down at 700mhz. It's sold as 1200mhz memory so 700 is way under clocked. It also shuts down with the mushkin 2x2gb kit installed. I've used that memory in half of the above boards with never any issues. It shuts down at default settings as well, that is no overclock on memory or cpu. It runs memtest86 at better than rated timings with no errors. If i was basing stability on just one of the above statements i could understand why you would say it could be memory. but taking all of the above statements in to consideration, I can say with certainty that memory is not causing the shutdowns.

Quote:

---

Originally Posted by Roger_D25

Nugzo/Bootsy,
This is just a suggestion but is there any chance that the PSU is playing an issue when the Quad Core is installed? I realize that sounds crazy given your Thermaltake supply/Silverstone supply but as many have already realized the boards are a bit finicky when it comes to PSU's? Plus an issue with the power supply might cause the type of issues your having (shut down)? With that said what supplys are you guys using that have successfully run Quad Cores on this motherboard? I can imagine how frustrating thsi would be so I'm just tying to help out! Good luck guys and keep us posted on progress!

---

Thanks for the input Roger. I intially thought it was a power supply issue after reading some threads at DFI club but I can fun my X6800 on this board with the same power supply at close to 500fsb priming all day. Last weekend I tested 3 different boards http://www.xtremesystems.org/forums/...d.php?t=143401 with my quad. So far I havent seen any success stories with quads on the new DFI boards.

Quote:

---

Originally Posted by nugzo;

I don't know if we are on the same page here, but I'll try and explain again why i don't believe it is memory. Memory is running as we speak with the E6600 at 950 4-3-3-6-2t 2.2v with prime running in background. I have used this same memory on 6 EVGA's, 2 Stikers, 2 Gigabytes 680i's, 2 Abit 680's and a foxconn 680i lt. Never any shutdowns. I stated in previous post it even shuts down at 700mhz. It's sold as 1200mhz memory so 700 is way under clocked. It also shuts down with the mushkin 2x2gb kit installed. I've used that memory in half of the above boards with never any issues. It shuts down at default settings as well, that is no overclock on memory or cpu. It runs memtest86 at better than rated timings with no errors. If i was basing stability on just one of the above statements i could understand why you would say it could be memory. but taking all of the above statements in to consideration, I can say with certainty that memory is not causing the shutdowns.

---

so look else where, psu, vga, timings are a good start...remember just because you could run 1 way on 1 board does not mean it will run the same on another. Also are you running the memory in a different way..IE are you running linked +sync and you were running unlinked on the other boards?

Quote:

---

Originally Posted by Tony

so look else where, psu, vga, timings are a good start...remember just because you could run 1 way on 1 board does not mean it will run the same on another. Also are you running the memory in a different way..IE are you running linked +sync and you were running unlinked on the other boards?

---

I hope it is not PSU. I have $800 in 2 power supplies. I just pulled a $400 enermax that works perfect on everything else, and installed a new $400 Thermaltake 1200. I 'll pull this board and install the EVGA today and try it with the new thermaltake power supply. I always run linked/sync. It shuts down linked/sync and auto/optimal. I never even changed the mem timings with the quad, left them at optimal. I ran each test 15x loop on 3dmark06 with the E6600@ 3800 linked/sync (1900/950) 4-3-3-7-2t . Scored 16,650. I'll be glad to try any tests recommended. Right now i'm taking the DFI out and installing the EVGA to make sure its not a PSU issue.

Quote:

---

Originally Posted by nugzo

I hope it is not PSU. I have $800 in 2 power supplies. I just pulled a $400 enermax that works perfect on everything else, and installed a new $400 Thermaltake 1200. I 'll pull this board and install the EVGA today and try it with the new thermaltake power supply. I always run linked/sync. It shuts down linked/sync and auto/optimal. I never even changed the mem timings with the quad, left them at optimal. I ran each test 15x loop on 3dmark06 with the E6600@ 3800 linked/sync (1900/950) 4-3-3-7-2t . Scored 16,650. I'll be glad to try any tests recommended. Right now i'm taking the DFI out and installing the EVGA to make sure its not a PSU issue.

---

Nugzo, perhaps you can post your BIOS settings with the quad... I'm going to finally get to OC'ing mine this evening and I can perhaps replicate the problem... or help in confirming that it's something particular to your setup.

I took the DFI out and installed EVGA today. 450*8 3600mhz 1.47v with the Quad. Prime for 2hours so far with no shutdowns. 18,100 3DMark 06. I guess I'm keeping this board for a while. Atleast untill the MSI Diamond makes it out. If it makes it out. I'll probably try the DFI again in a month or 2 after the bios matures. The EVGA wins again. Effortless Overclocking. Extremely Stable.

Quote:

---

Originally Posted by nugzo

I took the DFI out and installed EVGA today. 450*8 3600mhz 1.47v with the Quad. Prime for 2hours so far with no shutdowns. 18,100 3DMark 06. I guess I'm keeping this board for a while. Atleast untill the MSI Diamond makes it out. If it makes it out. I'll probably try the DFI again in a month or 2 after the bios matures. The EVGA wins again. Effortless Overclocking. Extremely Stable.

---

Crap! I assume you were shooting for similar OC on the DFI with similar settings? I'm not pleased to hear this. I hope I have better luck but now my expectations are low.

Does anyone know what EVGA did exactly on the A1 rev of those boards to increase Quad overclocking? I thought it was a GTL ref mod... if so, why can't we simply duplicate that on the DFI since it has customizable GTL?

Since Nugzo having problems, what powersupply do you guys recommend> something with a HIgh Amperage single 12v rail(70a) like the Silverstone Decathalon 850w? Going to run quadcore and duel GTX's.. Was going to get the Thermaltake 1000w toughpower but seems like thats out of the question now.

