DFI NF4 Ultra/SLi-D BUGS and Fixes Thread

[Lidocaine]

Trans, is it possible the DFI boards have issues with TCCD rams? my Patriot TCCD is giving me trouble when I install Windows with some files not copying over to the hard drive. Also I made a mistake initially plugging in the power LED and the HDD Led front connectors. Is it possible that I damaged it and that's why the LED's aren't on even though the CPU and the HDD is connected?

Try turning the little connectors upside down on the pins. This worked for me.....I had this same experience with my Neo2 boards...........and this DFI board.

[quicksilverXP]

I have restart problems everytime I use the 4v option. Can't seem to get it stable. Anyone else have any luck using the 4v option and some OCZ VXs?

[Orker]

No more problem with ram .. It appears it doesnt like the yellow dimms

[Blue078]

Trans, is it possible the DFI boards have issues with TCCD rams? my Patriot TCCD is giving me trouble when I install Windows with some files not copying over to the hard drive. Also I made a mistake initially plugging in the power LED and the HDD Led front connectors. Is it possible that I damaged it and that's why the LED's aren't on even though the CPU and the HDD is connected?

I cant get my 2x512 of PDP TCCD's to do 260 in this board, they were fine @275 in the NEO2. ????

I've been trying since 6:00 this evening and cant make them pass even 1 memtest run..

Stuck in some ram with Hynix chips and it good to go....

[Sgt_Strider]

I cant get my 2x512 of PDP TCCD's to do 260 in this board, they were fine @275 in the NEO2. ????

I've been trying since 6:00 this evening and cant make them pass even 1 memtest run..

Stuck in some ram with Hynix chips and it good to go....

Hopefully, DFI release a new patch that will fix the TCCD issues. I thought their latest revision fix that? Well, at least I can get Windows installed and surf the internet and stuff. I'm kinda iffy with using beta BIOS after my nightmare with the Asus A8N-SLI's beta BIOSES. Basically their beta bios caused the board to no longer detect my hard drive. It was a pain in the ass to get everything reinstall.

[Jimi...]

You might be able to get your RAM working properly by tweaking the individual settings.

Here's a good guide to RAM overclocking.

[jikdoc]

I have restart problems everytime I use the 4v option. Can't seem to get it stable. Anyone else have any luck using the 4v option and some OCZ VXs?

i have post problems with 4v option. board won't post when the jumpers are in the 4v config. using a zippy6701. opb told me he's having the same problem with his zippy and this board, something to do with the 5v sense line.

[quicksilverXP]

Yeah... I know what you mean. Sometimes it posts, sometimes it doesn't... but most of the time it doesn't. But if it does... I can get to windows and run Prime fine... but then when I want to restart it will not post again... it's getting annoying.

[ominae]

Everything is on Auto and I disabled Bank interleave.


Edit: When the computer boots up, the screen displays my timing as 1T even though I set everything to Auto. Could that be the reason why? How do I get rid of 1T and make it less aggressive to ensure stability?

yo, I thought enabling bank interleave increases performance?

[ViperJohn]

I'd like to share my current issues and also ask a question since the manual says NOTHING about it...

First my question:
- I have the Ultra-D board, and it has the 24-pin ATX connector, the 4-pin square one, and then it also has the FDD connector by the PCIe slots, and then the HDD connector near the nForce4 chip/fan. Is it REQUIRED to have connectors plugged into the FDD and HDD connectors to get this board to run stable? Unless I am blind, the manual only references these 2 connections but doesn't specify if/when you need to have power going into them. Thanks!

Brian

The info on the connectors is on page 56 of the PDF manual on the DFI CD. If your CD
doesn't have the PDF manual D/L it from DFI.

Viper

[D_o_S]

How do I know which PSU works with this mobo?

