Need help with TWINX2048-3200PT and DFI LANPARTY nf4 SLI-DR

[rye]

I'm having a serious stability problem I hope someone here can help with. I just purchased a DFI LANPARTY nf4 SLI-DR mobo along with a set of TWINX2048-3200PT RAM. Each stick runs fine by itself, but when I throw both sticks in, I get random BSoDs, lockups, errors and such bad stuff. I purchased this RAM for this mobo because it's listed in the Corsair memory selection tool at http://compatible.corsairmemory.com/memorysearch.aspx?modelid=1212#othermodules. I've tried manually setting the mobo to run the RAM at 2.5-3-3-6 and both 2.6 and 2.7 volts, as I've seen elsewhere, but it's still very unstable (though somewhat less so than before). I've flashed the DFI BIOS and updated the nforce drivers to latest versions and those both helped a little bit.

 

Should I return this RAM for something else or is there something I can do to make it stable? If you think I should return it, what RAM works best for this mobo in dual-channel mode that gives me 2GB is 2 sticks? Thanks in advance for your help!

[Garvin]

Did you install windows with one stick of ram or both sticks in dual channel? Some people have had problems, such as yours, with either windows Xp or media center installs hanging on them or random BSOD's and instability occuring after Xp or media center is installed while using a dual channel setup.

[rye]

Did you install windows with one stick of ram or both sticks in dual channel? Some people have had problems, such as yours, with either windows Xp or media center installs hanging on them or random BSOD's and instability occuring after Xp or media center is installed while using a dual channel setup.

Actually, I've tried it both ways. In either case it's unstable as heck once I've got both of them in.

[Garvin]

Have you tested each stick with memtest86? You state that each stick runs fine by itself; this would tend to point to something else being wrong. Could you list all the components in your system including the power supply. Also, how is/are your hard drive(s) hooked up?

[rye]

Have you tested each stick with memtest86?

I ran memtest86 for an hour on the pair of them with no errors, but I haven't run it on the modules individually. Would it be useful to do so?

[Garvin]

I ran memtest86 for an hour on the pair of them with no errors, but I haven't run it on the modules individually. Would it be useful to do so?

No. If the pair passed memtest, then we'll have to look elsewhere for the problem. Need system component info. How are the drives hooked up?

 

ps: jeeze you post fast.

[Wired]

The timings for TWINX2048-3200PT are:

CAS Latency: 3

RAS to CAS Delay (tRCD): 3

Row Precharge (tRP): 3

Active to Precharge Delay (tRAS): 8

[rye]

No. If the pair passed memtest, then we'll have to look elsewhere for the problem. Need system component info. How are the drives hooked up?

 

ps: jeeze you post fast.

I've got four Western Digital Raptor (WD740GD) 10K SATA drives. I tried running them through the SiS controller in RAID5, then RAID1+0, then through the nforce controller in RAID1+0 and finally just a single drive with no RAID through the nforce controller (current config). I've had instability in all cases, except when I run with only one DIMM or the other.

 

Other componants are as follows:

-Lian Li PC-7077A Silver Aluminum ATX Full Tower with all four fans running

-AMD Athlon 64 3500+ ClawHammer/Hammer 1GHz FSB Socket 939

-GIGABYTE GV-RX80L256V Radeon X800XL 256MB GDDR3 PCI-Express x16

-Antec TruePower 2.0 TP2-550 EPS12V ATX12V 550W

-two Plextor PX-716A drives

-four Western Digital Raptor WD740GD 73GB 10,000 RPM Serial ATA150 drives

-DFI LANPARTY nF4 SLI-DR Socket 939 NVIDIA nForce4 SLI ATX AMD mobo

-D-Link DWL-G132 802.11g 108Mbps USB wireless adapter

-random floppy drive

-WXP SP2

 

P.S. You sure do reply with helpful questions and comments fast! :biggrin:

[rye]

The timings for TWINX2048-3200PT are:

CAS Latency: 3

RAS to CAS Delay (tRCD): 3

Row Precharge (tRP): 3

Active to Precharge Delay (tRAS): 8

Yep, I ran with those for quite a while with no luck. The mobo detects those timings when everything's set to automatic. After I changed to 2.5-3-3-6 things got better, but it's still very bad.

[Garvin]

Did you delete the RAID array before hooking up the drive without RAID? There's another thread around here somewhere with similar problems with instability with the raptors. Cleared up by using the Western digital utilities for that make of hard drive to wipe the drive so that they can be repartitioned/formated.