Anybody have an ETA on this board ? :hump:

Quote:

---

Originally Posted by Colossous

Anybody have an ETA on this board ? :hump:

---

I assume you are talking about the UK? Have you tried contacting the DFI European office? As for NA... you probably know it's been available through Motherboardpro.com for about a week and eWiz apparently started selling it a couple of days ago.

Quote:

---

Originally Posted by vincey2kr1

Since Nugzo having problems, what powersupply do you guys recommend> something with a HIgh Amperage single 12v rail(70a) like the Silverstone Decathalon 850w? Going to run quadcore and duel GTX's.. Was going to get the Thermaltake 1000w toughpower but seems like thats out of the question now.

---

Not sure... All we know for certain is that an Enermax will not allow the board to post. I don't think it's been determined that the PSU is the problem with Nugzo's situation.

mine works find with 1k watt dxx guys donno what hes doing

Using what powersupply?

Quote:

---

Originally Posted by momomo6789

mine works find with 1k watt dxx guys donno what hes doing

---

I'm afraid i do know what i'm doing. I build, rebuild, or repair 30-50 computers a week between my 2 repair stores.Call DFI and ask them if there are any issues with enermax PSU's and DFI Boards.

Quote:

---

Originally Posted by nugzo

I don't know if we are on the same page here, but I'll try and explain again why i don't believe it is memory. Memory is running as we speak with the E6600 at 950 4-3-3-6-2t 2.2v with prime running in background. I have used this same memory on 6 EVGA's, 2 Stikers, 2 Gigabytes 680i's, 2 Abit 680's and a foxconn 680i lt. Never any shutdowns. I stated in previous post it even shuts down at 700mhz. It's sold as 1200mhz memory so 700 is way under clocked. It also shuts down with the mushkin 2x2gb kit installed. I've used that memory in half of the above boards with never any issues. It shuts down at default settings as well, that is no overclock on memory or cpu. It runs memtest86 at better than rated timings with no errors. If i was basing stability on just one of the above statements i could understand why you would say it could be memory. but taking all of the above statements in to consideration, I can say with certainty that memory is not causing the shutdowns.

---

1. Which BIOS? The differences between 0503 and 0504 are significant.

2. Do you have the X-FI installed? If so, remove it and see what happens.

3. Did you do a clean OS install? I have noticed the same issue moving from one 680i board to another.

Quote:

---

Originally Posted by bingo13

1. Which BIOS? The differences between 0503 and 0504 are significant.

2. Do you have the X-FI installed? If so, remove it and see what happens.

3. Did you do a clean OS install? I have noticed the same issue moving from one 680i board to another.

---

Is the 0504 the bios to install :confused:

Quote:

---

Originally Posted by bingo13

1. Which BIOS? The differences between 0503 and 0504 are significant.

2. Do you have the X-FI installed? If so, remove it and see what happens.

3. Did you do a clean OS install? I have noticed the same issue moving from one 680i board to another.

---

0504 for quad. Yes i do have a X-Fi installed. I did do a clean install. I never tried it with the X-Fi out. I also had an Areca 1210 in lowest PCI-X slot.

Quote:

---

Originally Posted by virtualrain

Not sure... All we know for certain is that an Enermax will not allow the board to post. I don't think it's been determined that the PSU is the problem with Nugzo's situation.

---

I would definitely not say that is a certainty. I have had absolutely zero problems from the first boot on using an Enermax 720 Watt Infiniti.

my full retail 680lt is arriving soon I will run some tests and report back to tech any issues I find..

Nice board:) Gotta get one.

Quote:

---

Originally Posted by century child

I would definitely not say that is a certainty. I have had absolutely zero problems from the first boot on using an Enermax 720 Watt Infiniti.

---

True... all we know with certainty is that Nugzo's 1000W Galaxy wouldn't allow him to post.

so how's everybody coming along on their quad-clocks ?? on meh qx6700 i'm having the most success with 1:1 memory (2T). i run into trouble trying link + sync 1T memory mode dropping the multiplier to 8 and raising the FSB.

anyone having good success with link+sync and quads with ~400 FSB ? i can't seem to get close to 400 without something going wrong =/

Quote:

---

Originally Posted by nugzo

0504 for quad. Yes i do have a X-Fi installed. I did do a clean install. I never tried it with the X-Fi out. I also had an Areca 1210 in lowest PCI-X slot.

---

Can you pull the X-FI and Areca out and try it? I had an issue with the X-FI in testing. Also, the 0504 BIOS completely changed the way the board handles memory now. Forget about sync and 1T being stable around 900~950 with it. It appears they have gone back to high DDR2 speeds with 2T being optimal (not really if you know what I mean). Run sync and 2T with tight timings down low or unlink with 2T and increase your RAM speed to get the best performance out of the 0504 release.

I'm trying to find a good stable overclock on the CPU before I go anywhere with the RAM (it's underclocked at the moment).

What GTL settings are good for Quad's?

Right now I'm trying Tony's settings...

Quote:

---

A sweet spot you may want to try is VTT 1.4 to 1.44V...GTL1=170, GTL2 95 and NB GTL 85

---

With high Vcore and NB Voltage.

I notice that if I set 1.535V in BIOS, I get a reading of 1.5 in Windows, and it droops to 1.47 under load.

Quote:

---

Originally Posted by virtualrain

What GTL settings are good for Quad's?

---

I´m at VTT: 1.40V GTLs: 180/100/100 right now with a X3210, GTLs need a lot of tweaking ... as not 1 CPU behaves the same (i have some 3210 and 322ß here ...).

NB 1.48V
VCore 1.45V in Bios (more than needed for sure ...)

8x413 stable and aiming for more right now :D

Quote:

---

Originally Posted by virtualrain

I'm trying to find a good stable overclock on the CPU before I go anywhere with the RAM (it's underclocked at the moment).

What GTL settings are good for Quad's?

Right now I'm trying Tony's settings...



With high Vcore and NB Voltage.

I notice that if I set 1.535V in BIOS, I get a reading of 1.5 in Windows, and it droops to 1.47 under load.