[Orker]

I have had everything setup for a couple of hours now and was playing css fine

Now when i play it, the screen violently shakes

EDIT: some good news.. it only happens with css.. so its definately a software issue

[cupra]

have week 40 and it seems max stable at 3.1 any lower or any higher = errors in mem86

i cant work with memtest86 right, its dont recognize the nf4 chipset so i cant
see the mhz and the timing of my memory.
i tray the ocz memtest and the last memtest86.
in which memtest86 you use for your tests trans am ?
and its work fine?

edit : and same as opp my fx-55 and my vx works fine with the sli-dr r.01 25/1/05 bios

[HARDCORECLOCKER]

i cant work with memtest86 right, its dont recognize the nf4 chipset so i cant
see the mhz and the timing of my memory.
i tray the ocz memtest and the last memtest86.
in which memtest86 you use for your tests trans am ?
and its work fine?

edit : and same as opp my fx-55 and my vx works fine with the sli-dr r.01 25/1/05 bios

In my opinion forget memtest. I tried several hours to get the max stable settings with memtest - then was very lucky to found some and in the end the rig didn't boot at all.........

[Anakin_Skywalke]

In my opinion forget memtest. I tried several hours to get the max stable settings with memtest - then was very lucky to found some and in the end the rig didn't boot at all.........

As the name of the test implies, it is JUST a memory test to see whether the modules will be up to the voltage and/or mhz. It does not strain the processor or other parts of the board. It CAN be used though to know which are the max memory timings and frequency for your memory. Please use it accordingly and don't expect it to be a panacea.

[HermS]

I was having some bad issues with the 25/01 bios. No matter what I did the rig would not boot with any kind of memory overclock, it kept hanging at the NV bootagent and the SATA drives weren't showing up at all. It was strange as when I booted at stock settings and clocked it up with clockgen it displayed no stability issues whatsoever. Anyways to cut a long story short I flashed to the 01/02 bios and the boot issues magically disappeared.

[joe2004]

As the name of the test implies, it is JUST a memory test to see whether the modules will be up to the voltage and/or mhz. It does not strain the processor or other parts of the board. It CAN be used though to know which are the max memory timings and frequency for your memory. Please use it accordingly and don't expect it to be a panacea.

It is not just a memory test, it is also a processor cache test. This is the best test around for memory and if you have problems/errors with this test, your memory/CPU subsystem does not work properly, that is all.

[mikead_99]

In my opinion forget memtest. I tried several hours to get the max stable settings with memtest - then was very lucky to found some and in the end the rig didn't boot at all.........

I've found on several platforms now that once I get my mem to pass memtest86+, it may still need a bit more voltage/tweaking to get stable in Windows. It is OK to get a "feel good" before going into Windows, but it ain't the be all/end all test either.

[Wingz]

it be good if someone can use a multimeter to measure the true vdimm voltage and see if it under/over volts.

[HARDCORECLOCKER]

I've found on several platforms now that once I get my mem to pass memtest86+, it may still need a bit more voltage/tweaking to get stable in Windows. It is OK to get a "feel good" before going into Windows, but it ain't the be all/end all test either.

Yep - that's exactly what I tried to say - had better experiences for stability test in Windows by using SuperPi.

[eva2000]

I've found on several platforms now that once I get my mem to pass memtest86+, it may still need a bit more voltage/tweaking to get stable in Windows. It is OK to get a "feel good" before going into Windows, but it ain't the be all/end all test either.

that's what memtest is for - initial testing.. b4 going into windows for more testing

[SVTSnake]

Got my board installed yesterday, went through several MAJOR curse out battles, pissed the wife off until she finally left to go out, and finally calmed down.

My board came from Hong Kong as a good friend of mine was on a business trip there. Not having a complete printed manual, my 1st question was rather or not we had to plug in the 12v FDD and HDD connectors to the board. A quick posting to the site referred me to the DFI online manual. After reading it, I went ahead and plugged in the extra 2 power connectors even though I am using the ATI X850XT PE. I figure if it is going to help system stability with SLi, then it might help flatten out the voltage in normal mode. So far, its running great, after of course going through the headaches below...

