Sign in with
Sign up | Sign in
Your question

New K7N2 Delta-L system hanging in Prime95

Last response: in Motherboards
Share
July 23, 2003 12:16:34 AM

I got a new system with an MSI K7N2 Delta-L motherboard and AMD XP2600+ last week, and it's been hanging on me in Prime95 after between 1 and 8 runs of the stress test (around 5 minutes). I've heard that the NForce2 chipset is picky about memory, but I've run Memtest-86 and SiSoft Sandra successfully, and I've tried it with numerous types of memory (listed below) and the results are the same (Prime95 runs a little longer with some, but never more than 10 minutes, and the games still hang after a half hour. I wouldn't care about Prime95 hanging but when my son can't play his games for more than a half an hour without the machine hanging, it's pretty useless. I believe that the PNY memory I tried does claim to work with the NForce2 chipset, so having that hang is particularly annoying. I have another piece of PNY being delivered tomorrow from Office Depot, and I'll try that as well.

Any ideas what the problem might be? I do have a couple of sticks of Micron PC2100 memory that I can take out of another machine to try in this one, and that may be my next approach. I also have not tried reseating the processor.. might that help? And are there some BIOS settings I could try changing that might fix this? I had to reset to BIOS defaults at one point so if there is something that might have been set when the machine was put together (I got it barebones but with the processor installed) I might need to do that again. I'm also not adverse to underclocking to try and get it stable for a bit.

Oh, and it fails with both the case closed and open, so I'm fairly sure it's not an overheating problem. PC Health4 doesn't show any problems until the time that it just hangs along with the rest of the system.

Bill

Here's the list of SDRAM I've tried in the machine:

PNY D256M6421OPT
Chips: "S" logo
AE 0314
S80032VMCTW-6A

Kingston KVR-PC2100DDR/256
Logo: circle with 9 dots making up a square in top left corner
V58C2256804SAT7
0308PL P050413IJALA

K-Byte PC2100 (DOUBLE SIDED)
Spectek sticker
Chips: "S" logo
FT 0230
S80016VH7TW 75A

K-Byte PC2700 (single sided)
Sticker: PMI DDR-333 256mb T286A-03B3 MD3456UPV
Chips: PQI logo
PQ3D328V6
0319

Kingston KVR333/512R
Kingston on chips
D328DL-60
0315-PT16 (also 0316, 0317, 0318)
July 23, 2003 1:42:12 AM

more system specs...why have you tried pc2100 with a 333mhz fsb processor...

Here's what i would do...go into the bios and set a memory divider so that the memory runs at a diffrent speed then the processors bus speed...

Underclock the processor to 133mhz fsb or even 100mhz fsb just to make sure that is not your problem...

You said the system temps looked ok but you failed to mention what they were...i'm just curious...

What are your powersupply voltages (also what brand/model PSU are you using)...it could very well be a power issue...

Make sure you have all of the latest drivers installed...

Try running prime95 from safe mode...could it be a software incompatability?

will give more sudgestions if you could tell me a bit more about your system...also just remembered...you do have the chipset drivers installed...right?


There is no smell better than fried silicon :evil: 
July 23, 2003 2:21:24 AM

Quote:
more system specs...why have you tried pc2100 with a 333mhz fsb processor...

I'm trying PC2100 because I was able to get it cheap/free and figured I would be ok with the performance loss in exchange for not having to spend more money on memory at this time. Actually, it's hanging with both PC2100 and PC2700 memory (what I have in now is the Kingston 512mb/PC2700 and while it lasts a little longer in Prime95, it still dies within 10 minutes).
Quote:
Here's what i would do...go into the bios and set a memory divider so that the memory runs at a diffrent speed then the processors bus speed...

Underclock the processor to 133mhz fsb or even 100mhz fsb just to make sure that is not your problem...

Please tell me how to do this, I am BIOS-settings challenged.
Quote:
You said the system temps looked ok but you failed to mention what they were...i'm just curious...

Right now, 46C CPU, 46C System, per PC Alert4, with MemTest running for about an hour (back down to 41/44 after having stopped MemTest and having the case open for a couple of minutes to get the PS brand).
Quote:
What are your powersupply voltages (also what brand/model PSU are you using)...it could very well be a power issue...