---

that gtl table is direct from Oskar, aim for the 1.4 vtt mark, though setting each range will give almost the same results, as you are keeping the same variance level to give the same effect overall....


as said by pete it is quite variable, for me 168/96/85 worked well with vtt 1.41, this is how the 467 fsb run came into play... again tune it for your cpu.



Obviously bios updates bring changes as Bingo has already pointed out. I will definitely be keeping a copy 0503 for benching...

regards
Raja

When tweaking GTL, what approach should I take?... is there any rhyme or reason to it? I guess I'll have to study the articles on TheTechRepository, but if someone has some recommendations on the best way to tweak, I'm all for it.

One of my problems is that I'm now stable (Windows, Super PI, etc.) except for PRIME... so testing anything for stability takes a long time. I can go anywhere from 15min to an hour before PRIME errors (usually on core 4). Thus it will take me eons to tweak every darn setting to find ultimate stability.

Quote:

---

Originally Posted by virtualrain

One of my problems is that I'm now stable (Windows, Super PI, etc.) except for PRIME... so testing anything for stability takes a long time. I can go anywhere from 15min to an hour before PRIME errors (usually on core 4). Thus it will take me eons to tweak every darn setting to find ultimate stability.

---

That`s a Overclockers Life .... Testing ;)

VTT 1.4 GTL 1750/95/85 should be a good start, also - you did not mention your actual GTL/VTT Settings ...

Try the settings and see how Core 4 behaves, normally when 1 Core crashed here i`ll add the corresponding GTL +2 and redo the testrun (in that case 170/97/85). Step by step you get closer to the sweetspot - but it takes some time :rolleyes: as every CPU is a bit different.

Quote:

---

Originally Posted by virtualrain

When tweaking GTL, what approach should I take?... is there any rhyme or reason to it? I guess I'll have to study the articles on TheTechRepository, but if someone has some recommendations on the best way to tweak, I'm all for it.

One of my problems is that I'm now stable (Windows, Super PI, etc.) except for PRIME... so testing anything for stability takes a long time. I can go anywhere from 15min to an hour before PRIME errors (usually on core 4). Thus it will take me eons to tweak every darn setting to find ultimate stability.

---

I use OCCT 30 mins test, I find this just as good, actually better than prime, as it is a quick hit. The 30 min test is good enough for me if it passes, generally everything I throw at the pc runs..

Rhyme or reason is in the Gtl article by Kris as you thought. With regards to formula's Gtl is in the 2/3 VTT ballpark as Kris has told, the rest is down to onboard resistor/component variance and tolerance levels, including thermal drift, system specific rfi/emi etc, not forgetting individual cpu variance...

regards
Raja

Quote:

---

Originally Posted by bingo13

Can you pull the X-FI and Areca out and try it? I had an issue with the X-FI in testing. Also, the 0504 BIOS completely changed the way the board handles memory now. Forget about sync and 1T being stable around 900~950 with it. It appears they have gone back to high DDR2 speeds with 2T being optimal (not really if you know what I mean). Run sync and 2T with tight timings down low or unlink with 2T and increase your RAM speed to get the best performance out of the 0504 release.

---

LOL, this board is looking more and more like a no go for me... first of all my Galaxy PSU looks like its got compatibility issues, and now im hearing theres an issue with the X-fi! Typical...

Ok, I got an OCCT stable overclock using what I would call a brute force methodology of going with aggressive voltages. My next step will be to scale back voltages to see what the OC is sensitive to so that I know what to emphasize when I drive higher...

I'm interested in your opinions on what voltages may be excessive at this point. (see below)

Details:
- X3220 at 3600 MHz (400x9)
- Voltages in screen shot below
- Cooling CPU on water (load temps 58-63-deg C) and NB via fan
- No OC on memory (using link/sync and optimized so memory is at 800 5-5-5)

http://members.shaw.ca/virtualrain/RainmakerOC1003.JPG

Note the differences between settings and BIOS measurements.

Also, I found that in Windows under load, the Vcore would droop as low as 1.46V from this value!!!

More screen shots here.

Quote:

---

Originally Posted by virtualrain

I'm interested in your opinions on what voltages may be excessive at this point. (see below)

Details:
- 3600 MHz (400x9)
- Voltages in screen shot below

---

NB Core looks a bit tooo high for 400FSB, right now i use 1.48 for 8x400 and 8x413 without a Prob. - but it´s an approach. If it works with these settings - start to lower the voltages notch by notch and rerun OCCT 30 Min test and work your way down ;)

Correction: Although it passed OCCT, it failed Prime. I guess I just got lucky. I assume OCCT uses similar code base to Prime?

someone needs to test a seasonic power supply on this board to see if they have any issues.

Quote:

---

Originally Posted by safan80

someone needs to test a seasonic power supply on this board to see if they have any issues.

---

I'm using a Corsair which I recall reading is OEM'd from Seasonic. No problems with it so far.

Quote:

---

Originally Posted by MikeMK

LOL, this board is looking more and more like a no go for me... first of all my Galaxy PSU looks like its got compatibility issues, and now im hearing theres an issue with the X-fi! Typical...

---

perhaps get one of the other 680i LT boards.. (foxconn variants, eVGA or XFX) I hear they rock with quads and don't have any compatibility issues with xfi that I know of.. I have a friend who just got one to replace his Abit 32x max and he's running his qx6700 at 400 on the fsb all day no problems.. says it'll do upwards of 440 stable... pretty nice. anyways..

edit: oh crap.. I thought you were quad'ing it.. hehe.. nm then.. but still and all, those board look like nice ocers.

Quote:

---

Originally Posted by revenant

I have a friend who just got one to replace his Abit 32x max and he's running his qx6700 at 400 on the fsb all day no problems.. says it'll do upwards of 440 stable... pretty nice. anyways..

---

ask him for a CPU-z screenshot.

better than cpu-z would be an actual 3d bench...

Quote:

---

Originally Posted by raju

better than cpu-z would be an actual 3d bench...