At 1st power up, I almost had a heart attack. The screen turned on, saw the pretty but useless LanParty boot screen, didn't go any further, and the fans didn't sound right, so I looked down and noticed not only the rear case 120mm fan not spinning, but the CPU fan not spinning as well Immediately pulled the power thinking that the CPU was toast. Check all the cables, everything was fine, turned back on the system, and then the CPU fan spun After about the 3rd lockup at the bios post screen, I was finally able to get in. My board shipped with the 01/25/05 bios. Using OPB's settings, I was able to boot, but it locked up after the IRQ listings. I was only giving my TCCD 2.7v as thats all it needed on the Neo2. I remembered reading a post Friday night from someone saying he needed 3.1 volts to get his TCCD to boot, so I changed the voltage to 3.1v and voila, I was able to pass Memtest at 300 2.5/4/7/4 (Memtest ran for over an hour with no errors on test #5). Of course my crappy 3500+ wasn't up to the par for running @ 300 The one thing that kept bothering me was I kept getting lockups at the bios post screen right after it identified the CPU. Flashed the bios to 1/26, same problem, flashed to 2/01, same thing, I was like DOH! Finally, I unplugged my USB Logitech Momo Force Feedback steering wheel, and the lockups disappeared. Never had this problem before on ANY other mobo. That was a major pain trying to solve. Got Windows installed, installed SP2, installed the 6.39 nForce drivers, and noticed that my nForce network controller wasn't working? I had it turned on in the bios, the driver was loaded, Active Armor was running, just no internet connection. This is a 1st, but you have to MANUALLY enter in the MAC address from the sticker on the enet port in the bios. Once I did that, it works like a charm. Now why doesn't DFI just do that autimatically in the bios? Every other vendor I have ever used, I have never had to enter in the MAC LAN address. Oh well!

I could never get my 3500+ to be stable enough for Prime on this setup no matter what voltage I used, so for good laughs, I threw back in my 3000+ that would only do 280 on my Neo2 @ 1.65v. She is chugging along right now at 290 @ 1.552v, Prime stable overnight! May try for the magic 300 later today, but I am kinda exhausted after all the troubles yesterday brought. Out of all the systems I have ever built (100+), this is by far the 1 that caused me the most headaches. However, in all due to DFI's defence, we are the pioneers of their Socket939 adventure, and they have giving us what everyone else wouldn't in terms of overclockability and settings. Now that the boards are out, and they can be run with all the various hardware out there, it will hopefully be easier for them to fix the quirks.

I am still stumped on the voltage required to run the TCCD? Most 437 Samsung chips like low volts, as was the case with my Neo2. Using 3.1 volts on them, they are still cool to the touch, as well as the mofsets around them. If I go under 3.1 volts, the board locks after the IRQ boot screen. And yes, my 5v jumper is in the stock position for 3.2v max. For those having TCCD boot issues, try changing your voltage to 3.1v and see if it helps with the post lockups. Also, try unplugging your USB periphs. Like me, my Logitech Momo steering wheel caused hard lockups during post. I have tried setting the USB jumpers onboard to both +5v and +5vsb with the same result.

Using OPB's memory settings from this thread.

For everyone out there going in the unchartered waters with the DFI, here's one to ya In all, once you get it setup, its a really nice rig!

[SVTSnake]

It is not just a memory test, it is also a processor cache test. This is the best test around for memory and if you have problems/errors with this test, your memory/CPU subsystem does not work properly, that is all.

I would NEVER use Memtest as a end of all true stability test! DOS does not stress out the cpu enough, even if it is using the cache controller on the CPU. With my 3500+, I could Memtest @ 275 with the 11x multiplier, but as soon as I went to boot WinXP, the system would reboot. This was after running Memtest over 4 hours using tests 5 & 6.

Use Memtest to find the MAX of your memory. Lower your multipler down to 5 or 6, then crank up the HTT til you start getting errors in test 5 or 6. Keep adjusting memory settings until you find MAX stable memory.

[scottk]

that's what memtest is for - initial testing.. b4 going into windows for more testing

Yep, if it'll pass all tests 1-7 and survive test #5 looping, it's usually good enough for a clean boot and shutdown. A good way to avoid trashing your OS install. I usually have to back down 4 - 5 notches on FSB/HTT from "memtest-stable" to stabilize in P95 and 3D01.

[Blue078]

Does any one know yet why it takes 3.1vdimm to get TCCD's to clock on this board ??

I have some PDP with XBL and it ran just fine in the NEO 2 on 2.7v
But couldnt get the ram to 260 on 2.7 even with OPB settings on the SLI-DR
After reading you guys posts about haveing to use 3.1V I caved in and tried it, and at least memtest is runing now.

Strange ?