Vcore 1.63, 3.3V 3.33, +5V 5.11, +12V 11.72

The power supply is a cheap one, I am assuming. Duro brand, 400w allegedly. I actually would be pleased to hear that this is the problem as the dealer sold me the case/power supply and he recommended it.
Quote:
Make sure you have all of the latest drivers installed...

Definitely have that, I used MSI LiveUpdate2 to update everything.
Quote:
Try running prime95 from safe mode...could it be a software incompatability?

I could try that but I won't be happy if that is the solution (although I guess it would point towards a possible fix). I should say that I took a HD from an AMD K6-450 system and put it in here and let Windows discover the new hardware (the video card remained the same). I am assuming this is ok and it generally seems ok.

<b>Update</b>: Safe Mode did not fix the problem, it died very quickly.
Quote:
will give more sudgestions if you could tell me a bit more about your system...also just remembered...you do have the chipset drivers installed...right?

The system is pretty simple. The only board I have installed is an AOpen GeForce2 MX400 64mb video card. I have a bunch of USB adapters (network is Orinoco Gold USB) but i can't imagine those coming into play.

So the story is that Memtest-86 v3.0 and memTest 2.0 (HCI Design) both run cleanly. SiSoft Sandra runs cleanly. Prime95 dies quickly. NBA Live 2001 and Nascar 2003 both die after roughly 30 minutes.

<P ID="edit"><FONT SIZE=-1><EM>Edited by misterbill on 07/22/03 10:37 PM.</EM></FONT></P>
Related resources
July 23, 2003 6:42:08 AM

If i did not know any better i would say it was either heat...or power supply...just relised that after trying that many memory modules...memory is probabbly not it (also if memtest says that its ok...then it is OK...)


Swipe a power supply from another computer...see if it runs ok...your 12v line is a bit low...not terrible but still it is worth a shot...forton...enermax...and antec all make quality power supplies...forton is the cheepest but still retain damn good quality

There is no smell better than fried silicon :evil: 
July 23, 2003 6:48:21 AM

reinstalling windows would be a convinient solution (what version i assume 95 or 98)...

Just noticed that your system temp is a bit high other than the cpu heatsink fan do you have any other case fans...if not i would recomend a low rpm 80mm case fan that sucks air through the case

There is no smell better than fried silicon :evil: 
July 25, 2003 6:12:05 AM

Oops.. somehow the system did not notify me I had new replies so I have not been back. Oh well.. I have had an eventful two days.

On Wednesday, Mr. UPS brought me the PNY memory I had ordered from OD. Much to my delight, it had Samsung chips, which I had been told works with the NForce2 board. It seemed to work ok, in that it did not hang immediately, but then after a while it hung again, although possibly in a different way (the Prime95 stuff had gotten past the initial test). Anyway, Thursday night I tried the following:

1) set the BIOS to 133FSB (was 166), processor now claims to be an XP2000+
2) left the case open
3) turned off all Windows power saving options, set to "Always On" (I did this more so I could see the screen when the error occurred, I don't think this would cause the problem to go away)

The current state is that Prime95 has been running for 2-3 hours. I am cautiously optimistic and am looking for an analysis of what this implicates. The other thing I did was move the piece of memory back to the 3rd slot from the 2nd. I did not think this would be a factor but I thought it ran ok initially when I had it in #3 and for some stupid reason I moved it to #2 and then I started having problems. But it might also have been that I closed the case after that.

I do *not* have a case fan. The guy who sold me the barebones system said not to bother, I didn't need one. Bad move, I guess.

Oh, and I'm running XP Home.

Assuming it's still running in the morning, I wonder if I should get daring and reinstall one of the pieces of K-Byte memory and see if it still runs ok with that memory at the slower processor speed?

Oh, as of a few minutes ago, here is some output from Sandra:

Temperature Sensor(s)
Board Temperature : 39.0°C / 102.2°F
CPU Temperature : 34.0°C / 93.2°F
Power / Aux Temperature : 32.5°C / 90.5°F

Voltage Sensor(s)
CPU Voltage : 1.66V
Aux Voltage : 1.71V
+3.3V Voltage : 3.30V
+5V Voltage : 5.08V
+12V Voltage : 12.04V
-12V Voltage : -12.12V
-5V Voltage : -5.20V
July 25, 2003 7:33:34 AM

He told you you dont need a case fan? That is a big error, even little computers need airflow... 46system is really hot, so Im thinking heat is your problem

<i>Royal Fusileres, Company C</i>
July 25, 2003 2:43:47 PM

I need to try it back to normal speed with the case open, and vice versa, and see if it fails one way and not the other. Will do that later..
July 26, 2003 9:50:59 PM

I tried changing the FSB to somethine between 133 and 166 (150, and I had to go into advanced mode to do that).. but the memory ends up at 240. How do I get it to stay at 266? I guess I could try going back to 166 with the case open and see if that will solve my problem, in which case I know I just need to get a case fan (hopefully).