---

ok a 3d bench with cpu-z ;)

Hi and I hope someone is looking at the boards today,

I am now gone to install the DFI 680i, when it comes to chip set heat sinks I am a noob. The owners manual comes with a supplemental page on how to install the heat sink but it states the NB already comes with thermal paste already on it, and they tell you not to spread it out as the heat sink will take care of that. Well it does not come with paste already on it or at least mine did not, so my question is do I just use Artic Silver and if so how much and do I spread it out covering the chip or should it be a rice size drop:confused:

Thanks for any help.

you should have a tube of paste with the heatsink to use...

If not use arctic ceramique in that area as it is non conductive/capacative.....

regards
Raja

Quote:

---

Originally Posted by raju

you should have a tube of paste with the heatsink to use...

If not use arctic ceramique in that area as it is non conductive/capacative.....

regards
Raja

---

God I'm so stupid:fact: Thanks you are so right. Should I use just a drop or spread a thin layer?

Quote:

---

Originally Posted by revenant

perhaps get one of the other 680i LT boards.. (foxconn variants, eVGA or XFX) I hear they rock with quads and don't have any compatibility issues with xfi that I know of.. I have a friend who just got one to replace his Abit 32x max and he's running his qx6700 at 400 on the fsb all day no problems.. says it'll do upwards of 440 stable... pretty nice. anyways..

edit: oh crap.. I thought you were quad'ing it.. hehe.. nm then.. but still and all, those board look like nice ocers.

---

Hi rev... I will be quad'ing it shortly... getting either a Q6600 or QX6700 to play around with ;) Thats y im looking at other boards atm, as well as the fact my IN9 just crapped out on me and is @ RMA.

Really from what ive seen its between the EVGA or the DFI... personally i would prefer the DFI as Ive always had good results with the various DFI's Ive had, but obviously its a no go if there are compatibility issues with my other hardware.

Quote:

---

Originally Posted by palese

God I'm so stupid:fact: Thanks you are so right. Should I use just a drop or spread a thin layer?

---

I would use a good sized blob in the middle, as the instruction leaflet says the pressure from the heatsink will auto spread the paste for you..

Quote:

---

Originally Posted by raju

I would use a good sized blob in the middle, as the instruction leaflet says the pressure from the heatsink will auto spread the paste for you..

---

Thanks raju:)

If you are interested, I posted some pics of my bench water cooling setup (actually on the dining room table) in this thread over in the water cooling forum... http://www.xtremesystems.org/forums/...d.php?t=144079

looks like dfi really needs new bios....to state the bleeding obviarse.
the evga with p28 looks like a no brainer (for quad oc) according to nugzo's experience.

hey VR thats some nice info you're accumulating over on your blog.

at the moment i've become very gunshy about overclocking this board. i was tweaking bits and pieces here and there and not getting close to the voltages you posted a bit ago in your screenshot.

something went wrong and the BIOS on boot was telling me the system was in safe mode and to check the CMOS. i loaded settings i had saved which i knew worked and it kept saying it was booting in safe mode. so i went and loaded optimized defaults, everything at stock speeds and it still reported it was in safe mode on boot.

so i figured something was wrong with the BIOS and i should flash it again to make sure. the flash went fine, it completed and told me to reboot. after rebooting i got a blank screen. nothing. i powered down and when i turned it back on a horrendous beep was coming from the rig. the post code showed "02" which is a reserved bios code. i tried clearing the CMOS by holding down reset/power for a few seconds, that didn't help. tried the cmos clear jumper and that didn't help either.

to make a long story short (if possible) i had to remove the CMOS battery for a few minutes to get the thing up and running again. i'm not sure why all this happened but i'm glad its working now even if its just at stock. i'll need to build up some courage before trying again =/

I guess I take a different approach to overclocking than most... I basically crank everything up to safe maximums and see what the processor will do and then scale things back one at a time... that's why my initial OC settings seem like really high Voltages... they are! :D

It turns out the max my X3220 will do under coarse adjustments right now is about 3.5 GHz (390x9) - I was hoping for 3.6 and perhaps that will come with some fine tuning.

I'm now in the process of scaling back my voltage settings to see what the minimums are to keep it stable at 3.5GHz and then I'll see if I can tweak higher.

I've found the BIOS to be pretty solid... if a setting is off and it won't post, the easy reset works for me. Most of my problems have been in getting Prime stable which is time consuming. I've found that using small FFT's is the most taxing on the CPU and if it's going to fail Prime, it will do so on this test in about 5-10min. The blend test takes longer because it doesn't hit the small FFT's until about 10 min. in.

Sometimes when the board fails, it fails hard!!! As Nugzo reported, it will simply shut itself off. I found this happened most often if I was pushing voltages to the extreme (like CPU VTT at 1.575 or so).

I hope to have some quad settings to share tomorrow, but I'm still confused about the GTL settings. They don't make any sense why there is a "CPU GTL 1" and a "CPU GTL 0/2/3"... the articles I've read seem to imply that 0/1 and 2/3 are ususally tied together and should have close to the same value. With the board, the recommended value for GTL 1 is twice the value of the 0/2/3 value. :confused:

The GTL options are labelled wrong in bios.

Option 1 is core 0 and 1
Option 2 is core 2 and 3
Option 3 is the NB

There is as far as I know no way to have the GTL voltages working how thwy are labelled in bios.

If i get time I will alter the wording with modbin so it coorect reflects what is going on

Is it true that GTL 0/1 should be twice the value of GTL 2/3? This doesn't make sense to me either unless the voltage bias is skewed to begin with.

We need some frame of reference like this in order to intelligently tweak... http://csd.dficlub.org/tikiwiki192/t...page=GtlRef&bl

Quote:

---

Originally Posted by virtualrain

Is it true that GTL 0/1 should be twice the value of GTL 2/3? This doesn't make sense to me either unless the voltage bias is skewed to begin with.