 

ps: the nforce and sis RAID instructions stink don't they.

[rye]

Did you delete the RAID array before hooking up the drive without RAID? There's another thread around here somewhere with similar problems with instability with the raptors. Cleared up by using the Western digital utilities for that make of hard drive to wipe the drive so that they can be repartitioned/formated.

 

ps: the nforce and sis RAID instructions stink don't they.

Yep, I deleted the array, disabled the appropriate RAID in BIOS and low-level formatted the drives before switching between one setup and the next.

 

Yeah, the directions that came with the mobo aren't very good. However, some further online reading and previous experience helped it all make more sense.

[Garvin]

When loading windows did you press F6 and load the RAID driver third party software? It's also functions as a memory controller program in non-RAID setups , without it Xp or mediacenter or the chipset drivers will try to handle the job and acheive poor results. How are the Plextor's hooked up? Yep, they could also be causing the instability you're seeing.

[rye]

When loading windows did you press F6 and load the RAID driver third party software? It's also functions as a memory controller program in non-RAID setups , without it Xp or mediacenter or the chipset drivers will try to handle the job and acheive poor results. How are the Plextor's hooked up? Yep, they could also be causing the instability you're seeing.

When loading WXP, I did hit F6 and load the appropriate RAID drivers, but *only* when using RAID. The directions don't say anything about loading them when not using RAID. Are you sure they're needed in non-RAID modes? If so, can you point me to any documentation? (Not because I doubt you, but for future reference and understanding.)

 

The Plextors are on seperate IDE channels as masters (hard-jumpered and at cable end). I've tried running with just one of them, then the other and then with both as slaves (at cable-middle). No difference in any case.

 

Here's the odd bit of info. After reading around these forums last night, I noticed that everyone else was running memtest86 for a lot longer than my one hour. I ran it overnight last night and have a bunch of errors this morning. My results show 360 errors in 8h13m. All errors are in test #8. Could this be anything other than bad RAM?

[RAM GUY]

Rye,

The correct settings for TWINX2048-3200PT would be Cass 3-3-3-8, these modules are not rated for Cass 2 or Cass 2.5. In addition, I would not suggest running test 8 on an AMD platform unless you know exactly what you are doing with this software.

Also you might want to read thru What is memtest86+, what do I use it for?

[rye]

Rye,

The correct settings for TWINX2048-3200PT would be Cass 3-3-3-8, these modules are not rated for Cass 2 or Cass 2.5. In addition, I would not suggest running test 8 on an AMD platform unless you know exactly what you are doing with this software.

Also you might want to read thru What is memtest86+, what do I use it for?

I ran it at 3-3-3-8 most of the time. I only bumped it to 2.5-3-3-6 to try something I saw on some forums (maybe at DFI). The memtest I ran was with the settings back at 3-3-3-8.

 

I'm curious why you suggest against running test #8. The memtest86 FAQ says the following about it:

Test 8 [Modulo 20, ones&zeros]

Using the Modulo-X algorithm should uncover errors that are not detected by moving inversions due to cache and buffering interference with the the algorithm. As with test one only ones and zeros are used for data patterns.

 

Based on what I have learned, I'm gong to run memtest86 on the individual modules and see what happens. If one module gives errors and the other doesn't, I'll assume a bad stick. Does this seem reasonable to you? If they both give errors, I'm not sure what to do. Suggestions? (Sadly, I'm out of town for a few days and won't be able to do that until I return.)

[RAM GUY]

Yes, but not if you are only getting errors with Test 5 or 8 as that may suggest some other device is failing not the memory.

[Garvin]

For information on test 8 I'd google on over to the Memtest86 Website and read the warning about this test with the speedier AMD processors, under trouble shooting. Three passes of Memtest86 are sufficient to uncover virtually all ram problems, running it longer is a waste of time and very taxing on the microprocessor, chipset, and ram. As a sidenote, a problem with a memory controller (either on the northbridge chip of Intel systems or on your AMD 3500 processor) can masquerade as a ram problem. If memtest reports an error on subsequent passes at the same address in the same test then it could be a ram problem as long as there isn't a power supply or insufficient voltage problem, starving the ram sticks. If memtest reports an error on subsequent passes at different addresses this would tend to point to either a defective memory controller or once again a power supply problem or insufficient voltage to drive the memory controller or in your case your AMD 3500 cpu. As to documentation on the Raid driver question, it would be hard to say, I know it was mentioned at least once on this site and I've seen it mentioned before in any of a couple of dozen forums I drop in on from time to time. Remembering where a few thread entries appeared out of the 12,000+ entries I've read in the last four months is nye near impossible. May I suggest Google.com.