PS: running at 133 (and case open), the machine was rock solid -- Prime95 torture test ran for 36 hours without any errors.
<P ID="edit"><FONT SIZE=-1><EM>Edited by MisterBill on 07/26/03 05:52 PM.</EM></FONT></P>
July 27, 2003 2:34:26 AM

I've found that giving the cpu a bit more on the vcore helped mine (2500+ @200fsb=3200+) to run stable under load. My power supply is not enough powerful when the 3d chip on the video card kick on to keep the system stable.

So, I gave the cpu a .05v boost and a .1v boost to my memory and my system is stable. It might be a bit hot (around 60 at full load) but I will add a better cooler later. I'm running the stock HSF.

-Always put the blame on you first, then on the hardware !!!
<P ID="edit"><FONT SIZE=-1><EM>Edited by pat on 07/27/03 09:06 PM.</EM></FONT></P>
July 28, 2003 12:28:11 AM

After thinking that I had it beat, my system hung again overnight running with the default speeds. Sadly, my son did not write down the numbers and reset it. Then I started it up again and when it failed again after a while, it was on 46c system again (38 CPU). This was with the case open. So I finally decided to pull a system fan (Antec, as it turned out) out of the other system I was replacing, and put it in this one. I'm hesitant to say I'm optimistic (because of past failures), but at the moment the system temp is only 42, and it's well into the Prime95 test cycle.
July 28, 2003 12:33:25 AM

WTF pat...do you want him to fry his cpu and possibly kill his memory?

On air cooling 2.15 volts would kill a cpu!
and adding .2v to the memory would certainly void his memory warantee and shorten the life of his memory chips...

Also he already has a hot system temp...do you think a 70c cpu will help...

So easy to fry yet tastes so good...
Silicon: The other, other, other white meat :evil: 
July 28, 2003 1:05:45 AM

Thanks for pointing out my error...I should have write 0.05v.....I'll edit my post with new value.

So the cpu will have 1.70v...as for memory, I dont think that running them at 2.7 will hurt them for a while.

-Always put the blame on you first, then on the hardware !!!
July 28, 2003 12:32:15 PM

Well, I think I have finally shot this one.. my system has been running Prime95 for 12+ hours and the system temp is just 42 (CPU is 35). That's with the case fan installed but the case still open (it occurs to me that that might not be optimal since it's not able to suck only the hot air out, although it does get more cool air from outside the case so maybe it's a wash.

Anyway, I may try the old memory in it a little later, although I don't think that will work.
July 28, 2003 10:02:32 PM

i think it will work...your computer must have had an ovheating component...

So easy to fry yet tastes so good...
Silicon: The other, other, other white meat :evil: 
July 28, 2003 10:11:18 PM

I dunno.. it was failing pretty quickly before (with the non-Samsung memory), long before it would have had a chance to overheat. That's why I had a good feeling once I put the Samsung in, since the failure changed.

Meanwhile my coworker brought in his spare PNY 256mb SDRAM today but it was Spectek :-(. He has one coming from Office Depot today or tomorrow and will let me know what he gets.. if it's Samsung, he'll let me have it (very generous of him :-) ).
July 28, 2003 11:04:25 PM

yes...indeed very generous...if you get the chance i would just try a diffrent memory module to be sure...good luck

So easy to fry yet tastes so good...
Silicon: The other, other, other white meat :evil: 
July 29, 2003 5:00:27 PM

I find it very strange that the CPU temperature is below the Mobo/Case temperature. I don't have experience with the newest stuff but older stuff the board is always several degrees cooler than the CPU???

The loving are the daring!
July 29, 2003 6:38:30 PM

actually i beleive it because my mobo temperature is also warmer than the cpu...it baffled me for a while...until i found where it was reading the temp...right next to my north bridge that normally gets up around 40c...