We need some frame of reference like this in order to intelligently tweak... http://csd.dficlub.org/tikiwiki192/t...page=GtlRef&bl

---

your response is at the club, for further insight ask @ TTR, kris is the man for gtl..


regards
Raja

One thing I will give Evga with their revision A board, it sure is easy to overclock:fact: I ran my Xeon 3220 27/7 @ 3.2Ghz with out a hiccup, I wish I could say the same for my new DFI:slapass: Hell I just want a simple OC but nothing is simple with this board.

Yeah, I'm not a big fan of GTL... which is what makes this board so complicated.

I don't know how EVGA and Intel boards get great quad core overclocks without it, but there has to be a better way than relying on users to tweak GTL. It will be interesting to see if people with EVGA boards can match or exceed their quad clock on the DFI or if they just give up in vein and go back to their EVGA. Giving people the tools to achieve great overclocks (such as tweakable GTL) is only good if the tools are well understood and/or some guidance exists... otherwise, they are more of a hinderance. I must admit, I'm finally making some headway, but it's been a frustrating weekend with this board to be sure and none of the current guidance was much help.

I think if you can overclock a DFI board, you should get some kind of Scout's honor or badge or something because this ain't no basic rope work here boys and girls! :D :p:

Quote:

---

Originally Posted by virtualrain

Giving people the tools to achieve great overclocks (such as tweakable GTL) is only good if the tools are well understood and/or some guidance exists... :D :p:

---

OR when it can get you higher than the others by tweaking more. Otherwise when you know that you going to have the same result or less you going with the easiest solution.

Quote:

---

Originally Posted by virtualrain

Yeah, I'm not a big fan of GTL... which is what makes this board so complicated.

I don't know how EVGA and Intel boards get great quad core overclocks without it, but there has to be a better way than relying on users to tweak GTL. It will be interesting to see if people with EVGA boards can match or exceed their quad clock on the DFI or if they just give up in vein and go back to their EVGA. Giving people the tools to achieve great overclocks (such as tweakable GTL) is only good if the tools are well understood and/or some guidance exists... otherwise, they are more of a hinderance. I must admit, I'm finally making some headway, but it's been a frustrating weekend with this board to be sure and none of the current guidance was much help.

I think if you can overclock a DFI board, you should get some kind of Scout's honor or badge or something because this ain't no basic rope work here boys and girls! :D :p:

---

virtualrain do I need to change the default GTL settings just to get a mild overclock:confused: And if so does each core on the quad have it's own setting?

Right now when I bump up the FSB and clock speed to say just 3.0 the system goes into safe mode and says hit F1 if I want to continue, but when I do and it goes into windows nothing is changed:eek:

F1 will continue on in "safe mode" hence no change. It is there to allow you to recover and go into the bios and make changes.

Quote:

---

Originally Posted by palese

virtualrain do I need to change the default GTL settings just to get a mild overclock:confused: And if so does each core on the quad have it's own setting?

Right now when I bump up the FSB and clock speed to say just 3.0 the system goes into safe mode and says hit F1 if I want to continue, but when I do and it goes into windows nothing is changed:eek:

---

I'm still trying to find something that is Prime95 stable for at least a few hours but here's what was working for me...

Voltages:
-CPU VTT: 1.53x (something high around here - but not too high)
-CPU special add: Auto
-CPU Drive: Strong
-RAM Drive: Weak
-RAM Voltage: (whatever your RAM likes)
-NB Core: 1.5x (something high)
-SB Core: 1.62
-SB Dual: 1.52
-LDT Voltage: 1.22
-CPU VTT: 1.52
- GTL1: 180
- GTL0/2/3: 180
- NB GTL: 90

I know these GTL settings are totally wacked from what has been recommended by DFI and Tony, but they are the only settings that seem to provide any semblance of stability under PRIME. I choose equal values for GTL1 and GTL0/2/3 based purely out of frustration tweaking values closer to default. They seem to work better than anything else I've tried. Either I have a wacked CPU, got extremely lucky, or the implementation and guidance on GTL for this board is off... I have no idea what.

Note that at least we've had it clarified that GTL1 is actually the GTL setting for Core 0/1 and GTL 0/2/3 is the setting for Core 2/3. They are mislabled in BIOS.

Go with high voltages for your CPU and NB and choose a high FSB that you think your CPU can do. You can always scale back the voltages after you find stable GTL settings. You don't want to be fighting both low voltages and bad GTL settings at the same time in my opinion. You will find there is a large gap between what you set in BIOS and what you get (as shown in BIOS). Furthermore, there is significant droop under load.

I started at 400FSB but couldn't get it to be stable at any GTL settings (and I tried a lot!). I then lowered the FSB to 390 and got a lot more stable, particularly with the GTL settings mentioned above.

If you can't even boot windows with these GTL settings, then you should tweak them significantly to try and find some bootable values. I found that I could boot windows with a wide range of GTL settings... I just couldn't find any that were PRIME stable. For example, I could boot at Tony's recommended sweet spot and also the radical settings I have above.

Best I've done is 2hrs PRIME Stable (hadn't had a chance to test longer yet) at 390 FSB with above settings.

Use Prime95.2 on SMALL FFT to save time... I found it would error out by test #5 (about 5min.) if I had unstable settings. Every system is likely different in this regard, but you will get a feel for what will make your computer fail quickly soon enough! For long term stability testing, use the Blended variation as it will stress the NB more.

Don't overclock your RAM until you get a stable FSB overclock on your CPU. Give your RAM enough volts to keep it happy but not stress it and if you are using Sync/Linked on your FSB the RAM clock should be under your RAM's ability to perform. There's no reason to have RAM playing a role in stability while you are trying to find a stable FSB.

I really hope someone starts finding the key to GTL on quad core here soon. Post your results please!

Quote:

---

Originally Posted by OldGuy

F1 will continue on in "safe mode" hence no change. It is there to allow you to recover and go into the bios and make changes.