 

Ps: 10,000 of the entries were read before I ordered and assembled the rig I have now, made troubleshooting a breeze.

[RAM GUY]

Or I actually posted this the other day that will help explain this same thing as well.

[rye]

Many thanks to both Garvin and RAM GUY for the great information! I just got back from out of town and will see what happens tonight with memtest running against the individual sticks. More details soon...

[rye]

Or I actually posted this the other day that will help explain this same thing as well.

RAM GUY,

FYI, the link to http://www.houseofhelp.com/v2/showthread.php?t=33521 you've got in the memtest FAQ you've listed above is broken.

[Wired]

The FAQ forum has been updated recently, and as such some links have been broken.

[rye]

New test information follows. Can anyone help me determine the source of the problem of perhaps fix it by tweaking?

 

I ran a lot more memtest v3.2 tests this evening. Here's what I did and what I found.

 

1) Repeated testing of individual DIMMs in slot2 generated no errors whatsoever. (Machine won't POST with a single DIMM in slot4, so this wasn't tested.)

 

2) Repeated testing of both DIMMs in slots 2 and 4 (the dual-channel slots) generated easily respoducable errors during every pass, but only with test #8 (modulo 20, ones & zeros).

 

3) The errors always occur in bursts. Between 3 and 16 individual errors show up at a time, all at closely related addresses within the bursts. However, each burst can show up at just about any address, so far they've occured between 635.0MB and 2043.8MB in the same run, so both DIMMs ranges are reporting errors.

 

4) The errors show up regardless of having DIMM1 in slot2 and DIMM2 in slot 4 or of having DIMM2 in slot2 and DIMM1 in slot4. I've switched these back and forth several times.

 

5) All of the errors show the "good" column listing "00000000", "bad" cloumn listing "ff000000" and "err bits" cloumn listing "ff000000".

 

6) All test environments have the mobo auto-detecting the RAM settings as 3-3-3-8 1T.

 

7) Nothing in the system is or has been overclocked.

 

Any ideas? Thanks in advance for your help!

[Garvin]

I did some searching over at hardwareanalysis.com on your DFI Lanparty board and frorm what I was reading your board defaults to a RAM voltage of 2.6 volts. The link to the forum thread is:

 

http://hardwareanalysis.com/content/topic/40970/

 

Ram Guy will have to confirm this but you could be undervolting your ram, resulting in the instability your seeing. Also about 2/3 of the way down the thread page, Chris G has listed his working bios settings for this board, as well as his method for upping the ram voltage. You should probably compare your bios settings to his. Many other people have had the instability problems your experiencing from what I saw. There are a few other threads on this motherboard over at hardwareanalysis.com you might want to look at. I don't personally own this board; I have the Asus version of the nf4 so I can only point you in directions to look at troubleshooting wise having fixed a slew of different systems. I wouldn't be too concerned with your test 8 errors and think the errors are pointing you in the wrong direction. RMAing hardware and discovering the problem isn't fixed with the replacements because some little esoteric setting wasn't set correctly due to a poorly written manual is depressing to say the least. I'd check with Ram Guy for the correct voltage level for your sticks before upping the voltage and use his timings for the ones Chris G is using as well.

 

ps: I don't work for Corsair or any computer related company, just think the problem lies elsewhere and isn't the ram itself.

ps2: Heres another link to a working bios configuration:

http://reviews.pimprig.com/motherboards/dfi_lanparty_ut_nf4_sli-dr.php?page=4

and another:

http://www.angrygames.com/sli-dr_1.htm

and another:

http://www.angrygames.com/nf4_ultra-x1.htm

 

all of them have Errata 94 Enhancement set to disabled, threads mention instability with this setting enabled. This board has a ton of Bios settings and from what I could see many owners are also having problems getting it set right with various component configurations.

ps3:Most motherboard manuals are poorly written or poorly translated and some are outright wrong in their bios configuration information. Just thought you should be aware of this.

[Wired]

Yep, tested @ 2.7 volts.

[Garvin]

Another possibility is the version of nvidea chipset driver your using with your raptors. People have reported problems with drivers after version 6.39 and raptors, especially 6.53.