3 386DX-25's...12 volts...glue and some ln2 and a wicked amount of overclocking and you get a willamantee minus 36 pins, 33.75 million transistors and a couple hundred mhz... :cool:
July 29, 2003 6:59:23 PM

Yeah, the problem was that the board was obviously overheating. Now with a case fan installed, the mobo temp seems to stay around 35c, while the CPU is about 10 degrees higher.
July 29, 2003 7:03:03 PM

I tried the K-Byte memory again last night (actually like 3am this morning :-( ). It failed for both of them, even when I tried them with the Samsung which works. So now I am running with just the Samsung. My friend who had offered to swap me the PNY Samsung when his came in got Elixir chips instead. I might take it and try to find an OD with Samsung this weekend. And I still have not tried pulling the Micron from my XP1800 system to try in this one. It's probably the easiest thing to do.. I just need to get around to opening it up, and at the same time installing the USB 2.0 card and extra HD I've had lying around for a while.
July 29, 2003 7:22:58 PM

this is increadbly strange...you have one picky mobo...just for the sake of trying...increase the vdimm (memory voltage) by .1 volts...if that does not help any try to dissable dual ddr...it helps very little on amd mobos anyway...

3 386DX-25's...12 volts...glue and some ln2 and a wicked amount of overclocking and you get a willamantee minus 36 pins, 33.75 million transistors and a couple hundred mhz... :cool:
July 29, 2003 8:06:01 PM

You must be kidding.. isn't the NForce2 chipset listed in the dictionary next to the definition for "picky motherboards"? It's not like I am the only one reporting problems with hangs due to non-conforming memory.. the MSI newsgroup is littered with them. Most are still at the "my system is hanging and I don't know why" state.

I sent email to the memory@nvidia address today that is listed on the NForce2 Memory Compatability website saying that I have had it with their stupid chipset and fail to see why they cannot make it more accepting of non-perfect memory. I also said that I would never buy another board with that chipset and would recommend against it if anyone asked. Not that I think it will make a big difference but at least I feel better. I also suggested posting the memory that failed as well as those that passed.

I can't see that dual DDR would make much difference since it's failed when the memory is installed alone and dual DDR is not an issue, but I can try that. Where is the memory voltage setting (what screen?).

To think at one point I wanted to try overclocking the memory up to PC2700. Actually, I would still like to try that :-).
July 29, 2003 8:39:48 PM

it would be under the frequency/voltage control menue in your bios labled as "dram voltage adjust"

Yes nforce boards are picky...but i never thought that with soo many diffrinet modules it still would not work...i don't know if i have recomended this yet but a bios update may help...

3 386DX-25's...12 volts...glue and some ln2 and a wicked amount of overclocking and you get a willamantee minus 36 pins, 33.75 million transistors and a couple hundred mhz... :cool: <P ID="edit"><FONT SIZE=-1><EM>Edited by pIII_Man on 07/29/03 05:00 PM.</EM></FONT></P>
July 29, 2003 10:25:39 PM

Oh, I am already at the latest BIOS.. that is like the first thing I tried. In fact, I considered going backwards to see if I could get something more tolerant. Funny thing is, before ordering this I thought I remembered reading that the latest BIOS did make it more tolerant of memory, but I may have been thinking of the ASUS A7V8X that I was originally planning to buy.

My problem is probably that I'm being cheap and I don't want to go out and spend money at Crucial, or buy the specific model # that Kingston says will work with this mobo. So I'm taking my chances with cheap RAM, and I guess I am paying the price for it.
July 29, 2003 11:24:39 PM

i think doing what i recomended above will get the board stable...

3 386DX-25's...12 volts...glue and some ln2 and a wicked amount of overclocking and you get a willamantee minus 36 pins, 33.75 million transistors and a couple hundred mhz... :cool:
July 30, 2003 12:59:31 AM

Ah ah! now you're stealing my idea of voltage increase!!! If that works, remember to give me full credit! :-)

-Always put the blame on you first, then on the hardware !!!
July 30, 2003 1:03:07 AM

stealing your idea my A$$...you basically told him to fry is cpu and kill his memory...

3 386DX-25's...12 volts...glue and some ln2 and a wicked amount of overclocking and you get a willamantee minus 36 pins, 33.75 million transistors and a couple hundred mhz... :cool:
!