---

So that about says it all then, if you do not fiddle with the GTL settings then your going no where:slapass:

enjoying your beta testing fellas? lol

INstOCK @Newegg $299.99 + shipping.

Quote:

---

Originally Posted by virtualrain

I'm still trying to find something that is Prime95 stable for at least a few hours but here's what was working for me...

Voltages:
-CPU VTT: 1.53x (something high around here - but not too high)
-CPU special add: Auto
-CPU Drive: Strong
-RAM Drive: Weak
-RAM Voltage: (whatever your RAM likes)
-NB Core: 1.5x (something high)
-SB Core: 1.62
-SB Dual: 1.52
-LDT Voltage: 1.22
-CPU VTT: 1.52
- GTL1: 180
- GTL0/2/3: 180
- NB GTL: 90

I know these GTL settings are totally wacked from what has been recommended by DFI and Tony, but they are the only settings that seem to provide any semblance of stability under PRIME. I choose equal values for GTL1 and GTL0/2/3 based purely out of frustration tweaking values closer to default. They seem to work better than anything else I've tried. Either I have a wacked CPU, got extremely lucky, or the implementation and guidance on GTL for this board is off... I have no idea what.

Note that at least we've had it clarified that GTL1 is actually the GTL setting for Core 0/1 and GTL 0/2/3 is the setting for Core 2/3. They are mislabled in BIOS.

Go with high voltages for your CPU and NB and choose a high FSB that you think your CPU can do. You can always scale back the voltages after you find stable GTL settings. You don't want to be fighting both low voltages and bad GTL settings at the same time in my opinion. You will find there is a large gap between what you set in BIOS and what you get (as shown in BIOS). Furthermore, there is significant droop under load.

I started at 400FSB but couldn't get it to be stable at any GTL settings (and I tried a lot!). I then lowered the FSB to 390 and got a lot more stable, particularly with the GTL settings mentioned above.

If you can't even boot windows with these GTL settings, then you should tweak them significantly to try and find some bootable values. I found that I could boot windows with a wide range of GTL settings... I just couldn't find any that were PRIME stable. For example, I could boot at Tony's recommended sweet spot and also the radical settings I have above.

Best I've done is 2hrs PRIME Stable (hadn't had a chance to test longer yet) at 390 FSB with above settings.

Use Prime95.2 on SMALL FFT to save time... I found it would error out by test #5 (about 5min.) if I had unstable settings. Every system is likely different in this regard, but you will get a feel for what will make your computer fail quickly soon enough! For long term stability testing, use the Blended variation as it will stress the NB more.

Don't overclock your RAM until you get a stable FSB overclock on your CPU. Give your RAM enough volts to keep it happy but not stress it and if you are using Sync/Linked on your FSB the RAM clock should be under your RAM's ability to perform. There's no reason to have RAM playing a role in stability while you are trying to find a stable FSB.

I really hope someone starts finding the key to GTL on quad core here soon. Post your results please!

---

CPU VTT: 1.53x seems high is this to cover droop?

Quote:

---

Originally Posted by palese

CPU VTT: 1.53x seems high is this to cover droop?

---

Yeah, on small FFT PRIME the Vcore will droop to around 1.44V with this kind of BIOS setting... at least on my board/CPU. Use Speedfan to monitor your Vcore. Let me know if you see the same thing.

Even after setting Vcore to 1.53x in BIOS when you reboot and come back to BIOS and look at the actual reported voltage (bottom of the Voltage Control screen) I see the actual Vcore is only 1.5!

I haven't got out the DMM yet... perhaps later today.

Most Nvidia 680i boards have large vrdoop. It kinda makes oc more of a challenge because you have to set the vcore so high to get enough voltage during load. Really suxs with "B" chips which tend to get hot with more vcore. I have to set a little over 1.5v on my Abit because it drops down to 1.42v @ load.

Interesting Nemesis, I was very suprised with the small amount of V-Droop my board has (even compared to my 3x prior boards)! Although I havn't seen it posted anywhere I was able to find a CPU voltage read point (after along time of testing different spots) and with BIOS set to 1.45v (actual voltage only 1.39v) it will only droop to 1.374v underload which is better than I have seen on most of my other boards (0.016v difference when underload).

I imagine guys that if most people find their boards drooping underload there will be a mod very soon for it (especially since this is an enthusiest board and will probably be very popular).

Virtualrain,
Thanks very much for taking the time to share your results with us, that info could save someone hours of tweaking (at least to get some baseline settings)! If you have the time please keep us posted with your progress, you'll be running at 450Mhz in no time!

Quote:

---

Originally Posted by palese

So that about says it all then, if you do not fiddle with the GTL settings then your going no where:slapass:

---

well at a minimum you have to fiddle with the VTT... for stability you might have to tweak the GTL i'm not sure. but that is all for lower-end overclocks. for a decent overclock it seems both the VTT and GTL better be spot on.

i haven't read this thread in a bit... the posts recently have been very informative though. if i could pull off a 390 clock i'd be happy at this point.

i'm gonna start overclocking it again sometime this week. since my last post about the CMOS crash i'm still hesitant... having a stock system that runs seems better than having a dead board =/

Quote:

---

INstOCK @Newegg $299.99 + shipping.

---

I just ordered one from Newegg. I hope I’m not sorry I did.

I plan on using it with a X6800.

I forgot to ask, what’s latest on the X-Fi issues, if there are any?

Quick question. I have one coming in on Thursday.

Could someone measure the spacing on the NB holddown loops?

From the pictures they look identical to the spacing for the 975x chipset. I am wondering if I will be able to use my Chipset WB from my BX2.

Quote:

---

Originally Posted by The Nemesis

INstOCK @Newegg $299.99 + shipping.

---

didnt even see your post. just checked newegg and they had it so i ordered it.
ill be using it with a E6700 with mushkin redline.
also i have a x700 pro in my old rig would it be beneficial to use it as a physics card.

is anyone else having trouble getting 400fsb with an E6700 ive tried up to 1.4v on the northbridge but it gets more unstable after adding more than that. whats everyone getting with single dual core cpu's for fsb on these boards i think im doing something horribly wrong

ok got another question i have a 7900gs in the other rig would it be benneficial to pick up another 7900gs and run it in sli mode, instead of just using the x1900xtx .

Quote:

---

Originally Posted by Spazilton

Quick question. I have one coming in on Thursday.

Could someone measure the spacing on the NB holddown loops?

From the pictures they look identical to the spacing for the 975x chipset. I am wondering if I will be able to use my Chipset WB from my BX2.

---

The NB uses loops which form a square that is 50x60mm. More tips, tricks, etc. in my blog here... http://virtualrain.blogspot.com/2007...and-known.html (which is also linked in my sig)

Well, I have the P965-S and its an allright motherboard, I think I'm correct in saying this 680i LT from DFI would have the same issues with Slow Raid performance compared to the ICH8R as all nVidia based chipsets. I have an E6400 that will run orthos stable at 475x8 (3800) on the P965-S and also 544x7 (3808) on my friends EVGA 680i A1, The Dark will not allow MY rams to run beyond 900 so my stiuation is either new ram or new board, I happen to like DFI since the old 939 days, and well, I ordered this board from the Egg five minutes ago... Hope it shapes up to be as good as my friends evga for dual core..

I had a brief discussion today with Tony and Raju about quad-core overclocking on this board... here's the outcome...

I'm going to try a new approach. If you read back a page, my previous approach was to set high voltages and a reasonably aggressive FSB and try to tune GTL that way. Well as you can imagine, that's like throwing darts in the dark. I ended up with rather Bizarre GTL settings that seem reasonably stable but I'm going to try it a different way this time.

I'm going to start with some aggressive voltages again, but this time, I'm going to start at 1.4V VTT and suggested GTL values with FSB=300MHz and work up in 10MHz increments. Every time I increase it and it fails stability testing (more on that below), I'm going to tweak GTL2 upwards a bit... since the key is to get Core 2/3 "in sync" with Core 0/1. When I find stable settings, I'll add another 10MHz. This might take awhile.

If my wacky GTL settings are the outcome again, it really means my core dies are not speaking well with each other via the FSB and I likely have a less-than-ideal CPU. I tended to agree that a good part of the problem in overclocking Kentsfields are the dual-die's are both trying to use the same bus and so synchronizing levels is critical. It may take some finesse to get them both talking to an overclocked FSB... that's the way it is.

As for stability, I've been encouraged to use OCCT as it apparently is more reliable and quicker than PRIME. This will be a tough one to let go of! :)

That's about all for now. Back to the drawing board... I'll report back when I have something useful.

Quote:

---

Originally Posted by Wiggy McShades

is anyone else having trouble getting 400fsb with an E6700 ive tried up to 1.4v on the northbridge but it gets more unstable after adding more than that. whats everyone getting with single dual core cpu's for fsb on these boards i think im doing something horribly wrong

---

475*8 on my E6600 with no issues. Make sure u have the C2D bios and not the Quad.

What kind of NB voltages are you guys using to get to 500fsb and beyond? What kind of heat is that generating? Is the board stable at 500fsb or beyond for Core2Duo, not quad?

I thought about 490FSB was tops for this board due to its LT chipset. There are boards that have done 500FSB , but I think I remember Tony stating that 465FSB was the sweet spot & over 500 yeilds a massive drop in bandwith. I intend to lower the multiplier of my 6600 "B" to 8 X 465 since its already proven stable @ 412 X 9.

allright, I can work with that, only reason for this one is sli, you think running P965-S clock for clock with this 680iLT aside from sli as a benefit.. which one performs better? (running both raid-0(system) and raid-5(storage))

Srry about the newb Qs

Quote:

---

Originally Posted by nugzo

475*8 on my E6600 with no issues. Make sure u have the C2D bios and not the Quad.

---

what voltages are you using? and whats the northbridge temp?

Quote:

---

Originally Posted by virtualrain

I had a brief discussion today with Tony and Raju about quad-core overclocking on this board... here's the outcome...

I'm going to try a new approach. If you read back a page, my previous approach was to set high voltages and a reasonably aggressive FSB and try to tune GTL that way. Well as you can imagine, that's like throwing darts in the dark. I ended up with rather Bizarre GTL settings that seem reasonably stable but I'm going to try it a different way this time.

I'm going to start with some aggressive voltages again, but this time, I'm going to start at 1.4V VTT and suggested GTL values with FSB=300MHz and work up in 10MHz increments. Every time I increase it and it fails stability testing (more on that below), I'm going to tweak GTL2 upwards a bit... since the key is to get Core 2/3 "in sync" with Core 0/1. When I find stable settings, I'll add another 10MHz. This might take awhile.

If my wacky GTL settings are the outcome again, it really means my core dies are not speaking well with each other via the FSB and I likely have a less-than-ideal CPU. I tended to agree that a good part of the problem in overclocking Kentsfields are the dual-die's are both trying to use the same bus and so synchronizing levels is critical. It may take some finesse to get them both talking to an overclocked FSB... that's the way it is.

As for stability, I've been encouraged to use OCCT as it apparently is more reliable and quicker than PRIME. This will be a tough one to let go of! :)

That's about all for now. Back to the drawing board... I'll report back when I have something useful.

---

virtualrain there is nothing wrong with your CPU trust me, just put it on a EVGA Rev1 and you will see.

I just wasted 4 more hours on this board to get a boot with minor OC and no luck. DFI can shove their GTL tweek:fact:

I can do 3.2 with this CPU 24/7 on the Evga with out ever touching the NB and running the CPU at 1.35v max.:fact: The DFI board is rock solid at everything stock I will give it that but thats not fun.:rolleyes:

Quote:

---

Originally Posted by palese

virtualrain there is nothing wrong with your CPU trust me, just put it on a EVGA Rev1 and you will see.

I just wasted 4 more hours on this board to get a boot with minor OC and no luck. DFI can shove their GTL tweek:fact:

I can do 3.2 with this CPU 24/7 on the Evga with out ever touching the NB and running the CPU at 1.35v max.:fact: The DFI board is rock solid at everything stock I will give it that but thats not fun.:rolleyes:

---

I hear you! This board takes time, and even then, I'm not yet convinced it will yield optimum results.

I'm going to give this a go. Problem is, who trusts OCCT as a stability test? I'm not sure I trust it more than running a few instances of Windows Calculator. :p: (EDIT: after using it for a bit, I'm convinced OCCT is running PRIME code under the covers anyway - it has a similar cycle of loading the CPU as PRIME does with different FFT sizes).

What do EVGA users use for testing Quad-Stability?

Quote:

---

Originally Posted by virtualrain

I hear you! This board takes time, and even then, I'm not yet convinced it will yield optimum results.

---

well ya that is the question in the end. will the fine tuning tweaks actually get us better results than with the other boards.

although it is frustrating and very bad PR that its hard to achieve even a moderate overclock without getting into complicated settings. if you can do it on the EVGA boards its obviously possible.

but i guess this is all expected. evga had a 6 month lead developing their BIOS. people have touted oskar wu as a genius though, at the moment its hard to justify. mis-labeling the GTL bios options and the PWM temp readout seem like silly mistakes.

I want the last 11 hours of my life back:slapass: This board is going in storage for the next 6 months like my Evga did until its fixed:slapass: and the Evga is coming back with open arms:woot:
When will I ever learn:fact:

Quote:

---

Originally Posted by palese

I want the last 11 hours of my life back:slapass: This board is going in storage for the next 6 months like my Evga did until its fixed:slapass: and the Evga is coming back with open arms:woot:
When will I ever learn:fact:

---

So palese, what CPU were you clocking on the DFI and what settings did you try?

Also, what were you able to achive on the EVGA?

I'd like to know your experience on both boards in more detail (as painful as it may be for you to re-live! :p: ).

Quote:

---

Originally Posted by palese

I want the last 11 hours of my life back

---

lol... i can relate. i went against my better judgement and tried a bit of oc'ing tonight. i got to a 350 FSB pretty easily with 1.4 VTT and the GTL's from the chart. multiplier at 8 so it really wasn't much of an overclock but was nice to see the FSB raised.

i bumped it to 360 and things didn't go too well so i'm done for the night.

on a related note am i the only one to have heard the shreiking beep and C1 post when the system is mad at me ? thankfully tonight i was able to clear it by holding down power / reset for 4 seconds. i'm back running stock woohoo 1 ! !

Still have my DFI in the box, after reading this post i'm afraid to open it... Anyone still interested in buying it pm

I can do "almost" everything with my q6600@3.2 except prime. It failed very hard, shut itself off....

Quote:

---

Originally Posted by The Nemesis

There are boards that have done 500FSB , but I think I remember Tony stating that 465FSB was the sweet spot & over 500 yeilds a massive drop in bandwith.

---

If I remember right this was the sweet spot for the ICFX 32000. Is it the same again?

Quote:

---

Originally Posted by nop

I can do "almost" everything with my q6600@3.2 except prime. It failed very hard, shut itself off....

---

what FSB ?

Quote:

---

Originally Posted by Jobeo

what FSB ?

---

9x356 (1424)

Quote:

---

Originally Posted by Apollonas

If I remember right this was the sweet spot for the ICFX 32000. Is it the same again?

---

Quote:

---

Originally Posted by Tony

the NB chipset is the key to doing 500fsb, most will run 480 or so with a few doing 500 and higher. My board will do 480 no problem...500 no way but with the strap change at 500 that goes a lot looser why would i want too?.

raja hinted at 450fsb...sync mode, trust me thats where you want to run the board ;)

you guys are all fixated with 500+fsb, you really should be looking at where the boards make the best speed and buy the cpu to suit.

---

I'm mistaken, It would seem 450FSB sync is what's hinted as being the performance spot. 465FSB was the RD600.

Quote:

---

Originally Posted by Wiggy McShades

what voltages are you using? and whats the northbridge temp?

---

Not sure of the NB temps, i had a 40mm fan on the HS, was cool to the touch. vcore 1.5v bios 1.46 after droop.

Quote:

---

Originally Posted by nop

I can do "almost" everything with my q6600@3.2 except prime. It failed very hard, shut itself off....

---

Yeah same EXACT issue here. I yanked the DFI and re-installed the EVGA. 3600mhz (450*8) on quad no problems, definately no shut offs. Coming frome someone who has had every 680i released, the EVGA is by far the best for quad. The DFI and EVGA were neck and neck with the C2D though.

It looks like the only advantage to this board over the Evga (which I already own) is a better layout and better cooling.

When I say better layout I mean when you are using two cards in SLI. With the Evga 680i I have to remove the top side heat sink from one of my 8800GTX KO’s to get my X-Fi card to fit in between. With the DFI board it looks like I will not have this problem.

Since I’m only going to use an X6800 with this motherboard I should probably keep it when it arrives.

has anyone figured out how to tell weather their cpu needs a strong or weak driving strength? ive gotten no different results using both? would one be unstable while the other stable at the same speeds?

Quote:

---

Originally Posted by nugzo

Not sure of the NB temps, i had a 40mm fan on the HS, was cool to the touch. vcore 1.5v bios 1.46 after droop.



Yeah same EXACT issue here. I yanked the DFI and re-installed the EVGA. 3600mhz (450*8) on quad no problems, definately no shut offs. Coming frome someone who has had every 680i released, the EVGA is by far the best for quad. The DFI and EVGA were neck and neck with the C2D though.

---

I'm sure Mr. Wu is still tweaking this as we speak. As long as he sticks to it, I believe we'll see some good Quad OC in the future.